Key Takeaways
- B2B programmatic works best when marketers target accounts, roles and buying groups instead of relying on person-level signals and click-based metrics.
- Intent data, long-form media engagement and protected internal information can help marketers focus spend on in-market buyers and improve account-level decisioning.
CTOs rarely get the stage to themselves— as Red Kite’s Julian Lloyd Evans joked, they tend to tell the truth. Kicking off the B2BMX Summit at Advertising Week New York, Funnel Fuel CTO Adam Johnston did just that. He explained why programmatic sits at the heart of B2B advertising and what it takes to make consumer-built pipes work for buying groups.
Johnston first big realization after moving to B2B, having spent most of his career in B2C that include building ad tech at Sky and serving as CTO at a DSP, was simple: programmatic runs on individual transactions, while B2B runs on organizations.
That gap pushed Funnel Fuel to build layers on top of standard infrastructure to represent accounts, roles and buying groups. “We don’t try to talk to individuals. We’re not interested in named people,” Johnston said. “We’re interested in roles. We’re interested in accounts.”
The heavy lifting happens directly with DSP and SSP partners, so account-level impression data is native to trading rather than bolted on later. Sometimes all that engineering produces one new spreadsheet column. That column tells you which account you actually reached.
Clicks Are a Proxy, Not the Prize
Nobody buys enterprise tech on impulse. B2B buying cycles run long, involve committees and demand repeated touchpoints over months. “It’s not like we’d have a one-click pair of trainers, right?” Johnston said. “We have to engage customers in multiple creative moments over a period of time.”
For marketers, that means clicks and impressions are signals along the journey, not the finish line. Real success shows up as account engagement that builds toward a buying decision.
Harder Because It’s Easier
Big B2B budgets attract a crowd. Venture-backed challengers now compete with incumbents, and Johnston admitted differentiation is tough, especially in the crowded U.S. market Meanwhile, AI has supercharged the build cycle. Johnston wrote his own code until a year ago. Now AI handles much of it, running analysis on historical data that no human team could finish in any feasible time frame.
“I think it’s a combination of: it’s harder because it’s easier,” he said. You can prototype faster, but so can everyone else.
Only about 5% of B2B buyers are in market at any moment. Funnel Fuel’s new partnership with Bombora delivers raw intent data at the account and device level to help find them. That data, Johnston said, “gives us a great opportunity to really zone in on that 5% and not expend client media money on the other 95%.”
Once you find them, channel matters. Buyers still stream video, watch CTV and listen to podcasts, sometimes on weekends. A contextually relevant ad break in a business podcast is fair game, and long-form audio and video deliver stronger engagement signals, as long as you can tie them back to the account.
Your Briefs Are Data, Too
Johnston’s top tip for attendees was stop treating delivery logs as the only data that counts. Campaign briefs, internal documents and firmographic and technographic data all carry information AI can mine. “Protect your data, treat your data as your first-class citizen,” he said. He even prefers the word “information,” since “data” feels like raw currency.
At Funnel Fuel, he built an AI playground holding all of it, then asked the machines what had worked. The answers surfaced patterns the team hadn’t consciously built: which creative won with which publishers, regions, devices, times of day and accounts.
Inventory forecasting has frustrated Johnston at every stop. Now he’s testing whether AI can make it even 5% to 10% better, forecasting in accounts and buying groups instead of standard DSP units as well as exploring agents that understand organizations, enforce contracts and decide whether an impression deserves a bid.
The guardrail isn’t optional. “Put your data in a super safe environment where if it does get destroyed, it’s fine,” Johnston said. “You just restore it and you go again.” Never point experimental AI at production systems.
Can Location Prove the Account?
In answering inquires from the audience, location signals do flow through the bid stream, and tying an impression to a specific company building would be a powerful account signal. The catch: publishers don’t emit those signals consistently. Funnel Fuel saw real success in early tests, then cross-checked the results. “Some of the impressions are being distributed, when we cross tested them, not where the programmatic system told us they were coming from,” Johnston said.
Testing continues, including a close look at Esri’s location data. Until verification is consistent, treat location as a supporting signal, not the source of truth.
As for where Should B2B AI Live, Johnston stated if agents will transact B2B media, someone has to decide where that intelligence sits. Johnston didn’t pretend to have the final answer. “Where should that technology be? Should it be inside the DSPs, having some great conversation with SSPs?” he asked.
Advertisers are pushing from their side, too. Clients increasingly want impression-level data piped straight into their own systems, ending decades of copying, pasting and pivoting. The answer may be all of the above. Johnston’s parting advice: choose excitement over fear and focus on “making these machines work for us.”