The Ultimate Account-Based (ABM) Measurement Glossary - Every Term You Need To Win the Pipeline Attribution Fight

There’s a meeting that repeats in almost every B2B organisation I’ve ever worked with. Marketing presents a pipeline influence number. Sales presents a completely different one. The CFO trusts neither, the CMO leaves with a smaller mandate than they walked in with, and the whole thing reconvenes next quarter to have the identical argument with fresher data.

Whatever the CRO’s views on marketing performance, and I’m sure we’ve all seen some strong views, the disagreement begins with definitions and the success criteria that fall out of them. Definitions that were never jointly agreed before the dashboards got built. Both teams pulled from the same CRM. Both numbers are usually “correct.” They’re answers to different questions wearing the same name. And we still talk about sales and marketing alignment.

Big statement: account-based measurement is the single most confused reporting topic in B2B.

Nearly every definition in circulation was written by a vendor with a platform to sell, so nearly every definition flatters whatever that platform happens to measure well. Amazing at capturing web analytics data? Guess where the metrics will focus. Awesome at capturing advertising insights? You get the picture.

Nobody has written the whole thing out neutrally, from the perspective of the people who actually have to defend these numbers in front of a board.

So here it is. Structured as an actual reference: every entry follows the same format (what it means, where it gets confused, the operator’s take), the head-to-head terms are laid out in strict parallel, and there’s a copy-paste definitions block at the end that you’re actively encouraged to lift straight into your own reporting deck. Use it, don’t credit it, I genuinely don’t mind. The point is that the argument stops.

What you’ll learn in this article:

• The foundational units of account-based measurement (ICP, TAL, buying groups, ABM tiers, the dark funnel) and why getting the unit wrong corrupts every metric downstream

• How engagement and qualification actually work: engagement scores, intent data, vendor-branded states like the 6QA, and the full MQL/MQA/SQL/SAL/SQA/PQA stack, term by term

• The progression metrics almost nobody reports properly: coverage, buying group penetration, and velocity

• Pipeline sourced versus pipeline influenced versus multi-touch attribution, and why conflating them manufactures the quarterly marketing-versus-sales fight

• The correction layer: self-reported attribution, incrementality testing and MMM, and when each earns its place

• A complete reporting framework by audience and cadence, plus a copy-paste definitions block for your next board deck

How to use this

Read it once top to bottom, because the sections build on each other. After that, treat it as a lookup. Every term is defined in a self-contained entry, so you can land on any single definition cold and it’ll still make sense. That’s deliberate. It’s the difference between a glossary and an essay with headings.

If you’re reading this in a browser, bookmark it now. That’s what it’s for.

One prerequisite before any of it makes sense: get clicks out of your measurement stack. Click-optimised systems learn to find clickers rather than buyers, a big share of programmatic clicks carry no intent at all, most legitimate B2B inventory gets filtered out before a click is even possible, and entire channels your buying committee actually lives in (audio and CTV among them) don’t produce a click in the first place. I wrote the full argument here: The Observer’s Paradox. The one-line version for this glossary: if a metric still has “click” anywhere in its lineage, treat it as noise. And ask your platform vendor what percentage of their engagement score is click-derived. The answer, and how long it takes them to produce it, will tell you a lot.

---

Part 1: The foundations

Account-based measurement fails most often not at the metric level but at the unit level, because teams keep measuring accounts with instruments designed for individuals. Five terms establish the unit properly.

ICP (Ideal Customer Profile)

What it means: The firmographic and behavioural definition of the companies you should be selling to. Size, industry, region, technology stack, business model, and increasingly signal-based criteria like observed in-market behaviour.

Where it gets confused: Teams treat the ICP as a list. It isn’t. It’s a filter, a written definition against which any company can be tested. The list it produces is the TAL, below.

The operator’s take: The most common ICP failure is building it from aspiration (”who we’d love to sell to”) rather than evidence (”who actually closes, retains and expands”). Rebuild it annually from your own closed-won and churn data, not from the pitch deck. The other common mistake is having one big ICP when the business actually serves three or four distinct buyer types. Break those out into their own ICP clusters and everything downstream sharpens up.

