The Ultimate B2B Programmatic Glossary - Everything You Need To Know About B2B's Most Under-Utilised Channel
Built to be bookmarked. Written from the operator's desk.
A special edition of The B2B Stack. Not the Friday briefing.
Most adtech glossaries are written by vendors trying to make their product sound clever, or by SEO content farms paraphrasing the IAB. Neither group has bought a deal ID in anger, reconciled a curated marketplace against the publisher ad server it claimed to deliver, or sat in front of a CFO explaining why the take rate looks the way it does.
This one is written from that desk.
115 terms across nine clusters, plus a benchmarks coda with operating numbers from the work: typical CPMs by channel, realistic match rates by environment, the working media percentage you should actually be hitting, the supply-path hop counts that explain where your money has gone, the frequency caps that keep B2B presence on the right side of harassment. Roughly forty minutes if you read it through. Ten seconds if you jump to the A–Z and look up the one term you came for.
Where the industry uses a term loosely, it gets flagged. Where the B2B definition diverges from the generic one, the divergence is called out. Where a term has quietly changed meaning over the last twelve months, you’ll see that here before you see it anywhere else. Where there’s a specific number from inside the work that helps you calibrate what you’re seeing, the number is here, hedged where it deserves to be and unhedged where it doesn’t.
There is no shortage of definitions on the internet. There is a shortage of definitions that tell you what to actually do with them. That gap is what this is for.
It is designed to be bookmarked, returned to, and updated as the vocabulary shifts. The regular Friday briefing returns to its usual cadence this week. Reply to this email with corrections, gaps, or anything you’d argue with — the document gets better when operators help keep it honest.
What you’ll learn in this article
The auction and bidding mechanics that govern every impression you buy, and the bid-shading savings most teams don’t see
The supply-side terminology that defines what inventory actually is, including the curation, MFA, and verification layers that quietly determine campaign quality
The buying-side architecture, fees, and economics that compose the programmatic tax — and the working media benchmarks worth measuring against
The identity and audience plumbing that has been quietly rebuilt around the cookie’s collapse, with realistic match rates by environment
The B2B-specific vocabulary that separates real operators from people who learned the trade in B2C, including the frequency and signal-decay metrics that don’t appear in any DSP report
The measurement, attribution, and signal language that lets you tell working media from theatre
A benchmark coda with the operating numbers (CPMs, match rates, working media, supply-chain hops, incrementality thresholds) you should be testing your own stack against
How to use this glossary
Read it through once if you’re newer to the space. Bookmark it if you’re not. The A–Z index below lets you jump straight to a term you came in for. Each entry stands alone, so dropping in from a search engine or an AI answer gets you the full context without backtracking. Terms in italics inside an entry are defined elsewhere in the glossary.
If you find something missing, wrong, or fashionable nonsense, reply to this email. Every correction gets read and considered.
A–Z index
If you know the term, jump straight to it.
A · Account-Based Marketing (ABM) · Account graph · Account-level frequency · ads.txt and sellers.json · Ad exchange · Ad server · Agentic ad tech · Answer Engine Optimisation (AEO/GEO) · Audience containerisation · Audience extension · Authenticated audience · Authenticated traffic
B · B2B Marketing Operating System · Bid duplication · Bid shading · Bid stream · Bombora · Brand safety · Brand suitability · Buying committee
C · Closed-loop measurement · Connected TV (CTV) · Contextual · Conversion API (CAPI) · Cost per Mille (CPM) · CPC, CPA, CPL, CPV · Curation · Customer Data Platform (CDP)
D · Dark funnel · Data clean room · Data co-op · Data CPM · Data Management Platform (DMP) · Data onboarding · Deal ID · Demand generation · Demand-Side Platform (DSP) · Digital Out-of-Home (DOOH) · Display
F · Fingerprinting · First-party data · First-price auction · Firmographic · Floor price · Frequency cap
G · Generative Engine Optimisation (GEO)
H · Hashed email (HEM) · Header bidding
I · ID5 · Ideal Customer Profile (ICP) · Identity graph · In-app · Incrementality · Insertion Order (IO) · Intent data · Invalid Traffic (IVT)
L · Lead generation · Lift study · Lookalike modelling
M · Made-for-Advertising (MFA) · Marketing Mix Modelling (MMM) · Match rate · Mobile Advertising ID (MAID) · MQL · Multi-Touch Attribution (MTA)
N · Native
O · Open exchange · Open web · OpenRTB · OTT
P · Pacing · Pipeline attribution · Pixel · Preferred Deal · Private Marketplace (PMP) · Privacy Sandbox · Programmatic audio · Programmatic Guaranteed (PG) · Programmatic tax · PPID
R · RampID · Reach · Real-Time Bidding (RTB) · REDS · Remarketing · Retargeting · Return on Ad Spend (ROAS)
S · Second-party data · Second-price auction · Self-service vs managed service · Signal · Signal decay · Signal envelope · 6sense and Demandbase · Supply Path Optimisation (SPO) · Supply path reconciliation · Supply-Side Platform (SSP) · Surge
T · Take rate · Target Account List (TAL) · Technographic · Tech fee · Third-party cookie · Third-party data · Topic taxonomy · Total Addressable Market (TAM)
V · Verification · Video · Viewability · Viewable CPM (vCPM) · View-through
W · Walled garden · Win rate · Working media
1. Auction and bidding
The mechanical layer beneath every programmatic impression. Most marketers never look here. The ones who do tend to spend their budgets better.
Real-Time Bidding (RTB)
The auction mechanism that runs underneath open programmatic. Every time an ad slot becomes available, an auction request goes out to bidders, who respond with a price within milliseconds. The highest bidder wins, the creative is served, the page loads. All of this happens in roughly the time it takes you to blink.
In practice. RTB is the default assumption when someone says “programmatic,” but increasingly it isn’t where the interesting buying happens. Curated marketplaces, programmatic guaranteed, and preferred deals all sit alongside open RTB, and for B2B specifically, the share of spend going through pure open auction has been falling for years.
OpenRTB
The technical specification that defines how RTB requests and responses are formatted. Maintained by the IAB Tech Lab. The reason a DSP built by one vendor can talk to an SSP built by another.
In practice. You don’t need to read the spec. You do need to know it exists, because every conversation about new signals, new identifiers, or new auction mechanics ultimately comes down to whether the OpenRTB schema supports them. If a vendor tells you they’re sending a new audience signal in the bid stream, the right question is which OpenRTB field.
First-price auction
An auction format where the winning bidder pays exactly what they bid. The default across most of programmatic since around 2019. Replaced second-price auction as header bidding broke the assumptions that made second-price coherent.
In practice. First-price changes how you bid. In a true second-price world you could bid your maximum and trust the auction to clear at the next price down. In first-price, every penny you bid is a penny you pay, which is why bid shading exists.
Second-price auction
An auction format where the winner pays one cent more than the second-highest bid. Theoretically encourages truthful bidding because there’s no penalty for bidding your maximum. Largely extinct in open exchange, though it persists inside some private marketplaces and walled gardens.
In practice. Worth knowing about because the assumptions it created (bid your true maximum, don’t worry about the clearing price) still leak into how some buyers and DSPs reason about bidding, years after the auction format itself was replaced. If you’re being told to “just bid higher” by a vendor, check what auction the inventory is actually clearing in.
Bid shading
An algorithmic adjustment, usually applied by a DSP, that lowers a buyer’s effective bid in a first-price auction based on what the model expects the winning bid to be. The idea is to keep some of the surplus that first-price would otherwise transfer to the publisher. An Adtech take rate booster
In practice. Every major DSP has its own bid-shading logic, and none of them will tell you exactly how it works. Typical shading savings in open-exchange first-price auctions sit between 5-12% of the gross bid, with the highest savings on inventory with the most price volatility. PMPs and direct deals compress that spread to low single digits. Worth knowing the mechanism exists, worth being sceptical when your DSP claims efficiency wins that look suspiciously round.
Floor price
The minimum bid a publisher or SSP will accept for an impression. Set per-deal, per-format, per-buyer, or globally. Can be hard (no bid below this clears) or soft (bids below this enter a lower-priority pool).
