The week your data layer got a landlord + LinkedIn launches + Longtail publishers get hit by Google again
What happened in B2B paid media this week
Back to the usual Friday format today, where we look at what’s happened across the B2B tech stack this week, with a special focus on data driven ad channels. It was a big week, the week impartiality left the building when it comes to first party data on ramps
Unlike recent weeks we were starved of our usual agency-versus-DSP punch-up this week. No earnings shocker, exactly. The story that matters happened on a quieter register, and most of the trade coverage filed it under M&A maths and agentic buzzwords. That framing misses the point.
For about a decade, the identity and onboarding layer underneath digital advertising has behaved like neutral infrastructure.
LiveRamp sat in the middle of the pipes. Everyone routed through it. The holdcos used it, the independents used it, in-house teams used it, and nobody had to ask whose side it was on, because the answer was nobody’s. That was the whole value of it, because any tech tasked with handling first party data is always going to carry special sensitivity…
This week one holdco bought it. Publicis is acquiring LiveRamp for 2.2 billion dollars, and a neutral piece of plumbing just picked a side. If you buy B2B media and you are not a Publicis client, your identity resolution layer now has a landlord who also competes for your budget.
Omnicom worked that out within days and started heading for the exit, which tells you the conflict is not hypothetical.
Underneath the marquee deal, the same theme kept surfacing all week. The foundations B2B paid media stands on are quietly being bought, stretched, and questioned.
LinkedIn admitted the launch-day model is dead. The open web’s discovery layer kept collapsing under AI search. And the ratings currency that underpins CTV buying turned out to be measuring pirated streams.
Four stories that change the ground under your media plan.
We’ve got an obvious starter though
1. Publicis buys LiveRamp, and neutrality leaves the building
The story. Publicis announced on 17 May it is acquiring LiveRamp in an all-cash deal worth 2.2 billion dollars in enterprise value, around 2.5 billion in equity, at 38.50 dollars a share. That works out at roughly 12.3 times forward adjusted EBITDA on a business doing about 126 million dollars of EBITDA this year.
LiveRamp connects more than 25,000 publisher domains and 500-plus data and technology partners across 14 markets, with RampID as its durable identifier.
Publicis dressed the rationale in agentic language, calling it “data co-creation for smarter agents.” Then, on cue, Digiday reported Omnicom accelerating its exit from LiveRamp, citing the obvious problem of leaning on critical infrastructure owned by a direct rival.
My take. Strip out the agentic wrapper and this is a control play, not an AI play. The interesting bit is not what Publicis gains. It is what everyone else loses. And that’s around $600m of non Publicis revenue leaving the building and without a truly obvious home to head to
LiveRamp’s entire commercial logic was that it sat above the fight. It was the Switzerland of identity. RampID was useful precisely because it was vendor-agnostic plumbing that resolved a person or an account across walled gardens and the open web without an agenda. The moment a holdco owns that plumbing, the agnostic part is gone, whatever the press release promises about continued independence.
Omnicom did not wait around to test those promises, and neither should anyone running serious B2B budgets through clean rooms and onboarding flows.
For B2B specifically this lands harder than it does for consumer brands. Our entire targeting model leans on matching, onboarding, and account resolution, because we are chasing 200 named accounts and 11 people inside each of them, not a lookalike pool of two million.
When that resolution layer is owned by an agency that wants your media spend, “neutral match rates” becomes a phrase you have to take on trust. The independents in this market have spent years arguing that data ownership and execution should not sit in the same hand that bills your media. This deal just made that argument for us. The value of working with a partner whose data access does not route through a competitor’s balance sheet went up this week, quietly, for everyone outside the Publicis tent.
It is one reason we have always preferred raw signal access we control directly over processed scores rented from a layer somebody else owns. This was fundamental to how I built FunnelFuel’s data strategy, and I’m glad I did
Watch for the second-order move. If LiveRamp is now a Publicis asset, the other holdcos and the big independents need their own resolution layer or a credible neutral alternative, fast. That is the consolidation wave to track over the next two quarters, not the headline multiple.