TAL (Target Account List)

What it means: The named list of specific companies that pass the ICP filter and that you’ve deliberately chosen to pursue in a given period.

Where it gets confused: The TAL is the denominator for almost every metric in this glossary, which is why a badly built one corrupts everything downstream very quickly. If the TAL contains accounts that will never buy, your qualification rates, coverage figures and engagement trends are all being measured against a work of fiction.

The operator’s take: Most measurement problems that get blamed on attribution are actually TAL hygiene problems in disguise. Before commissioning an attribution project, audit the TAL. It’s cheaper, and it’s usually where the bodies are.

Even in the second half of 2026 it astonishes me how poor many TALs are. No standardisation, no refresh cadence, stale entries from two planning cycles ago, and millions being spent against them anyway.

The single biggest win available to most B2B vendors is investing proper work in their TALs. Plural is not an accident either: they should be segmented, and they should be fresh. At FunnelFuel we built a system with over 700 live connectors for exactly this reason, to join and combine sources, interrogate them, match them to ICP, and make sure the matching engine reflects the intent behind the TAL, meaning who you actually want to reach, rather than producing an inflated match rate that looks good in a QBR and targets nobody in particular.

The TLDR: nail your TAL, everything else swims off it. And if you want help thinking yours through, my inbox is open.

ABM tiers (1:1, 1:few, 1:many)

What it means: The standard structure for segmenting an account-based programme by investment level. Tier 1 (1:1) is a handful of accounts getting fully bespoke treatment. Tier 2 (1:few) is clusters of very similar accounts getting lightly personalised programmes. Tier 3 (1:many) is programmatic ABM, the full TAL receiving targeted but scaled activity.

Where it gets confused: Teams define the tiers for activation and then forget them at measurement time, reporting a single blended number across all three. But the tiers have completely different economics and should carry different expectations. A Tier 1 account might justify a year of patient buying-group development. Tier 3 lives or dies on efficient qualification rates.

The operator’s take: Never report a blended ABM number. Blending Tier 1 and Tier 3 produces a figure that describes neither. We regularly see lists where accounts genuinely worthy of Tier 1 attention get bundled into one big undifferentiated pile, because big is always better, right?

I always advocate for TAL splitting by at least one or two firmographic dimensions alongside the tiers. Breaking banks out from healthcare brands, for instance, opens up an easy win: you can run more scaled but still nuanced 1:many that starts to borrow from the 1:few playbook, cast for dramatically more signal, and feed the 1:1 accounts with better data as a result.

My bet is that technology erodes true 1:many over the next couple of years and lets world-class paid media run against 1:few audiences at 1:many economics, which is genuinely exciting. But only if the up-front investment in the TAL gets made. It always comes back to the TAL.

Buying group (buying committee)

What it means: The set of people inside an account who collectively make the purchase decision. In enterprise B2B this routinely runs to double digits across multiple functions, with different members researching different questions at different times, most of it invisibly. In 2026 Forrester estimates that buying groups are 13 people across 4 departments, sometimes with upwards of 9 external stakeholders brought in to aid the decision

Where it gets confused: The buying group is the reason lead-level measurement breaks in account-based motions. One enthusiastic individual tells you almost nothing about whether an organisation is in-market or where the entirety of a consensus driven committee’s collective heads are at. Moderate engagement from four different functions in the same account tells you a great deal. Every serious account-based metric aggregates across the buying group; every legacy metric doesn’t.

The operator’s take: Take the highlighted point above, and this in a nutshell is my argument for where programmatic falls over itself for B2B. Programmatic and many ad buying platforms are designed to hit individual users but they entirely lack the ability to map them up into a buying committee, and then up into the business they work for. The buying committee construct is critical and is the reality of B2B buying journeys.

If mapping your ICPs correctly to output the best possible set of TAL’s is critical, and it really is, then the next big job is to really understand who is making buying decisions.