In practice. Floors are the publisher’s main lever against bid suppression. They’re also the thing that makes “the CPMs in this PMP are too high” a slightly more nuanced complaint than it sounds.
Bid stream
The flow of bid requests sent from publishers and SSPs to buyers, containing the metadata that lets a DSP decide whether and how much to bid. Includes the URL, user signals, device, geo, and increasingly first-party audience data layered in via curation. Individual sell side platforms like Index Exchange handle upwards of 700 billion bid requests per day
In practice. Almost every interesting innovation in programmatic in the last five years has come from changes to what’s in the bid stream. Bombora surge data in the bid stream is a meaningfully different product to Bombora data sitting in a separate audience platform.
Win rate
The percentage of bid requests a buyer wins out of the requests they bid on. Reported by every DSP, interpreted by almost no buyer.
In practice. Low win rates can mean you’re underbidding, or that you’re bidding into supply you’ll never win because someone with better data is always above you. The diagnostic only works if you compare like-for-like supply paths, which most teams don’t.
In B2B, we deal with nail in a haystack audiences, so a low win rate can mean we are missing out on highly valuable audience opportunities. Its common that buyers bid too low to reach account based audiences in premium environments, and end up clearing with lower CPMs but in environments like online gaming, where CPMs are cheap but intent is non-existent. There’s a complex dynamic here to be worked through at a buyer level
Header bidding
A publisher-side technology that runs an auction across multiple SSPs and exchanges simultaneously, in the user’s browser or on the publisher’s server, before calling the ad server. Replaced the old waterfall logic where SSPs were ranked sequentially.
In practice. A modern publisher header bidding setup commonly auctions a single impression to 8-15 SSPs simultaneously, sometimes more once resellers are layered in. Each path takes a percentage. This is the structural reason open-exchange working media looks the way it does, and the reason supply path optimisation became a discipline rather than an option.
When I first entered the ad tech industry 15 years ago, individual publishers would work with individual sell-side platforms, meaning that sell-side platforms had distinction and unique value in the supply that they represented. Header bidding, in reality, completely blew open the doors on that value proposition and commoditised the amount of inventory available.
That means that today, in modern programmatic, you’ve got upwards of 50 sell-side platforms who are pretty much all representing the exact same inventory, creating many layers of duplication across their respective bid requests and massively inflating the volumes of impressions that DSPs like the Trade Desk and DV 360 have to listen to. This has created a world of supply path optimisation, curation, audience containerisation and sell-side predicated rebate structures in order to try and drive distinction in what is, in reality, a highly commoditised part of the ecosystem
Bid duplication
The phenomenon where the same impression opportunity reaches the same DSP through multiple supply paths simultaneously, producing several competing bids on a single auction.
In practice. Header bidding plus reseller relationships routinely create 3-7 duplicate paths to the same impression. The buyer effectively bids against themselves, the SSPs take their fees on each path, and the winning path is often not the most direct one. This is the specific waste pattern that supply path optimisation exists to fix, and the reason a duplicated path can look efficient on win rate while quietly inflating the buyer’s effective cost.
Supply Path Optimisation (SPO)
The buyer-side practice of choosing which SSPs and exchanges to bid through for any given piece of inventory, usually with the goals of reducing duplicate auctions, lowering fees, and preferring more direct paths to the publisher.
In practice. SPO is one of the most underused levers in B2B. Most B2B buyers have no view of how many hops sit between their DSP bid and the publisher’s ad server. The honest answer is usually three to five. Each hop takes a percentage, each hop introduces latency, and at least one of them is usually arbitrage. Even worse, signal gets lost or broken across these hops, creating less accuracy. Cleaning up the top ten supply paths typically recovers 8-15% of working media in the first quarter of focused effort.
2. Supply and inventory
What you’re actually buying when you buy programmatic. The vocabulary here has shifted more than any other cluster in the last three years.
Supply-Side Platform (SSP)
The publisher’s counterpart to a DSP. Manages the publisher’s relationships with multiple exchanges and demand sources, runs auctions on the publisher’s behalf, and reports back on revenue. Examples: Magnite, Index Exchange, OpenX, PubMatic
In practice. For most B2B buyers, the SSP is invisible — the DSP handles which SSPs it bids into. That invisibility is the reason supply path optimisation exists and the reason fees compound in places you can’t see.
Over about my first twelve years in the adtech ecosystem, I would have said that the sell-side platform is in danger of becoming a redundant part of the ad tech stack, symbolic in many ways of the duplication and bloat that exists in a world that should have few intermediaries. However, over the past probably three or four years, five at a push, we’ve seen the emergence of genuine innovation coming from the elite, probably the top half a dozen or so, of the supply-side platforms, led by the likes of Index Exchange, who have recently launched their Index Cloud.
We’re now seeing a lot of genuine differentiation and unique value coming from the ability to construct audiences in a nearly latency-free environment, living in the sell-side infrastructure, which is allowing us to run DSPs in containerised environments, like Bedrock recently announced. We’re also increasingly seeing a real shortening of the time that it takes to run an open RTB auction, from 200 ms round trips down to a fraction of that, which is opening up genuine creative optimisation opportunities, which is using that time that’s being clawed back. This is real grass roots innovation in adtech and gives me cause to believe that the ecosystem will evolve beyond the models that personified it even 5 years ago.
As SSPs have converged towards DSPs, and with examples at the trade desk and their open path solution converging back towards the supply side, there is actually a lot less distinction today between an SSP and a DSP than there was 10 to 15 years ago. This is definitely a part of the ecosystem to keep a really close eye on, where genuine and true innovation is emerging
Ad exchange
A marketplace where publisher inventory and buyer demand meet. Often used interchangeably with SSP, though strictly an exchange is the auction layer and an SSP is the publisher-facing technology that plugs into it. I’d venture the ad exchange is redundant today because of header bidding and SPO efforts.
In practice. The distinction matters less in 2026 than it did a decade ago, because most major players run both. You will see them listed separately in reporting.
Demand-Side Platform (DSP)
The buying technology that places bids into auctions across multiple SSPs and exchanges. Holds the campaign setup, audience definitions, frequency caps, budget pacing, and creative. Examples: The Trade Desk, DV360, StackAdapt, Yahoo DSP, Adform.
In practice. The choice of DSP is the most consequential decision in a B2B programmatic stack and the one most often made for the wrong reasons. The right DSP for B2B is the one with the best access to the data signals you actually need to activate, the cleanest supply, the most flexible deal architecture, and a fee structure you can read without a lawyer.
The DSP has been the king of the ad tech world for the last 15 years, and that’s only really started to change since the emergence of innovation coming from the supply-side layers over the past half a decade or so. Since then, we’ve actually seen some real pressure coming onto the top DSPs for the first time. Trade desk, for example, has been the darling of the NASDAQ, representing ad tech with ferocious lower-left-to-upward-right growth over the past five years. That’s coming under some pressure now for the first time, and it’s symbolic really of changes that are coming between agentic buying, which is disintermediating a lot of hands on the keyboards with DSPs and with it some of their value, perceived or otherwise. The emergence of better curation and increasingly containerised audiences is now meaning that the reality of running a DSP is that you’re actually using it for a lot less than you used to use it for. If in the past we were leaning on it for setting up campaigns, building audiences, defining those, and optimising, we’re increasingly now doing a lot of that audience building and construction and refinement in the supply-side layers. What we need a DSP for is less and less, which is putting downward pressure on the fee structures, and we’ve seen that kind of leading to fewer players in the DSP market really over the past few years
Open exchange
The default inventory pool available to any buyer who can connect to an SSP. No private negotiation, no audience curation, no minimum spend commitment. Sits at the bottom of most modern supply hierarchies.
In practice. Open exchange in B2B is mostly a poor use of money. The supply is variable quality, the audience signal is degraded, and the MFA exposure is real. Most disciplined B2B buyers route ninety percent or more of their spend through curated or private inventory.
Open web
The collection of publisher websites and digital environments accessible through standard programmatic infrastructure, as distinct from authenticated, closed platforms. Where most independent publishing lives.
In practice. “Open web” is a term that has been doing more rhetorical work in the last few years as walled gardens have grown. For B2B buyers, the open web is where most of the high-trust trade and business press lives, and it remains the only environment where you can apply your own audience and curation logic at scale.