2. LinkedIn quietly buried the launch-day model
The story. On 26 May, LinkedIn published a three-phase B2B launch framework arguing that product launches built around a single announcement moment leave most of the buying committee untouched. The numbers it leaned on are the real story. LinkedIn cited the average B2B buying journey at 211 days from Dreamdata’s 2025 benchmark, then the more recent March 2026 data stretching that to 272 days and 10 stakeholders, with up to 76 touchpoints across channels. Same month, LinkedIn launched an agency certification programme, citing lengthening buying cycles as the reason to raise standards among practitioners running its campaigns.
My take. Yes, LinkedIn has an obvious incentive here. “Launch day is dead, you need sustained presence” is a remarkably convenient conclusion for a platform that sells always-on budgets. Treat the framing with the scepticism it deserves.
The way their new algo is performing for organic posts could lead me to think you’d need to plan dozens of posts to land the message, a point which is a bit snide but the volatility in reach is real. That’s an aside to the bigger story
But the underlying data is not LinkedIn marketing. It is Dreamdata’s, drawn from millions of real customer journeys, and the direction of travel is the part worth sitting with.
The cycle did not just stay long. It got longer, year over year, from 211 days to 272. That is the opposite of what everyone predicted AI tooling and self-serve research would do to B2B buying. More information, more available, faster, and the committee is taking longer, not less.
This has been validated across scores of other research I’ve seen so I’m inclined to believe it
The practitioner read is straightforward. If the average deal now spans the best part of nine months and involves ten-plus people, then any paid media plan that spikes around a moment and goes quiet is buying impressions against a committee that has not started caring yet, and is dark by the time they do.
The money is not in the launch. It is in being consistently present and measurable across the dark stretch in the middle, where the committee is forming a view without ever clicking anything you can attribute.
This is the same point we keep making about high-value actions and the dark funnel. The signal that matters is rarely the conversion, as important as that obviously is, but the wider point is you need to build memory a long way upstream of hand-raising.
It is the quiet, repeated engagement from accounts that are months from a decision and will never fill in your form until they have already decided. Plan for the 272 days, not the launch week.
And as we’ve said repeatedly - when it comes to hitting the buying group, LinkedIn will over index in commercial roles and under represent back office personas. We need orchestrated coordination around the buying group triggered by cross pollinated data - engagement and intent signal coordinated between LinkedIn and other channels that represent the rest, which is where omnichannel stands heads and shoulders ahead
3. The open web’s discovery layer is collapsing in real time
The story. Google’s May core update began rolling out on 21 May and was being felt across the publishing world by the bank holiday weekend. Just what the premium publishing market needs, more volatility.
On 27 May, Google brought Preferred Sources into AI Overviews and AI Mode, added “Highly Cited” labels, and rolled out topic carousels, deepening the shift from blue links to synthesised answers.
Digiday ran a piece the same week on small publishers watching their traffic drain as AI reshapes discovery. And Comscore data showed Claude growing 130 per cent month over month in March, with ChatGPT still the dominant assistant, as the assistant layer keeps eating the top of the research funnel.
My take. This is not a B2B story on the surface, which is exactly why most B2B marketers are sleepwalking through it. Two things are happening at once and both matter to anyone running content-led demand gen.
First, the supply side of the open web is thinning. B2B programmatic runs on the open internet, on the trade press and the niche publishers and the long-tail business sites where buying committees actually read. These are the richest source of fresh third party intent
If AI Overviews keep absorbing the clicks and the small publishers keep folding, the quality inventory we target contextually gets scarcer and the survivors consolidate pricing power.
The “open web” half of the open web is quietly shrinking, and that flows straight into your CPMs and your reach against the accounts you care about.
Second, the discovery behaviour of your buyers is moving inside the assistants. When a procurement lead researches CRM platforms or a CMO scopes a martech category, an increasing share of that now happens in a conversation with ChatGPT or Claude, not a Google results page or a vendor site.
That research is invisible to your analytics and unservable by most of your paid media.
Google adding “Highly Cited” labels is a tell about where authority is being arbitrated now. The practical implication is uncomfortable but clear. Being the answer the model cites is becoming as important as being the result the buyer clicks, and almost nobody in B2B has a plan for it. Start treating model citation as a distribution channel, because your competitors who publish primary, citable expertise are already being read by the machine your buyer is asking.