Now different organisations will approach this differently. Some are smaller and leaner in their decision making, which doesn’t have to mean in their total org size. Some small businesses are slow and ponderous, and may have huge committees. The avatar of decision maker and influencers can at best be approximated and I love a signals based approach.

That means, which deviceIDs are showing intent, and they get mapped, where possible into people using an account graph. This then lets us understand the types of roles that are coming on your website, mapped to accounts and what we know from a firmographic POV about those businesses. This, with machine learning approaches, allows an estimation of buying committee avatars. CRM compounds the workflow. Then we take a broader approach to communicating with these avatars, which means casting a bigger net, because the reality remains that we do not know if any one organisation has a faster and looser, or wider and slower approach to buying - so we should err on the side of caution and use advertising to try and find out

To finish this off, I internalise that the account is the unit and the buying group is the texture within it, and the rest of this glossary follows logically. Every vendor dashboard you use will keep trying to pull you back down to the individual, because individuals are what cookies and form fills can see. Resist it.

Dark funnel

What it means: The portion of the B2B buying journey that happens on surfaces your measurement stack cannot instrument: peer recommendations, Slack and WhatsApp groups/threads, podcast mentions, community discussions, internal meetings, and increasingly the LLM a buyer consults before ever touching your website. None of it fires a pixel. Research across the intent vendors consistently puts the majority of the buying journey in this category, and buyers routinely arrive with a shortlist already substantially formed.

Where it gets confused: Teams treat the dark funnel as a measurement problem to be solved with more software or even just more effort. It cannot be. No platform can track a Slack recommendation. The dark funnel is a structural condition to be corrected for, which is what Part 4’s correction layer is for.

The operator’s take: The dark funnel is not an excuse for unmeasurable marketing but equally we have to accept real structural limitations, and not waste time trying to work around them where there is no viable way - that is not defeatist or small minded, it is pragmatic and realistic. It is the reason the correction layer (self-reported attribution, incrementality) has moved from nice-to-have to mandatory. If your measurement stack has no mechanism that can see the invisible journey, your dashboards are precise descriptions of the visible minority.

---

Part 2: Engagement and qualification

Account engagement score

What it means: A composite number, usually 0 to 100, representing how “engaged” a target account is with your brand, aggregated across every buying group member and every measurable channel, over a rolling window. Common inputs: website visits and content consumption, email engagement, ad exposure, event attendance, and third-party intent signals.

Where it gets confused: There is no industry-standard formula, and most platforms will not fully disclose theirs. The weighting logic varies wildly on five axes:

• Recency decay - does a whitepaper download from six weeks ago still count, and at what discount to yesterday’s?

• Committee weighting - does an engaged VP score the same as an engaged intern? Should they?

• Channel weighting - does a demo request live on the same scale as a viewable ad impression?

• Anonymous resolution - how much of the score rests on de-anonymised account-level activity versus known contacts, and how confident is the match?

• Normalisation - is the score comparable between a 200-person account and a 200,000-person account, or only meaningful against that account’s own trend line?

This is why the same account can legitimately score 85 in one platform and 40 in another with no data quality problem anywhere. Both are “correct.” They are different instruments measuring different things under the same label.

The operator’s take: Stop searching for the platform with the “right” formula. There isn’t one. Pick a methodology, document its logic in plain English so anyone can defend it in a leadership meeting, and use the score as a trend indicator for a given account over time. An account moving from 40 to 70 over six weeks is signal. The number 70 itself, in isolation, is decoration.

My approach with scoring has been to make it composable, which means the vendor teams can input and shape the scoring that we run for them. Use my tech to capture data you cannot capture today, use my tech to pipeline data between systems like web analytics and media buying which is very difficult to do, but bring your own scoring model to make your own reality out of this data pipelining.

Intent data and surge scores

What it means: Third-party signals indicating that an account is actively researching a topic, typically derived from content consumption patterns across networks of B2B publishers and research sites. A surge score flags that an account’s consumption of a given topic has spiked meaningfully above its own historical baseline.

Where it gets confused: Two big ways.

First, topic-level intent is not vendor-level intent: an account surging on “account-based marketing” is researching the category, not necessarily shortlisting you.