Walled garden
A closed platform that controls its own inventory, audience data, identity, and (usually) measurement, with limited ability for external tools to operate inside. The largest are Google, Meta, Amazon, and LinkedIn, with TikTok and Reddit increasingly in the same category.
In practice. B2B campaigns spend a meaningful share of total budget inside walled gardens, especially LinkedIn. The trade-off is structural: better authenticated targeting and inside-the-platform measurement, against limited cross-platform reporting, no log-level transparency, and audience data that doesn’t leave the garden. The right B2B stack treats walled gardens as one tool among several, not as the whole tool.
Private Marketplace (PMP)
An auction-based deal between a publisher (or curator) and a defined set of buyers, accessed via a deal ID. More restricted than open exchange but still auctioned. Usually has an agreed floor price and audience or content scope.
In practice. PMPs have largely become the default in B2B because they’re the easiest way to enforce content quality and apply curated audience layers without committing to fixed media volume.
Deal ID
A unique alphanumeric identifier issued by a publisher, SSP, or curator that gives a buyer access to a specific package of inventory, audience, pricing, or deal mechanics. The plumbing that makes everything other than open exchange work.
In practice. Most B2B deal IDs in 2026 aren’t really about premium publisher access. They’re audience containers. The publisher list inside the deal is often a thousand sites long; what makes the deal valuable is the data layer applied to it. Judging a deal ID by its publisher roster misses the point.
Programmatic Guaranteed (PG)
A non-auctioned deal where a buyer commits to a fixed volume of impressions at a fixed price, executed programmatically through a deal ID. The closest programmatic gets to a traditional insertion order.
In practice. PG is useful when you need delivery certainty: a launch window, a heavyweight execution, a specific high-impact placement. It’s expensive per impression and inflexible on optimisation, so most B2B campaigns use it for ten to twenty percent of budget at most.
Preferred Deal
A non-guaranteed, non-auctioned deal between a buyer and a publisher at a fixed CPM, where the buyer has first refusal on the inventory before it goes to PMP or open exchange. Has largely fallen out of fashion.
In practice. Preferred deals are rare in modern B2B stacks because they offer the worst of both worlds: no auction efficiency, no delivery guarantee.
Curation / Curated Marketplace
A practice where a third party assembles a package of inventory across multiple publishers, applies an audience or content layer, and exposes it to buyers as a deal ID or PMP. Examples of curators: Audigent, Chalice AI, Multilocal, and a growing roster of agency- and operator-owned curation businesses.
In practice. Curation is the single most important shift in B2B programmatic of the last three years. The mechanics are simple: take inventory the buyer could already access, layer signal the buyer couldn’t easily apply themselves, expose it as a single deal ID with predictable economics. Most curated marketplaces in B2B carry an additional 10-25% on the media cost in exchange for the audience and content layer. Whether that’s value depends entirely on what the curator is actually applying. The difference between a thoughtful B2B curation, built on raw signal access and properly maintained allow-lists, and a generic audience overlay sold under the same name is a factor of three in performance and often the reason a campaign clears or doesn’t.
Made-for-Advertising (MFA)
Sites that exist primarily to generate ad impressions rather than to host genuine editorial. Characterised by heavy ad-to-content ratios, slow load times, clickbait headlines, low session quality, and outsized share of programmatic inventory volume.
In practice. Research from Adalytics, the ANA, and Jounce Media across 2023 to 2025 estimated MFA absorbing 15-21% of open-exchange spend industry-wide. The pattern inside a single B2B campaign is usually worse than the average: MFA exposure concentrates in a small number of high-volume domains that look respectable in a list and disappear under a competitive set check. I’ve reviewed B2B open-exchange buys where a single unfamiliar domain accounted for nearly 20% of impressions, and the buyer had never heard of it. The defences are unglamorous: a clean exclusion list refreshed monthly, a preference for curated supply, and at least quarterly log-file audits.
Brand safety
The discipline of preventing ads from appearing alongside content that’s harmful to the advertiser’s reputation. Includes hard exclusions (illegal content, hate speech) and soft preferences (avoiding war coverage for a leisure brand, for example).
In practice. B2B brand safety is mostly a B2C-defined toolset applied to a different problem. The harder question for B2B is brand suitability: whether the publication is one your buyer respects. That’s a different filter to whether the article mentions a sensitive topic.
Brand suitability
The discipline of selecting environments that are positively appropriate for an advertiser’s brand and audience, rather than simply not unsafe. Distinct from brand safety, which is the floor; suitability is the ceiling.
In practice. Most modern B2B buyers should care more about suitability than safety. The risk in B2B advertising isn’t usually that your ad runs next to upsetting content; it’s that your ad runs across a long tail of low-attention, low-credibility sites that quietly erode the perception your media is trying to build. Suitability filtering is what curated marketplaces and quality-led private deals are really selling.
Invalid Traffic (IVT) / ad fraud
Traffic on advertising inventory that isn’t a real human seeing the ad. Includes bot traffic, scripted impressions, hidden ads, click farms, domain spoofing, and various flavours of inventory fraud. IAB Tech Lab distinguishes General Invalid Traffic (GIVT, detectable by basic filters) from Sophisticated Invalid Traffic (SIVT, requiring more advanced detection).
In practice. Industry-reported IVT in open programmatic has ranged from low single digits to double digits depending on methodology and year, with Pixalate, DV, and IAS reporting somewhat different numbers from somewhat different vantage points. The honest read is that B2B campaigns running through curated supply and verified inventory see meaningfully lower IVT exposure than equivalent open-exchange buys, often by a factor of three or more. Most B2B teams have never asked their DSP to report IVT separately by supply path, which is the only view that tells you which paths are actually problems versus which are clean.
Verification (DV, IAS, MOAT)
Third-party services that independently measure viewability, IVT exposure, brand-safety adjacency, and other supply quality metrics for advertisers. The three major players are DoubleVerify, Integral Ad Science, and Oracle MOAT.
In practice. Verification has been a category of structural friction in programmatic for over a decade. The fees are real, the protection is real, and the over-blocking is real too. Most enterprise B2B campaigns run with at least pre-bid verification on; whether you also run post-bid measurement is usually a budget question.
Publisher-Provided Identifier (PPID)
A persistent identifier set by a publisher for its own users, passed into the bid stream to enable identity continuity without third-party cookies. Implementation patterns vary by publisher.
In practice. PPIDs have quietly become one of the more important post-cookie identity signals in premium publisher environments. For B2B, the value sits in how well the publisher’s authenticated user base maps to your target audience. A trade publication with ninety percent registered users and a strong B2B reader base is a different identity proposition to a general-interest site with low log-in rates.
Viewability
The measure of whether an ad was actually rendered in a viewable position on the page. The IAB standard is fifty percent of pixels in view for one continuous second for display, and two continuous seconds for video.
In practice. Viewability is a hygiene metric, not a performance one. High viewability is necessary but not sufficient. Most B2B buyers should be targeting seventy percent or higher on display and ninety-plus on video, and treating sub-fifty percent placements as something to diagnose, not optimise around.
ads.txt and sellers.json
Publisher-side files that declare who is authorised to sell a publisher’s inventory programmatically. ads.txt is a plain-text file in the root of the publisher’s domain. sellers.json is the SSP-side counterpart that declares the seller identity behind each authorised path.
In practice. Together they let buyers verify that the path they’re bidding through is legitimate. Useful technically. Less interesting commercially than they sound, but absence of either is a hygiene red flag.
3. Buying-side architecture and economics
The tech and money flowing inside the buyer’s half of the stack. Where most of the fees live, and where most of the fees aren’t visible.
Ad server
The technology that holds the ad creative, decides which version to serve in any given impression, and tracks delivery. Sits between the DSP decision and the user’s screen. Examples: Google Campaign Manager (CM360), Flashtalking, Sizmek (now Amazon Ad Server).
In practice. Most B2B buyers use a single ad server across all DSPs, which is the right call because it makes cross-DSP reporting possible.
Insertion Order (IO)
The contractual document covering a media buy. Specifies the campaign, the budget, the flight dates, the targeting, and the creative. The bridge between sales and execution.
