The Bid Request Has Receipts: Eric Tilbury on Clean Supply, ID-Free Buying, and the Craft of Programmatic

Full analysis

Eric Tilbury, who runs programmatic ops and solutions engineering at Inuvo, spent an hour on Signal & Noise with hosts Rio Longacre and Brett House laying out a plain argument: most programmatic buyers are running their campaigns wrong, and the two biggest mistakes are letting the DSP optimize to a vanity number and buying ID-dependent inventory while believing they are buying contextual. There's no deal, no funding, no earnings print here. The value is diagnostic. So the question for an operator isn't "what just changed" but "is the diagnosis right, and does it change where I spend the next quarter's effort?"

How hard is this to undo? Easy. Nothing here forces a move. An operator can test one of Tilbury's claims on a single campaign line and reverse it by Friday. That lowers the bar for trying something and raises the bar for the claims themselves.

What's actually being decided: whether the ID-free, iOS-is-underbought, incrementality-over-ROAS worldview is a real edge an operator should chase, or a well-argued pitch from a vendor whose business depends on it being true.

The Market Analyst

Follow the money through the conversation. Inuvo is pivoting its whole business away from search arbitrage and into IntentKey, its ID-free contextual product. Every major claim Tilbury makes points customers toward that product. ID-free inventory is undervalued. iOS is abandoned. Probabilistic ID matching is "complete bullshit." Open-web auction volume is declining as users sit inside ChatGPT. Each of those, if true, raises the value of contextual decisioning. That doesn't make them false. It means grade on evidence, not conviction.

For an informed non-specialist: Tilbury sells a product that works without tracking people, and his whole argument is that tracking people is broken. Convenient, and possibly correct.

The Ruggable story in the saved reading is the real-world version. A direct-to-consumer brand cutting Meta performance spend because costs rose and returns fell. That's the demand-side pressure that makes contextual pitches land.

The Skeptic

Steelman the case against Tilbury and it mostly holds, which is the problem. The claims that would most help Inuvo are the ones he can't source. ID-bearing traffic monetizes "30 to 40% higher" than ID-free: unattributed. Rio Longacre's "70% drop in clicks from Google in 2026": the episode's own fact-check calls it almost certainly overstated. "99% of buyers don't know to flag ID-free bid requests": a round number that means "almost everybody" and nothing more precise.

But strip the oversell and a hard core survives. DSPs will hit a CPA goal by finding bots if you give them no supply controls. That's mechanically true. sellers.json, the file that discloses who actually sold an impression, really is buried or absent in most DSP interfaces. CTV fraud really does concentrate outside the top apps. Those three don't depend on Inuvo being right about anything.

In plain terms: the sales pitch is inflated, the engineering diagnosis is sound.

The Operator

Tuesday morning, what do you actually do with this? Three things, in order of how fast they pay back. First, pull sellers.json and supply-path reporting on your live CTV buys and see how much spend sits outside the top 10 to 20 apps. If Tilbury is even half right about fraud concentration, that audit finds waste in an afternoon. Second, find the ID-free bid request flag in your DSP and check whether it's on. Third, run one iOS line item against a holdout before you kill it for showing no conversions.

What breaks at 90 days: the incrementality religion. Tilbury's own line is the giveaway. "If buyers actually understood measurement, half the industry would be screwed." The people presenting ROAS in the quarterly review are the people whose jobs depend on ROAS looking good. Auditing supply is cheap. Telling your client their Meta spend drove zero incremental sales is a career event.

For the non-specialist: finding the waste is easy. Telling the boss the waste exists is the hard part.

The CFO

The incrementality argument is the one that moves real money, and it's the oldest story in the deck. The Uber case Tilbury cites, where pausing Meta spend showed no drop in installs, is years old and the actual findings were more nuanced than "zero lift." Media-mix modeling, the statistical method for estimating what each channel actually caused, has been the correct answer for a decade. What's new is that LLM-assisted analysis drops the cost of doing it. You no longer need a team of data scientists to build the model.

That cost collapse is the genuinely interesting part for a P&L. The reason most advertisers run on platform ROAS instead of incrementality was never that they thought ROAS was true. It was that real measurement was expensive. Take the cost out and the excuse goes with it. Whether buyers actually switch is a different question, because the switch exposes spend that was never working.

The Customer / End User

Put yourself on the receiving end as a DTC advertiser. You've been told for years that Meta and performance CTR are where growth comes from. Ruggable just tore that up. The customer here isn't asking for ID-free targeting. They're asking why their costs went up and their returns went down. Contextual and incrementality are answers to that question, which is why the pitch works now and didn't in 2021 when performance spend still paid.

For the non-specialist: brands aren't suddenly in love with privacy-friendly ads. They're unhappy that the old ads stopped working and they're shopping for a reason.

Where the council splits

Two real disagreements. The Skeptic and the Operator agree the supply-hygiene diagnosis is solid, but split on the ID-free opportunity: the Operator says test iOS cheaply and find out, the Skeptic says the headline numbers propping up that opportunity all come from a man selling ID-free inventory. The CFO and the Operator split on incrementality: the CFO says the math is finally cheap enough to do, the Operator says doing it surfaces results nobody in the room wants to own. That second tension is where the whole episode actually lives. The barrier to real measurement was never capability. It's that honest measurement embarrasses the people who commission it.

What it hinges on

Strip it down and operators face two separate questions wearing one coat. Is the supply-path and CTV-hygiene critique right? Almost certainly yes, and it costs an afternoon to verify, so verify it. Is ID-free the undervalued goldmine Tilbury claims? Unproven, sourced mostly to a vendor, and testable on one holdout line before you believe the CPM-premium math. The council leans: act on the hygiene, pilot the ID-free, discount the sourcing on anything that happens to sell IntentKey.

Prediction: Inuvo's full-year 2026 revenue, reported in its Q4 2026 results around February 2027, will be lower than its 2025 revenue, as the IntentKey pivot fails to replace the search-arbitrage business it is walking away from fast enough.

Confidence: Medium. The pivot is deliberate and self-described, but the timing of the crossover is hard to call.

Why: Tilbury says plainly that Inuvo is pivoting fully toward IntentKey and away from its legacy search-arbitrage business, and the macro reason he gives is that open-web auction volume and search clicks are falling as users move into LLM chat interfaces. That is exactly the revenue base the old business sits on, so it's shrinking under the company by its own account. Replacing a mature, cash-generating arbitrage line with a newer contextual product is a multi-quarter job, and the new product's whole market thesis is that buyers don't yet know to want ID-free inventory, which means a slow sales cycle. When a company abandons its revenue engine before the replacement is at scale, the transition year usually prints a decline. The less likely path is that IntentKey scales fast enough inside a single year to cover a search business that's eroding at the pace Tilbury himself describes.

Revisit by 2027-03-01: We're right if Inuvo's reported full-year 2026 revenue is below its 2025 figure. We're wrong if full-year 2026 revenue is flat or higher than 2025.