Second, intent providers differ fundamentally in how signal is collected (co-op publisher networks versus bidstream inference/modelling versus review-site activity), and those methodologies have very different noise profiles. Treating all “intent” as one substance is how bad TAL prioritisation happens.

The operator’s take: Intent data is an input to engagement scoring and TAL prioritisation, not a metric to report upward in its own right. A surge is a reason to act, never a result to claim. The moment “accounts surging” appears in a board deck as an outcome, the programme has lost the plot.

We need to be realistic about intent data in 2026 too. This is not having a go at any players in the ecosystem, and I partner with them all, and many are friends. However, the elephant in the room.

The same dark funnel behaviours which are impacting the tracking of your funnel is impacting the collection of intent data. Whether it is bid stream modelling (which we do a form of at Funnelfuel using audience containers to understand usage and intent), or publisher co-ops, these are two different ways to get to the same end point - understanding what people are reading on the content web. great - but - increasingly they use LLMs instead which sit squarely in the dark funnel. Buying committees are beginning their research journey 51% of the time in the dark funnel, and 94% - yes 94%! - of buying journeys now utilise LLMs. This is obvious in 2026.

However those 94% of journeys that use AI, or the 51% that start in an LLM used to start on observable content websites online, like G2, like IDG, like deeply specialist research sites which were capturable in 2020 via co-ops and modelling, and which are now feeding the funnel via citations which we never see. This data is thus thinner now then it was before LLMs. That is a fact

There is still value here but understand its limitations and it is another signal to use alongside others. That is why we say this is the signals era.

Vendor-qualified account states (6QA and equivalents)

What it means: Proprietary qualification thresholds productised by ABM platforms. The best known is 6sense’s 6QA: an account the platform’s model judges to have crossed into a late buying stage based on its combination of fit, intent and engagement, at which point it is flagged as ready for sales. Demandbase, Terminus and others run equivalent constructs under their own branding.

Where it gets confused: A vendor-qualified state is an engagement score with a threshold and a trademark. The model is proprietary, the weighting is opaque, and the qualification logic is configurable, which means a “6QA” at one company is not the same object as a “6QA” at another. Teams routinely compare these states across organisations, or present them to boards, as though they were standardised. They are not.

The operator’s take: These are useful operational triggers and I am not dismissing them at all; a well-configured one beats a home-rolled score for many if not most teams. But never let a vendor-branded state become a board-level metric without a plain-English definition of what it means in your configuration. “We had 40 6QAs this quarter” is not a sentence a CFO can act on.

What I am seeing is big enterprise clients, often using AI, understandably now believe that they can create their versions, without having to crowbar their own data into a pre-canned model. That is why more and more big players want lots and lots of raw data exhaust fumes, so they can run their own computations. A year ago we got a lot of asks to work with these platforms, today we definitely still do but the second cluster of DIY signal builders is growing by the week. This backs up the SaaS-apocalypse narrative.

The qualification stack: MQL, MQA, SQL, SAL, SQA, PQA

Six terms, because this is the layer that causes the most cross-functional friction, and it is almost always friction over definitions. For each one, hold three questions in mind: what is the unit (person or account), who does the qualifying, and what is actually being claimed.

MQL (Marketing Qualified Lead). Unit: a person. Qualified by: a marketing scoring threshold. The claim: this individual has shown enough interest to be worked. The trap: one junior researcher can generate an MQL while the five people who will actually sign the contract have never heard of you. That is not a data error; it is the structural limit of measuring individuals in a buying-group world.

The dream ticket would be migrating this from one person to a marketing qualified buying committee, especially if the committee had been mapped by marketing.

MQA (Marketing Qualified Account). Unit: an account. Qualified by: aggregate engagement across the buying group within a defined period. The claim: this organisation is showing coordinated signs of being in-market. The trap: check the aggregation logic, because one hyperactive contact should not be able to trip the threshold alone. If your MQA definition can be satisfied by a single person, you have renamed your MQLs, not replaced them.

Most scoring around this fails to take into account the buying committee in its broader context, so it really just becomes another term for account engagement score.