In practice. Programmatic was sold for years as the death of the IO. In B2B it never really happened. Most enterprise B2B campaigns still run on an IO, often a multi-channel one that wraps programmatic, search, social, and direct deals into a single document.
Take rate
The percentage of media spend taken by the DSP (or any other intermediary) as a fee. Disclosed in some commercial structures, opaque in others.
In practice. DSP take rates vary wildly. A self-service Trade Desk seat in a major holding company might run a take rate in the high single digits. A small advertiser on a managed-service contract can be paying twenty-plus percent before any other fees layer on.
Tech fee
A catch-all for any fee charged by a piece of technology in the buying stack. Includes DSP take rate, ad verification fees, brand safety tooling, audience platform fees, and identity resolution fees.
In practice. A meaningful share of B2B programmatic spend dissolves into tech fees that the marketer never sees on a single line item. Working through a programmatic invoice and rebuilding the fee waterfall is the single most useful exercise a B2B CMO can do this year.
Working media
The portion of total budget that actually reaches a publisher in exchange for an impression. The inverse of all fees, taxes, and intermediation costs above it.
In practice. Most B2B programmatic stacks operate at 55-70% working media when measured rigorously through to publisher payout. Genuine 80%+ working media exists only in heavily curated, direct, or short-supply-chain configurations. Anything claimed above 90% by a managed-service vendor without log-level evidence to back it deserves polite scepticism.
Data CPM
A fee charged per thousand impressions for applying a data segment, audience, or signal layer to a buy. Distinct from the media CPM.
In practice. Data CPMs stack. A typical B2B impression might carry the publisher’s CPM, plus a curation fee, plus a Bombora intent layer at a data CPM, plus a third-party brand safety verification fee, plus the DSP take rate. Each line is small. The compound is not.
Programmatic tax
Informal term for the cumulative load of fees, intermediation costs, and tech charges between the buyer’s budget and the publisher’s revenue. Sometimes called the “ad tech tax.”
In practice. The 2020 ISBA / PwC study put the programmatic supply chain tax at around fifty percent in some open-exchange paths, with around fifteen percent of spend unaccounted for entirely. More recent work from the ANA, Adalytics, and the IAB suggests the median has improved somewhat in B2C as direct deals have grown. Comparable B2B-specific data is thinner. Independent log-file analysis remains the only reliable way to know your own number.
Supply path reconciliation
The process of comparing what your DSP says it bought against what the publisher’s ad server records as delivered, then identifying the supply paths, fees, intermediaries, and audience layers that compose each impression.
In practice. Reconciliation is the only honest way to know what you actually bought, and it’s expensive, slow, and rare. The first reconciliation a B2B team runs almost always surprises everyone in the room: a path the agency swore was direct turns out to have two intermediaries, a CPM the buyer assumed was the publisher’s payout is actually the gross before fees, and the supply mix on a “premium” buy includes inventory the buyer would not have bought knowingly. The B2B teams that have done this exercise once tend to make different supply decisions afterwards.
Self-service vs managed service
Two commercial models for accessing a DSP. In self-service, the buyer holds the seat and runs the campaigns directly. In managed service, an agency or specialist operator holds the seat and runs the buying on the buyer’s behalf, usually for a percentage fee.
In practice. The right answer depends entirely on internal capability. Self-service looks cheaper on paper and is more expensive in reality if you don’t have a senior trader in-house. Managed service is more expensive on paper and is often the right call for B2B teams running campaigns at moderate scale who don’t want to staff a trading desk.
Pacing
The pattern in which a campaign’s budget is spent across its flight. Even pacing spreads spend uniformly; ASAP pacing spends as fast as supply allows; smart or accelerated pacing variants concentrate spend around predicted high-performance windows.
In practice. Pacing is one of the most under-managed dimensions of B2B campaign delivery. A B2B audience is rarely active uniformly across the working week, and almost never on weekends. Letting a DSP pace evenly across seven days means spending a meaningful share of your budget against a non-buying audience. Day-parted and weekday-weighted pacing remains a basic discipline most teams skip.
Pixel
A small piece of code, usually a transparent image or JavaScript tag, placed on a webpage to record events: page views, conversions, audience membership, time on page. The original mechanism for marketing measurement on the open web.
In practice. Pixels have been gradually displaced by Conversion API server-side equivalents as browser-level protections have eroded their reliability. They still work in many contexts and remain the default in plenty of campaigns. Treating pixel data as ground truth in 2026 is generous; treating it as one signal among several is sensible.
Conversion API (CAPI)
A server-to-server method for sending conversion events from an advertiser’s environment directly to an ad platform, bypassing the user’s browser. Implemented by Meta as the original CAPI, with equivalents across most major DSPs and walled gardens.
In practice. Conversion APIs have become essential as cookie deprecation and tracking-prevention features have made pixel-based event capture unreliable. For B2B, CAPI matters most for closed-loop signal back to walled gardens (especially LinkedIn) and for feeding cleaner data into optimisation algorithms that would otherwise be working from degraded inputs.
4. Identity and audience
The plumbing that lets you target a specific person, or company, or device. Quietly rebuilt over the last five years as the third-party cookie has eroded.
Third-party cookie
A small text file stored in a user’s browser by a domain other than the one being visited. The mechanism that underpinned most of cross-site tracking, retargeting, and audience matching in display advertising for two decades.
In practice. The third-party cookie is dead in Safari and Firefox and on its way out in Chrome, with Google’s deprecation timeline having shifted multiple times. Treating cookies as a primary identity strategy in 2026 is professional negligence. They still work in some contexts, but planning a programme around them is planning to be wrong in eighteen months.
Privacy Sandbox
Google’s collection of replacement technologies for third-party cookies in Chrome. Includes the Topics API for interest-based targeting, Protected Audience for remarketing, and Attribution Reporting for measurement.
In practice. Privacy Sandbox is real, it works at a basic level, and almost nobody in B2B is using it as their primary identity strategy because the targeting precision is far below what B2B campaigns require. It will matter more for B2C retail than for B2B.
Mobile Advertising ID (MAID)
Persistent identifiers issued by mobile operating systems for advertising purposes. GAID (Google Advertising ID) on Android, IDFA (Identifier for Advertisers) on iOS.
In practice. IDFA has been functionally dead since Apple’s App Tracking Transparency rollout in 2021, with opt-in rates running at twenty to thirty percent in most categories. GAID still works but Google has signalled changes. For B2B, mobile identifiers are rarely the primary signal anyway, because most of the relevant research and buying activity happens on desktop or in authenticated web environments where MAIDs aren’t the working identifier.
The practical value is mapping desktop signals to MAIDS, to extend targeting into mobile environments and to execute location based tactics
Hashed email (HEM)
An email address run through a one-way cryptographic hash function, producing a unique string that can be matched across systems without exposing the underlying email.
In practice. Hashed emails are the de facto persistent identifier in B2B because B2B is a logged-in, email-driven world. Most modern B2B identity resolution starts from a hashed email and builds outwards.
Unified ID 2.0 (UID2)
An open-source, deterministic identifier built from hashed and salted email addresses or phone numbers. Maintained by the IAB Tech Lab, originally developed by The Trade Desk. Designed as an industry-wide cookie replacement.
In practice. UID2 has real adoption in B2C, especially in CTV and authenticated environments. In B2B the picture is patchier because B2B sites have lower authentication rates than streaming services or retailers. Useful as one identifier in a stack, not as the only one.
RampID
LiveRamp’s persistent, people-based identifier. Built from authenticated and offline data sources, resolved into a single ID that can be activated across most major DSPs and SSPs.
In practice. RampID has the widest deployment of any commercial identity solution across the open programmatic ecosystem. For B2B specifically, the value is in how well RampID resolves your CRM list against the addressable identity universe inside major DSPs. Onboarding fees, ongoing activation CPMs, and minimum commitments layer on top of media costs in ways that aren’t always visible inside a managed-service IO. Worth a service-side resolution check against your CRM file before contract — the gap between “your data is supported” and “your data resolves usefully” is real, and a five-minute conversation can save a five-figure surprise.
As of May 2026 they have been purchased by Publicis, with unknowns over whether the technlogy will be gated for their customers or not.
ID5
A commercial identity solution that operates as a probabilistic and deterministic identifier across the open web, particularly in environments where authentication is low.