The key is recognising that citations won’t drive clicks like the blue links of old, but the marketers gamble is they do build trust, brand memory and authority at crucial junctures in the research journey
4. The ratings currency is measuring pirates
The story. Adtech auditor Adalytics published a report on 26 May finding that illicit livestream services such as WatchSports and StreamSports99 are redistributing high-profile sport to hundreds of thousands of devices. In one example, a single pirated stream of the Super Bowl off NBCU’s Peacock reached more than 100,000 devices, one of over 100 sites redistributing that game. Adalytics estimates this year’s Super Bowl ratings were likely off by one to two million viewers. The NFL had already put Nielsen on notice last September over systematic undercounting.
My take. B2B money is flooding into CTV on the promise of premium, brand-safe, measurable reach. This report is a useful cold shower on the “measurable” part.
CTV is woefully Immature and onwards of 70% of the ad requests are essentially empty of the metadata which makes programmatic so valuable. Big players like Teads are working on solutions that remind me of my first days in the space around 2010 - hacking solutions to fix fundamental premises.
The whole CTV value proposition rests on a ratings and measurement currency that buyers treat as ground truth. Adalytics just demonstrated that the currency is counting streams nobody licensed and missing viewers nobody can find. If the measurement of the single most-watched, most-scrutinised broadcast of the year is off by a couple of million either way, the idea that your B2B CTV buy is being measured to a clean, trustworthy standard deserves a hard second look.
For our market the lesson is not “avoid CTV.” CTV works for B2B when it is bought against the right accounts and tied to real outcomes downstream. Our FunnelFuel team had a great week in advancing the scale behind account based targeting and reporting of connected tv. I’m bullish on it
The lesson is to stop outsourcing your confidence to a ratings number and insist on outcome-based measurement that connects the impression to account engagement and pipeline. Reach you cannot verify is not reach. It is a story you are paying for. Measure the thing that pays you back instead.
In case you missed it
ChatGPT ads got cheaper and B2B still has no way in. OpenAI removed its advertising spend minimum entirely on 5 May, having cut it from 250,000 dollars to 50,000 in April, and CPMs have slid from a 60-dollar launch price to as low as 25 through the Criteo pipeline. Impressive compression. But the platform’s allowed verticals still skew consumer (household goods, local services, travel, digital products), B2B is not really invited yet, and the deeper problem is unchanged. As TAU’s Robert Webster put it to Digiday, the real moat for Google and Meta is owning measurement, not the interface, and until someone independent can verify what a ChatGPT impression is worth, advertisers are taking OpenAI’s word for it. Much of the early spend is still test-and-innovation budget chasing FOMO.
Meta cut 350 Irish jobs, blaming AI. Around 20 per cent of its local workforce, with AI named as the driver. File under the slow reality that the efficiency story the platforms sell advertisers is the same story they are telling their own headcount.
The Trade Desk’s Q2 guide landed soft. A roughly 750 million dollar Q2 forecast was read as weak and the stock dropped after the 7 May call. We have covered TTD’s year to death, so just noting it. The pressure on the independent DSP from Amazon and the walled gardens is not letting up.
Dreamdata’s number worth pinning to your wall. 272 days. 10 stakeholders. If your reporting still judges a campaign on what converted this quarter, you are grading a nine-month process on a six-week sample.
Have a great weekend
Mike
Sources
• Publicis to acquire LiveRamp, SEC filings (Form 8-K / DEFA14A, LiveRamp Holdings), 17 May 2026
• “Ad Tech Briefing: Agencies reposition for the agentic era,” Digiday, 26 May 2026
• “LiveRamp announces Q4 and FY2026 results,” LiveRamp / SEC, 17 May 2026
• “LinkedIn’s three-phase B2B launch framework,” PPC Land, 26 May 2026; Dreamdata 2026 LinkedIn Ads B2B Benchmarks
• “Pirated Sports Streams Are Warping TV’s Most Important Ratings,” AdExchanger, 26 May 2026
• “AI ads, data deals, and a search shakeup define the week,” PPC Land, 28 May 2026
• “Advertisers still skeptical of ChatGPT ads,” Digiday, 26 May 2026; ChatGPT Ads guides citing Digiday CPM and minimum-spend reporting
• “By the numbers: Ad tech’s quarter of mixed fortunes,” Digiday, May 2026