SQL (Sales Qualified Lead). Unit: a person. Qualified by: sales, after an actual conversation. The claim: this individual represents a real, workable opportunity. In account-based motions the SQL is a legacy unit, increasingly absorbed into the SQA, but it survives in most CRMs and most compensation plans, so it is worth defining even if only to retire it deliberately.

SAL (Sales Accepted Lead, or Account). Unit: either, depending on your motion. Qualified by: sales, on receipt of the handoff. The claim is procedural rather than evaluative: sales formally acknowledges the qualified record and commits to follow up within an agreed SLA. This is the forgotten middle step, and its absence is why the two oldest complaints in B2B (”sales never followed up” and “marketing sends us rubbish”) are both permanently unfalsifiable in most organisations. Nobody signed for the parcel, so nobody can prove anything about what happened to it.

SQA (Sales Qualified Account). Unit: an account. Qualified by: sales, after investigation. The claim: there is evidence of a real buying process, meaning an identified need, a champion, budget movement or a defined evaluation underway. The critical word is accepted. An SQA is a joint claim, agreed by both functions, not a marketing claim awaiting validation.

PQA (Product Qualified Account). Unit: an account. Qualified by: product usage signals, meaning seat adoption, feature depth, trial behaviour. The claim: usage suggests conversion readiness. Only meaningful if you run a self-serve or trial motion alongside enterprise sales. Managed-service and pure enterprise teams can ignore it; it appears in vendor glossaries mainly as a completeness gesture.

Where it all actually breaks: Almost always at the MQA-to-SQA handoff, for the same reason the MQL-to-SQL handoff was always contentious: the two functions are rarely working from a jointly written, jointly signed threshold. Here is the test. Ask your sales leadership, without a deck in front of them, what specifically distinguishes a sales qualified account from a marketing qualified one in your organisation. If the answer takes more than two sentences, you do not have a qualification stack. You have a naming convention.

The operator’s take: The fix costs nothing: a one-page written definition of each state, agreed by both functions, before the dashboard is built rather than after the argument starts. And add the SAL step even if it feels bureaucratic, because it converts the eternal “sales never followed up” versus “marketing sends rubbish” stalemate into a measurable SLA with a name attached.

Many of these terms are as dated as the funnel concept, and like many scores, they are only as good as what can be tracked. The scores limiting factor is data.

---

Part 3: Progression metrics

Between qualification and pipeline sits a set of mid-funnel metrics that are genuinely useful and almost universally under-reported, mostly because they do not flatter anyone and as a result nobody is keen to own them

Account coverage

What it means: The percentage of your TAL in which you have the assets required to actually run the motion: identified buying group contacts, identifiers to reach them with ads, working contact data, and active engagement channels.

Where it gets confused: Teams report programme metrics against the full TAL when the programme is only operationally live in a fraction of it. A TAL of 500 accounts with real coverage of 120 is a 120-account programme wearing a 500-account budget, and where budget is getting pushed to potentially saturated accounts in the 120 list

The operator’s take: Report coverage before you report anything else, because every downstream percentage is computed against it. It is also the most honest early indicator of whether an ABM programme is real or aspirational.

Buying group penetration

What it means: Within a given account, the proportion of relevant buying group roles you have actually identified, reached and engaged.

Where it gets confused: This is the metric that catches the classic false positive: an account that looks hot on aggregate engagement where every signal traces back to one person. Engagement scores can be inflated by depth-on-one-contact in exactly the way penetration exposes.

The operator’s take: Penetration is the honest counterweight to the engagement score, and the two should always travel together. High engagement plus low penetration is not a hot account; it is a research project with one enthusiastic author.

Deal velocity (and stage velocity)

What it means: How quickly accounts move between defined stages: TAL to MQA, MQA to SQA, SQA to opportunity, opportunity to close. Usually expressed as median days per stage.

Where it gets confused: Velocity means nothing without a baseline. “Our MQA-to-opportunity velocity is 47 days” is not information; “it was 68 days before the programme started” is.