In practice. ID5 sits in a part of the identity stack that most B2B planners never directly interact with, but the resolution rate of the identifier in any given supply path can materially affect addressable reach. Worth asking your DSP or curator which alternative IDs are being recognised inside the bid request, not just which ones are notionally supported.
Identity graph
A database that links multiple identifiers (cookies, MAIDs, hashed emails, RampIDs, UID2s, account-level signals) to a single resolved person or account. The connective tissue beneath modern audience activation.
In practice. Every meaningful audience platform in B2B sits on top of an identity graph of some kind. The quality of the graph determines the quality of every downstream activation. The graph is also the most opaque part of most vendors’ stacks.
Match rate
The percentage of records in a target audience that can be matched to a usable advertising identifier inside the activation environment. Usually expressed as a percentage of the input list.
In practice. Match rate is the most-quoted and least-honest metric in B2B audience activation. Vendor-quoted match rates of 70-80% rarely survive an honest comparison to working ID-level activation rates at the impression level. Realistic ranges in our own book: 22-38% on open-web display, 35-55% on CTV in authenticated environments, 60-75% inside LinkedIn against a well-maintained CRM list, below 15% on mobile in-app. The honest test isn’t the vendor’s quoted match rate, it’s the impressions actually delivered against the target list. Most teams have never asked their DSP to report against that denominator. They should.
Authenticated traffic
Inventory served to users who are logged in to the publisher or platform, providing a persistent first-party signal that doesn’t rely on cookies or MAIDs. Includes most CTV, streaming audio, walled-garden social, and authenticated publisher environments.
In practice. Authenticated traffic is the only segment of the open programmatic supply where identity is genuinely durable. B2B buying stacks should be tilting hard towards authenticated supply where the audience fits.
Fingerprinting / probabilistic identity
Identification methods that combine non-personal signals (browser, device, IP, behaviour) to identify a user with high probability without using cookies or MAIDs.
In practice. Fingerprinting has been pushed to the margins by browser-level countermeasures and by regulator scrutiny. It still operates, particularly in measurement and fraud-detection contexts. Most reputable B2B buying stacks no longer rely on it for primary targeting.
Data onboarding
The process of taking offline or first-party data (a CRM list, a CSV of accounts, a customer file) and matching it into the digital identity environment so it can be activated across DSPs and walled gardens.
In practice. Onboarding is where match rate becomes a daily reality. The shorter the path from your first-party data to the impression decision, the better the match. Multiple onboarding layers compound match-rate decay.
Data Management Platform (DMP)
A category of technology, dominant from roughly 2012 to 2020, that aggregated audience data from multiple sources, built segments, and pushed them out to DSPs and other activation tools. Examples: Salesforce DMP (Krux), Oracle BlueKai, Lotame, Adobe Audience Manager.
In practice. DMPs have been quietly dismantled as a category. The third-party cookie collapse undermined the cross-site audience-building that justified them, and modern CDPs have absorbed most of the use cases that survived. The term still appears in B2B procurement documents and platform overviews, mostly out of habit.
Customer Data Platform (CDP)
A category of technology that unifies first-party customer data from across an organisation (CRM, web behaviour, email engagement, product usage, support interactions) into persistent, ID-based profiles, then makes those profiles available to activation channels. Examples: Segment, mParticle, Tealium, Treasure Data, Salesforce CDP.
In practice. CDPs have become foundational infrastructure for any B2B marketing organisation serious about activating first-party data into advertising. The hard part isn’t usually the CDP itself; it’s the data hygiene work upstream and the integration work downstream that makes the CDP useful. Buying a CDP without committing to that work tends to produce a more expensive version of the data mess you already had.
Retargeting / remarketing
The practice of serving ads to users who have previously interacted with a brand: visited the website, opened an email, watched a video, attended a webinar. Sometimes used interchangeably; technically Google reserves “remarketing” for its own ecosystem.
In practice. Retargeting has been the most over-relied-on tactic in B2B advertising for years. Most B2B retargeting audiences are small (because the qualifying behaviour is rare), saturate quickly (because the same handful of users are reached over and over), and produce attributed conversions that are heavily inflated by non-incremental activity. A reasonable rule of thumb: if a controlled holdout test on your retargeting line shows incremental lift below 30%, the line is mostly attribution theatre. Useful as one layer in a wider stack. Not useful as the spine of a programme, and not useful as a budget defence in front of a finance team that asks the right questions.
Lookalike modelling
A method for finding new audiences that resemble an existing high-value audience, using statistical similarity across behavioural, firmographic, or other attributes. The original at-scale audience extension technique.
In practice. Lookalike modelling in B2B is more limited than in B2C because the seed audiences are smaller, the firmographic universe is narrower, and the platforms with the best lookalike models (Meta, Google) aren’t usually where the B2B buying decision-makers live. Better deployed inside B2B-native data environments where the modelling can use firmographic and technographic signal, not just generic behaviour.
Audience extension
The practice of extending a known audience (a target account list, a CRM segment) into adjacent users or accounts that share defining characteristics but weren’t on the original list. Distinct from lookalike modelling in that extension is usually rule-based (industry, size, technology footprint) rather than purely statistical.
In practice. Audience extension is one of the more useful B2B audience disciplines and one of the most under-used. A target list of 500 accounts can be extended via firmographic and technographic logic into a workable activation universe of 2,000 to 5,000 accounts without sacrificing definitional precision. The constraint isn’t usually the data; it’s the planning rigour to define the extension rules properly.
5. B2B-specific vocabulary
The terms that don’t translate cleanly from B2C. This is where the B2B operator’s job really starts.
Account-Based Marketing (ABM)
A marketing approach that targets specific named accounts as the unit of attention, rather than individuals or generic personas. Coordinates marketing and sales activity around a defined account list.
In practice. ABM as a category includes everything from a sales-led named-account programme with no media at all, through to fully-funded programmatic ABM running against thousands of accounts in parallel. The category label is doing too much work. When someone says “ABM” the first question is which version.
Ideal Customer Profile (ICP)
A definition of the kind of account most likely to become a high-value customer. Built from firmographic, technographic, and behavioural attributes.
In practice. Most B2B teams have an ICP. Few of them have an ICP that’s been updated more than once a year. Static ICPs miss the second-order shifts that show up in pipeline data months before they show up in strategy decks.
Target Account List (TAL)
The named accounts a campaign is designed to reach. Sits downstream of the ICP, sometimes refined further by sales prioritisation or current pipeline status.
In practice. A TAL of one hundred accounts is a different campaign to a TAL of ten thousand. The ratio of TAL size to Total Addressable Market drives almost every downstream activation decision: channel mix, frequency, creative variation, measurement model.
Total Addressable Market (TAM)
The total universe of accounts that could in principle become customers, before any targeting or prioritisation is applied. The denominator that ICP, TAL, and account-list strategy all sit inside.
In practice. Most B2B TAM estimates are built top-down from industry and employee-count filters and tend to be either confidently wrong or comfortably round. A more useful definition for media planning is the operational TAM: the subset of total TAM that you can actually reach with available data, identity, and activation infrastructure. That number is usually meaningfully smaller than the strategy-deck TAM, and the gap is where most B2B media plans quietly break.
Firmographic
Attributes that describe a company: industry, employee count, revenue, geography, ownership structure. The B2B equivalent of demographics.
Technographic
Attributes that describe a company’s technology stack: which CRM they use, which marketing automation platform, which cloud provider, which DSP. Sourced from a combination of detection technology and self-declared data.
In practice. Technographic targeting is one of the most commercially relevant signals in B2B and one of the most variable in quality. Detection methods range from genuinely robust (job-listing scraping cross-referenced with technology footprint scans) to wishful (one mention of a product in a single web page).
Intent data
Behavioural signals indicating that an account is researching or considering a particular topic, product, or category. Sourced from publishers, B2B media networks, search behaviour, and proprietary research environments.
In practice. Intent data has become table stakes in B2B, and the quality variance between providers is the largest unexamined variable in B2B targeting. Same nominal data category from three different providers, three almost entirely different account universes. The difference between a co-op of a few hundred publishers and a network of several thousand is structural. The difference between raw signal access and a pre-scored aggregate is larger again. Run the overlap analysis yourself before believing any vendor’s coverage claim; the results tend to embarrass everyone involved.