The operator’s take: Velocity is the most CFO-legible metric in the entire account-based canon, because it converts marketing activity into a language finance already respects: time and rate. If nobody in your organisation captured a pre-programme baseline, that is this quarter’s first job, because every future efficiency claim depends on it. So in other words, for the CFO - if I give you $1m today, when can I expect to get returns on this?

---

Part 4: Pipeline, attribution and the correction layer

Now the main event: the source of the meeting from the opening of this piece.

Pipeline sourced vs pipeline influenced

The single most important distinction in this glossary, so it gets the table treatment:

Pipeline sourcedPipeline influencedDefinitionValue of opportunities where marketing generated the first touch that led to the opportunity’s creationValue of opportunities where marketing touched any buying group member at any point in the cycle, regardless of who originated the dealAttribution mechanicFirst-touchMulti-touch, typically over a lookback window running backward from opportunity creation and forward through to closeThe question it answersWhich activities create net-new pipeline that would not otherwise exist?Where is marketing accelerating and supporting deals across the whole motion?CharacterClean, conservative, deliberately narrowBroad, generous, deliberately inclusiveNatural audienceCFO, board (budget allocation)CRO, programme leads (programme effectiveness)Failure modeUnder-credits everything that isn’t first-touch, punishing brand and mid-funnel work”Influenced” can mean a single ad exposure to a single contact, so an unlabelled 70% influenced claim destroys the whole report’s credibility

These are not two versions of the same number. They have different definitions, different mechanics and different denominators, and most B2B attribution conflict is simply a marketing team presenting an influenced figure into a room that is mentally expecting a sourced one.

The operator’s take: Report both, always together, always explicitly labelled, never blended into a single “marketing pipeline” figure. Presenting one without the other is what manufactures the argument. And put a one-line definitions footer on every deck (there is one at the end of this piece you can lift verbatim).

Multi-touch attribution (MTA)

What it means: The rule set for distributing credit across touchpoints when more than one contributed to an outcome. It is the mechanism underneath the influenced number, not a separate metric. The standard models:

ModelHow it distributes creditWhat it’s forFirst-touch100% to the first interactionDemand creation questions (this is what “sourced” runs on)Last-touch100% to the final interaction before conversionDemand capture questions; systematically over-credits bottom-funnelLinearEqually across all touchesA neutral default when you refuse to make weighting decisionsTime-decayMore credit to touches closer to conversionLong cycles where recency plausibly mattersU-shapedHeavy credit to first touch and conversion touchBalancing creation and captureW-shapedHeavy credit to first touch, lead creation and opportunity creationMotions with a meaningful mid-funnel milestoneAlgorithmicModel-derived weights from your own dataSounds most sophisticated; hardest to explain in a boardroom

Where it gets confused: Two things every operator should internalise. First, the model choice changes the influenced number materially, which is why two teams both “doing multi-touch” can produce different figures and both be technically right. Second, and more fundamentally: MTA can only distribute credit across touches it can see, and in B2B most of the journey is dark funnel. MTA is a precise accounting of the visible minority.

The operator’s take: Pick one model, write down why, and stop shopping models until you find the one that flatters this quarter. The credibility cost of changing attribution models mid-year exceeds any insight the new model provides.

View-through measurement

What it means: Crediting outcomes to ad exposure without a click: an account was served impressions, and subsequently visited, engaged or converted. In a post-click world, and especially in channels like CTV and audio where clicks do not exist, exposure-based measurement is structurally necessary.

Where it gets confused: View-through is necessary and gameable at the same time. Loose view-through windows and generous matching logic let platforms claim credit for outcomes they merely coincided with. The question to ask any platform reporting view-through conversions: what is the window, what is the match methodology, and what does the same report look like with the window halved?

The operator’s take: Accept view-through as a directional input to account-level measurement, interrogate its settings, and never let a platform grade its own homework as the final number. Which is the perfect segue to incrementality.

Self-reported attribution (SRA)

What it means: An open-text “how did you hear about us?” field on every conversion form, and the same question asked verbally on every first sales call, with the answers logged and reviewed as first-class data.