Surge
A spike in intent activity on a particular topic by a particular account, relative to that account’s own baseline. The mechanism by which intent providers separate “engaged” from “always engaged.”
In practice. Surge is the most-quoted metric in intent reporting and also the most coarse. A surge based on three articles read by one anonymous user at a 500-person company is treated identically to a surge based on twenty articles read across multiple users at the same company. The underlying volume matters.
Bombora
The largest B2B intent data co-op in the market. Operates a network of around 5,000 B2B publishers and provides intent topic data at the company level. Most major B2B platforms (6sense, Demandbase, ZoomInfo, and others) license Bombora data as a component of their offering.
In practice. Most B2B teams encounter Bombora data through a downstream platform (6sense, Demandbase, ZoomInfo, or similar), not directly. The taxonomy covers roughly 14,000 B2B topics; surge is calculated against a rolling account-level baseline of around 12 weeks. The downstream platforms see the surge score and the topic, then apply their own modelling. Operators with raw signal access see additional dimensions (signal weight, contributor diversity, topic clustering, source publisher) that downstream platforms quietly compress out. The shift to raw signal-level Bombora access has been the basis of meaningful B2B advertising differentiation for the small number of operators with that relationship.
Dark funnel
The portion of buyer behaviour that happens outside any channel you can directly measure: Slack communities, peer recommendations, podcast listening, AI-driven research, private analyst conversations. Sometimes called “dark social” or “self-directed research.”
In practice. Dark funnel activity is now the dominant mode of B2B buying for most categories, particularly software. Marketing teams that build their plans around what they can attribute, attribute themselves into smaller and smaller corners of the actual buyer journey.
Buying committee
The group of stakeholders involved in a B2B purchase decision. Typically six to twelve people for enterprise software, more for capital purchases, fewer for SMB or self-serve products.
In practice. Most B2B campaigns are built creatively for one persona — usually the perceived decision-maker — and then served to whoever is in the audience. Better practice is to plan for the committee from the start: different creative for different roles, different signals, different sequencing.
Account graph
A data structure linking individual identifiers (cookies, IDs, hashed emails, mobile IDs) to the company-level entity they belong to. The B2B-specific equivalent of an identity graph.
In practice. The account graph is the prerequisite for any serious B2B audience activation. Without it, you can target known persons (which is small) or anonymous traffic (which is unfocused). With it, you can target accounts.
Signal
Loose industry term covering any behavioural or attribute data point indicating a possible buyer state. Includes intent topics, surge data, technographic changes, news events, hiring signals, funding signals, web behaviour, and email engagement.
In practice. “Signal” has become the dominant vocabulary in B2B marketing because it captures the shift from MQL-style discrete events to continuous, multi-source evidence. The risk is that the term has been stretched so far that it now means everything and nothing. Useful signal is signal that’s specific, recent, and weighted against an account baseline.
Engagement scoring
A model that aggregates multiple signals into a single score representing the likelihood or readiness of an account to engage. Underpins most account prioritisation in modern B2B systems.
In practice. Most engagement scoring models are additive (more signal equals more score). The better models are multiplicative — they require multiple corroborating signal types to fire before scoring an account high. Multiplicative models produce smaller, more reliable target lists. Additive models produce big lists with a lot of noise.
Account-level frequency
The number of times any user inside a target account is exposed to a campaign, regardless of which specific person. The right unit of frequency measurement in B2B, and the one almost no DSP reports on natively.
In practice. A campaign that hits a target account 12 times across four different employees is a different experience to a campaign that hits the same employee 12 times. The first looks like presence; the second looks like harassment. Most DSPs cap at the cookie or device level, which is neither of these. Getting to true account-level frequency reporting usually requires either a B2B-native activation layer that resolves identifiers to the account before delivery, or a custom log-level analysis after the fact. Worth the effort: it’s the metric that turns “we’re running ABM” into “we know what our target accounts have actually been exposed to.”
Signal decay
The rate at which an intent or behavioural signal loses predictive value as time passes from when it fired.
In practice. Most B2B teams treat signals as durable for weeks or months because that’s the cadence at which platforms surface them. The honest decay curve is much steeper. Research-based intent signals typically lose half their predictive value inside 14-21 days; account-level surge signals can decay within a week if not corroborated by additional activity. Acting on a four-week-old signal is acting on a different account state than the one that generated it. The activation cadence should match the decay curve, not the platform’s refresh cycle.
6sense and Demandbase
The two largest B2B “ABM platforms,” each combining intent data, account identification, audience activation, and (in some configurations) advertising execution. Heavy enterprise footprint. Often treated as default infrastructure inside larger B2B marketing organisations.
In practice. Calling these platforms “ABM platforms” understates what they actually are: they’re data-and-orchestration layers that sit between a CRM, a marketing automation platform, and the activation channels. Whether you also use them as your advertising execution engine is a separate, often debatable, decision. The intent data refresh cadences differ from each other and from raw co-op feeds; for time-sensitive opportunities the latency between buying-committee research activity and signal appearing in the platform can be days to weeks. For an account already in active pipeline that’s fine. For competitive opportunities where you’re trying to surface a new account before a rival does, it usually isn’t.
Demand generation
Marketing activity aimed at building awareness, interest, and consideration across a target audience, typically over longer time horizons and broader audiences than direct response. The B2B equivalent of brand and mid-funnel activity combined.
In practice. “Demand gen” in B2B has come to mean almost anything that isn’t pure direct response or sales-led outbound. The risk is that the term swallows so much activity that none of it is held to a sensible measurement standard.
Lead generation
Marketing activity focused on capturing a specific contact’s details, usually via a gated asset, form fill, or sign-up. The historic engine of B2B marketing pipelines.
In practice. Lead generation as a standalone discipline has been quietly losing ground for years. Most B2B buyers will not fill in a form to access a piece of content that’s freely available elsewhere. Lead-gen as a tactic still works in narrow contexts (high-value gated research, webinars with genuine subject-matter authority). Lead-gen as a marketing organising principle is past its sell-by date.
MQL (Marketing Qualified Lead)
A lead that meets a defined set of behavioural or firmographic criteria, indicating to marketing that they should be passed to sales for follow-up. Historically the unit of handoff between marketing and sales.
In practice. MQLs were designed for a world where filling in a form was a strong signal of interest. That world isn’t this one. Many B2B teams have moved or are moving to account-level handoffs (Marketing Qualified Account, or MQA) or signal-based handoffs, where the unit isn’t a single contact but a pattern of activity at the account.
6. Data and signal sources
The categories of data that flow into a B2B activation, and the structures that hold them.
First-party data
Data collected directly by the company that owns the relationship: CRM records, website behaviour, email engagement, product usage, customer-service interactions. The most reliable data type, and the most under-utilised in most B2B stacks.
In practice. First-party data is the single most defensible audience asset a B2B marketing organisation has, and the one most often left sitting in a CRM rather than activated. The work to make it usable (clean records, persistent identifiers, identity resolution, consent management, activation pipes) is unglamorous and slow. The marketing teams that have done it have a structural advantage over the ones that haven’t.
Second-party data
Another company’s first-party data, made available to you under a direct commercial arrangement. Rare in B2B because most companies are wary of sharing customer-level data even with partners.
In practice. Second-party arrangements are growing in B2B as data clean rooms make them mechanically possible without raw data being exchanged. The most common B2B example is event partnerships, where a conference organiser shares attendee or session data with sponsors under a clean-room contract. Worth pursuing when the partner has audience overlap you can’t otherwise reach.
Third-party data
Data licensed from a vendor that has aggregated it from many sources. Includes intent providers (Bombora, G2, TechTarget), data brokers, firmographic providers (ZoomInfo, Dun and Bradstreet), and technographic providers (HG Insights, Slintel, BuiltWith).
In practice. Third-party data is the default in B2B activation because first-party data is usually too small and second-party data is structurally hard to share. The quality varies. Treating all third-party data as equivalent is one of the most common mistakes in B2B planning.
Data clean room
A secure environment where two or more parties can match and analyse data at the user or account level without either side seeing the raw records. Examples: Snowflake clean rooms, AWS Clean Rooms, LiveRamp Clean Rooms, Google Ads Data Hub.