Where it gets confused: Teams dismiss it as anecdotal precisely because it is analogue. Yet teams running software and self-reported attribution side by side routinely find radically different channel mixes: what software logs as “direct” or “organic search” was, by the buyer’s own account, a podcast, a peer recommendation or a community thread that triggered the branded search. Neither source is complete. Software tells you which touchpoints occurred; self-reported data tells you which ones mattered.

The operator’s take: Two implementation details determine whether SRA works or becomes theatre. Keep the field open text, because dropdown options collapse the interesting answers into “social media.” And look for buying-group patterns: three people from the same account citing the same source is a channel finding, not an anecdote.

Incrementality (lift testing)

What it means: The experimental answer to “did this activity cause pipeline, or just correlate with it?” Hold out a matched segment of the TAL from an activity, run it for the rest, and measure the difference in outcomes between exposed and held-out groups.

Where it gets confused: It is the only method in this glossary that measures causation rather than correlation, which is exactly why finance teams trust it and why it is quietly displacing attribution modelling as the top-of-house proof mechanism in sophisticated organisations. The honest caveats: it needs enough account volume to build meaningful test and control cells, it needs the discipline to genuinely withhold activity from the control group, and it answers big questions slowly rather than small questions instantly.

The operator’s take: Use incrementality for the material budget decisions, not the weekly optimisation. One well-designed holdout test per quarter, on the channel carrying the most budget or the most scepticism, is worth more board credibility than a year of attribution dashboards. A lift result is the only slide in the deck the CFO cannot argue with.

MMM (Marketing Mix Modelling)

What it means: Statistical modelling of the relationship between marketing investment by channel and business outcomes over time, using aggregate data rather than user-level tracking. The old brand-world technique, back in fashion because it needs no cookies and sees channels that attribution cannot.

Where it gets confused: MMM is regularly pitched to B2B teams whose data cannot support it. It wants years of weekly spend and outcome data with real variation in it, and meaningful ongoing investment to maintain. Below a substantial annual media budget, the model’s error bars are wider than the decisions it is meant to inform.

The operator’s take: For most B2B teams, MMM is a later-stage tool. Get SRA and periodic incrementality testing running first; they answer most of the same questions at a fraction of the cost. When media spend gets large enough that channel-level allocation errors cost real money, MMM earns its seat.

The layered stack, stated plainly

The mature measurement position is not one method but four layers, each doing what it is actually good at:

1. Platform measurement (engagement scores, qualification states) for operational speed

2. Sourced and influenced pipeline (separately labelled) for programme reporting

3. Self-reported attribution as the standing correction for what software cannot see

4. Periodic incrementality tests as the causal anchor for the numbers that decide budgets

Any one layer alone is either fast but shallow, or rigorous but slow. The stack is the answer.

---

Part 5: The reporting framework

Strip everything above into a cadence and it looks like this:

CadenceAudienceReportDo not reportWeeklyABM / demand teamEngagement score movement across the TAL, MQA and SQA counts, coverage and penetration by tierAnything as a revenue claimMonthlyCRO and sales leadershipSourced and influenced pipeline (separately labelled), MQA-to-SQA acceptance rate, stage velocity vs baselineBlended “marketing pipeline” figuresQuarterlyCMO and CFOSourced pipeline as the primary number, influenced as programme context, velocity trends as the efficiency story, incrementality results as the headline when you have themEngagement scores as anything other than a leading indicator

And one rule that governs all three rows: never present a metric to an audience that has not been given its definition. Most account-based measurement does not fail in the data. It fails in the room.

The copy-paste definitions block

This is the artefact the whole piece has been building to. Lift it verbatim into the footer of your reporting deck, adjust the lookback window to match your model, and the quarterly definitions argument ends:

Definitions used in this report. Pipeline sourced: opportunity value where marketing generated the first touch leading to opportunity creation (first-touch attribution). Pipeline influenced: opportunity value where marketing engaged any buying group member during the cycle (multi-touch attribution, [90]-day lookback, through close). These are different metrics answering different questions and are never blended. MQA: an account meeting the aggregate buying-group engagement threshold jointly agreed by marketing and sales on [date]. SQA: an MQA reviewed and accepted by sales as showing evidence of an active buying process. Engagement scores are leading indicators of account attention, not revenue claims.

Six sentences. It costs nothing, and it is the difference between a report that gets trusted and one that gets re-litigated every quarter.

---

The close

None of this is complicated once it is written down in one place. That is the entire problem with account-based measurement: every term above is well-defined by someone, somewhere, but almost never by the same someone twice, and nearly always by someone whose platform benefits from the definition. So every in-house team ends up rebuilding the Rosetta Stone from scratch, under deadline pressure, in a deck that half the room quietly distrusts.

Bookmark this one instead.

What is your organisation’s actual, written definition of “influenced,” and would your sales leadership agree with it without checking?

If this was useful, it was built to be kept, not just read. Subscribe below and The B2B Stack lands in your inbox on Fridays. And if the definitions block saves you one quarterly argument, forward it to whoever you usually have that argument with.

---

Vendor-qualified accounts / 6QA / engagement scoring

• 6sense Glossary (buying stages, 6QA definition): https://6sense.com/glossary/ and https://6sense.com/glossary/6sense-qualified-account/

• 6sense support docs — default 6QA qualification logic (buying stage, profile fit, configurability): https://support.6sense.com/docs/6sense-qualified-accounts-6qas

• 6sense on 6QA misattribution (useful supporting read for your attribution section): https://6sense.com/blog/the-6qa-mistake-how-teams-misinterpret-a-key-buying-signal/

MQA and the qualification stack

• SaaSTrack, “How Do You Define an MQA?” (MQA/SQA distinction): https://www.saastrack.ai/blog/how-do-you-define-an-mqa

• Archstone Digital MQA glossary (fit/intent/engagement model): https://archstonedigital.com/glossary/marketing-qualified-account-mqa/

• INFUSE MQA glossary: https://infuse.com/glossary/marketing-qualified-account-mqa/

• Clay MQA glossary (buying-committee rationale): https://www.clay.com/glossary/marketing-qualified-account

Sourced vs influenced pipeline

• Metadata.io, “Sourced vs Influenced Pipeline”: https://metadata.io/resources/blog/sourced-vs-influenced-pipeline/

• ZoomInfo Pipeline, marketing-sourced pipeline benchmarks and framework: https://pipeline.zoominfo.com/marketing/marketing-sourced-pipeline-trending-down

• Rework, “Marketing-Sourced vs Influenced Pipeline: The Distinction”: https://resources.rework.com/libraries/marketing-sales-alignment/marketing-sourced-vs-influenced-pipeline

• Saber glossary entries: https://www.saber.app/glossary/marketing-sourced-pipeline and https://www.saber.app/glossary/marketing-influenced-pipeline

• Prospeo, marketing-sourced pipeline (includes the Forrester 70%/48% tracking-gap stat): https://prospeo.io/s/marketing-sourced-pipeline

Self-reported attribution / dark funnel

• Refine Labs, Hybrid Attribution Framework study (the 90% measurement gap, podcast 53% vs 0% finding): https://www.refinelabs.com/article/hybrid-attribution-framework

• Refine Labs, “The Attribution Mirage” (methodology detail: mandatory open-text field, dark social close-rate analysis): https://www.refinelabs.com/article/attribution-mirage

• Prospeo, self-reported attribution guide (2026 summary of the same study): https://prospeo.io/s/self-reported-attribution

Incrementality / lift testing

• Measured, holdout test explainer (includes the catalog 14% vs 40% claimed example): https://www.measured.com/faq/holdout-test/

• AdSights incrementality glossary (test design best practice, minimum detectable effect): https://www.adsights.ai/resources/glossary/general/incrementality-testing

• Amplitude, incrementality testing overview: https://amplitude.com/explore/experiment/incrementality-testing

• Academic grounding if you want it: Lewis & Rao lineage via arXiv, “Latent Stratification for Incrementality Experiments” (calls holdout experiments the gold standard for causal inference): https://arxiv.org/pdf/1911.08438.