In practice. Clean rooms have moved from “interesting concept” to “infrastructure you need to plan around” in the last two years. For B2B specifically, the use cases are mostly around closed-loop measurement and audience overlap analysis with key publisher partners. The maturity of the tooling is improving fast.
Data co-op
An arrangement where multiple companies contribute data to a shared pool in exchange for access to the aggregate. Bombora’s B2B publisher network is the canonical example: publishers contribute reading behaviour, Bombora resolves it to account-level intent signal, all contributors get access.
In practice. The economics of co-ops only work if the data being contributed is roughly fungible. They tend to break down when one participant contributes substantially more or higher-quality data than they extract.
Topic taxonomy
A structured list of topics or interests against which intent or content data is classified. Bombora’s taxonomy covers several thousand B2B topics. The IAB Content Taxonomy is a broader-purpose framework.
In practice. The granularity of the taxonomy matters. Targeting “cloud infrastructure” is meaningfully different to targeting “Kubernetes managed services for financial services.” Some platforms allow custom-topic creation, most do not.
7. Channels and formats
The places programmatic shows up. Each channel has its own quirks for B2B specifically.
Display
Banner advertising on web pages. The original programmatic channel. Includes a range of standard IAB-defined ad sizes, with the medium rectangle (300x250) and the leaderboard (728x90) being the most common.
In practice. Display in B2B is rarely the channel that drives the headline number, but it’s almost always the channel that sustains presence at scale. The mistake is judging it on the same metrics as paid search.
Video (in-stream and out-stream)
Video advertising delivered programmatically. In-stream sits inside other video content (pre-roll, mid-roll, post-roll). Out-stream sits inside non-video environments, usually as a player embedded in an article.
In practice. In-stream video on premium B2B publishers is a strong format for awareness and brand-building. Out-stream is more variable, with quality dependent on the publisher’s player implementation and viewability discipline.
Connected TV (CTV)
Programmatic video delivered to a streaming environment on a television: smart TV apps, streaming sticks, gaming consoles. Authenticated, large-screen, full-attention.
In practice. CTV in B2B is mostly used for executive targeting and high-consideration awareness against named accounts. Reach is real but small relative to display. Typical B2B CTV CPMs in 2026 sit between £35-£55 for premium streaming environments, with sports and news inventory pushing higher. The economics make sense for high-ASP categories where individual account influence is worth £25-£50 per impression on a small TAL. They rarely make sense for high-volume mid-market campaigns where the same budget against the same outcome works harder in display and audio.
OTT (Over-The-Top)
A broader category covering any video content delivered over the open internet rather than through traditional cable or broadcast distribution. Includes CTV (television-screen viewing) as well as mobile and desktop streaming.
In practice. OTT and CTV are often used interchangeably in casual conversation; the distinction matters when you’re planning specifically against the big-screen environment versus the broader streaming universe. For B2B campaigns aimed at executive audiences, the CTV subset is usually the more relevant target.
Digital Out-of-Home (DOOH)
Outdoor digital screens (airports, taxis, billboards, office buildings, transit) bought programmatically. Often anonymised, but increasingly with location- and time-based audience signals layered in.
In practice. DOOH for B2B works in specific high-value contexts: executive commuting corridors, financial districts, conferences and event venues. The buying mechanics are still less mature than display, which is both an opportunity and a risk.
Programmatic audio
Audio advertising delivered programmatically: streaming music platforms, podcasts, broadcast radio simulcasts. Includes both pre-recorded host-read podcast ads (rarely truly programmatic) and dynamically inserted streaming audio (genuinely programmatic).
In practice. Audio for B2B is genuinely underused. Authenticated listener environments, long session times, and an attention regime that’s hard to find elsewhere. Suited to brand and senior executive reach.
Native
Ad formats designed to match the look and feel of the surrounding editorial content. Includes both feed-style native (in-feed content recommendation units) and brand-content placements styled as articles.
In practice. Native in B2B has a chequered reputation, partly because of the historical association with low-quality content recommendation networks. The format itself is fine. The supply quality is what matters. A native unit served inside a respected B2B trade publication is a different proposition to the same unit served inside a content-recommendation widget at the bottom of a tabloid article. Most modern B2B teams use native sparingly and only inside curated supply.
Contextual
Targeting based on the content of the page being viewed, rather than the identity of the user. Modern contextual is far more sophisticated than the keyword-matching of a decade ago, using natural-language understanding to classify pages by sentiment, topic, and entity.
In practice. Contextual is one of the few targeting methods that survives identity collapse intact. It’s also, despite years of vendor claims to the contrary, not a serious substitute for identity-based targeting at scale in B2B. The audiences are too broad, the targeting too coarse, the buying-committee resolution non-existent. Useful as a complement, not a replacement.
In-app
Advertising delivered inside mobile applications. Authenticated where the app requires sign-in, anonymous where it doesn’t.
In practice. In-app is a significant share of mobile programmatic supply but a relatively small share of B2B activation, because most B2B research and buying behaviour happens on desktop or in authenticated work environments. Worth including for completeness rather than as a B2B priority channel.
Reach / unique reach
The number of distinct users (or in B2B, distinct accounts) exposed to an advertising campaign at least once over a defined period. The complement to frequency in the reach/frequency planning model.
In practice. Most B2B campaigns under-report reach because they measure it at the cookie or device level rather than the account level. Account-level unique reach is the more useful metric for B2B planning, and the gap between the two can be substantial: a campaign that reached 50,000 unique devices might have reached only 12,000 unique accounts. The denominator matters.
8. Measurement and attribution
How you tell what worked. Where most B2B campaigns either over-claim or under-measure, often both at once.
Cost per Mille (CPM)
The cost of one thousand ad impressions. The default unit of programmatic media pricing.
Viewable CPM (vCPM)
The cost of one thousand viewable impressions, where viewability is defined by the IAB standard. Adjusts CPM for impressions that didn’t actually render in view.
In practice. Most B2B buying happens on standard CPM, but vCPM is the more honest unit for comparison across publishers. A 5 CPM at 50% viewability is more expensive in real terms than an 8 CPM at 90%.
Click-Through Rate (CTR)
The percentage of impressions that resulted in a click on the ad. Historically the headline performance metric, increasingly a vanity metric.
In practice. CTR in B2B display is structurally low (0.05 to 0.15 percent across most placements is normal) because the audiences aren’t there to click. Treating CTR as the primary indicator of programmatic success in B2B is a misalignment of the metric to the medium.
Cost per Click (CPC), Acquisition (CPA), Lead (CPL), View (CPV)
Pricing or reporting models indexed to the specific action. CPC for clicks, CPA for completed conversions, CPL for captured leads, CPV for completed video views.
In practice. B2B campaigns are almost always reported on a CPL or CPA basis internally, even when the underlying media is bought on CPM. The translation creates room for misunderstanding: a CPL of £200 against an average deal size of £80,000 looks expensive to a marketing operations team and looks like a rounding error to the CFO. Reporting on more than one cost basis is usually the right call.
Return on Ad Spend (ROAS)
Revenue attributed to advertising divided by ad spend. Most often used in B2C and retail. Less directly applicable in B2B where attribution windows are long and revenue attribution is multi-touch.
Multi-Touch Attribution (MTA)
A measurement approach that distributes credit for a conversion across multiple touchpoints in the buyer journey. Includes a range of weighting models: linear, time-decay, position-based, data-driven.
In practice. MTA in B2B is mostly broken. The cookie attrition that undermined B2C MTA undermines B2B MTA more severely because B2B journeys are longer. Most modern B2B measurement leans on a combination of incrementality testing and marketing mix modelling rather than MTA as the primary attribution engine.
Marketing Mix Modelling (MMM)
A top-down statistical approach to measuring the contribution of marketing channels to business outcomes. Uses aggregate spend, exposure, and outcome data, with no requirement for user-level tracking.
In practice. MMM has returned to prominence as identity has eroded user-level tracking. Modern MMM tooling (Meridian from Google, Robyn from Meta) is more accessible than the old enterprise-MMM consulting projects. For B2B, MMM is still under-deployed relative to its usefulness.
Incrementality
The measurement of whether an advertising exposure caused an outcome that wouldn’t have happened otherwise. Distinct from attribution, which assigns credit for outcomes that did happen.
In practice. Incrementality testing is the most honest form of advertising measurement available. It’s also under-used in B2B because the test design (split-cell, geo-holdout, conversion-lift study) requires either scale or patience that most teams don’t apply. The teams that do test rigorously are usually surprised by how much of their “attributed” volume turns out to be non-incremental.
Lift study
A specific kind of incrementality test, usually run inside a walled garden or DSP, that compares outcomes between an exposed group and a held-out control group.
REDS (Raw Event Data Stream)
The Trade Desk’s log-level data feed, providing per-impression, per-event data including bid, win price, supply path, audience match, and outcome. Available to advertisers and partners under specific agreements.
In practice. REDS data reveals two things every time you look at it for the first time. First, you bought more inventory from fewer supply paths than you thought, with the top ten paths typically accounting for 60-80% of impressions. Second, the paths that look statistically efficient on the surface (high win rate, low CPM) are often the same ones with elevated IVT and MFA exposure. Most B2B advertisers buying through The Trade Desk never request REDS access, which is one of the larger missed opportunities in modern B2B measurement. The access is contractual and the analysis is non-trivial, which is why most agencies don’t volunteer it.
Closed-loop measurement
A reporting structure that ties advertising exposure all the way through to a downstream business outcome (pipeline created, opportunity progressed, revenue closed). Requires connection between the ad-server data, the CRM, and ideally the financial system.
In practice. Genuine closed-loop measurement in B2B is rare because it requires data plumbing across functions that don’t normally share data. The teams that build it are usually able to make the case for their marketing budget on much firmer ground than the teams that don’t.
Pipeline attribution
The B2B-specific version of attribution that traces marketing activity through to qualified opportunities and revenue rather than stopping at leads or conversions.
View-through
A conversion or outcome credited to an ad that was viewed but not clicked. Distinct from a click-through, which requires a click.
In practice. View-through reporting in display is notoriously generous. Most reputable B2B measurement applies aggressive view-through windows (often single-digit days) and uses incrementality to validate that view-through claims are real rather than coincident.
Frequency cap
A limit on how many times a single user (or account, in B2B) is shown a particular ad over a defined period.
In practice. Frequency capping in B2B is harder than B2C because the underlying identity is more fragile and the unit (the account, comprising multiple identifiers across multiple people) is more complex. Most B2B campaigns are either under-capped at the user level (over-frequenting a small set of identified users while missing the rest of the buying committee) or over-capped at the account level (failing to build any presence with anyone). A defensible starting frame for B2B display sits around 8-12 impressions per user per week, with the campaign-level account-frequency target much higher than the per-user cap — see account-level frequency. The right answer is almost never the DSP default.
9. The terms reshaping the space in 2026
A short cluster of newer terms that didn’t exist or didn’t matter five years ago, and that are increasingly load-bearing now.
Agentic AI / agentic ad tech
Advertising technology where significant elements of campaign setup, optimisation, and decision-making are handled by AI agents acting semi-autonomously. Distinct from the broader use of ML in optimisation, which has been standard for a decade.
In practice. Agentic ad tech in 2026 is still mostly marketing language laid over capabilities that aren’t yet fully agentic. The trajectory is real, the current state is over-claimed. Worth tracking. Not yet worth restructuring a buying programme around.
Answer Engine Optimisation (AEO) / Generative Engine Optimisation (GEO)
The practice of structuring content to be discovered, cited, and surfaced by AI answer engines (ChatGPT, Perplexity, Claude, Google AI Overviews). The successor discipline to SEO in environments where the user gets an answer rather than a list of blue links.
In practice. GEO matters for B2B because the buying journey increasingly starts with a question typed into an AI, not a query typed into a search engine. The content patterns that get cited (structured, primary-source, definitive, recent) are different from the content patterns that ranked under classical SEO.
Authenticated audience
Audiences resolved through logged-in or otherwise authenticated identifiers, as distinct from inferred or probabilistic identifiers. Increasingly the only durable B2B audience category.
Audience containerisation
An emerging architectural pattern where audiences (defined by intent, firmographics, technographics, or behaviour) are packaged into deal IDs or other deliverable containers, so they can be activated across multiple buying channels with consistent definition and measurement.
In practice. Containerisation is the operational answer to the question “how do I activate the same audience across The Trade Desk, DV360, and StackAdapt without rebuilding it three times?” It also makes measurement comparison possible across DSPs in a way that traditional audience builds don’t.
Signal envelope
The cumulative pattern of signals attached to an account over time, treated as a structured object rather than as a feed of discrete events. The basis for the multiplicative engagement-scoring approach.
In practice. Signal envelopes are how modern B2B operators think about account state. Rather than reacting to each signal as it lands, you maintain a continuously updated picture of where each account sits and what would change that.
B2B Marketing Operating System
A term increasingly used to describe the integrated stack of audience, activation, measurement, and orchestration capabilities required to run modern B2B marketing. Distinct from individual point tools.
Benchmarks worth knowing
These are rough operating ranges from our own book and the wider public datasets, accurate to mid-2026. Treat them as the numbers you should be testing your own stack against, not as absolutes. If yours are materially worse, there’s probably a recoverable problem. If yours are materially better, double-check the methodology before you celebrate.
Working media
Open-exchange display through a typical agency-managed stack: 55-68%
Curated PMP via a competent operator: 68-82%
Direct programmatic with named publishers: 78-90%
LinkedIn (walled, all-in CPM): not directly comparable; treat as a separate line
Typical B2B CPMs
Open exchange display: £4-£10
Curated B2B PMP display: £6-£18
Premium B2B publisher PMP (FT, WSJ, Bloomberg): £18-£40+
Programmatic audio (streaming, B2B-relevant): £8-£18+
CTV (B2B premium environments): £35-£55+
LinkedIn (CPM-equivalent): £35-£90 depending on targeting depth
Match rates (against a well-maintained CRM list)
Open-web display: 22-38%
CTV in authenticated environments: 35-55%
LinkedIn: 60-75%
Mobile in-app: below 15%
Viewability
Acceptable display target: 70%+
Acceptable video target: 90%+
Below 50% on either is a problem to diagnose, not a number to optimise around
Supply chain hops between DSP and publisher
Best case (direct deal, named publisher): 1-2
Typical open exchange: 3-5
Worst case (reseller chains, MFA-adjacent paths): 6+
Frequency
Per-user-per-week display (B2B): 8-12 impressions
Per-account-per-month total (active campaign): 40-80 impressions across the buying committee
Anything north of 20 impressions per user per week is starting to burn the brand it’s trying to build
Retargeting incrementality
Anything below 30% measured incremental lift on a holdout test: mostly attribution theatre
30-50%: genuine but optimisable
Above 50%: rare in B2B, worth understanding why before scaling
Intent signal decay
Research-based intent half-life: 14-21 days
Account-level surge half-life (without corroborating signal): 5-7 days
Acting on a signal older than 30 days: usually acting on a different account state
These are the numbers we keep on a wall. If yours are dramatically different, the gap is the brief.
How this glossary will keep evolving
The vocabulary of B2B programmatic moves faster than the vocabulary of almost any other commercial discipline. Terms shift meaning every twelve to eighteen months. Some entries above will need rewriting by the end of the year. New entries will be added.
If you spotted something that’s wrong, missing, or outdated, reply to this email. Every correction gets read, considered, and (where warranted) folded in. The version above is 1.1; the next update will land in the same place. Bookmark the URL, not a screenshot.
Found this useful? Forward it to the colleague who keeps asking what a deal ID is, the one who insists CTR is a B2B performance metric, or the CFO who wants to understand where the programmatic budget actually goes. They’ll thank you. Possibly.
How to cite this. Harty, M. (2026). The B2B Programmatic Glossary. The B2B Stack. theb2bstack.com
Written by Mike Harty, fifteen-year veteran of B2B programmatic and founder of FunnelFuel.io. The B2B Stack is a practitioner-led publication for the people running modern B2B marketing stacks. FunnelFuel is the managed service that runs the stack described in this glossary, day in, day out, for B2B advertisers worldwide.
The regular Friday briefing returns this week.