Why do ad platforms hide performance data?

Every ad platform knows precisely how every campaign on it performs — the data exists, indexed, queryable, and reconciled to the penny for billing — and almost none of it is published. Withheld ad performance data is not an oversight or a feature nobody has requested. It is three separate incentives pointing the same direction, and only one of the three belongs to the platform. (Disclosure: this is published by BidSurvivor, which publishes its own numbers, for reasons that are also self-interested and stated at the end.)

Reason 1: averages sell, distributions don't

Platforms publish benchmarks. They publish average click-through rates, average costs per click, average conversion rates by industry — and those numbers are flattering by construction, because a small number of very large, very well-run campaigns pulls the mean well above the median. When an agency quotes you the 2026 Google Ads benchmarks — roughly $5.42 a click across all industries, an 8.18% average landing-page conversion rate — those are means. The typical advertiser is not the mean advertiser.

A distribution would answer the question the mean is standing in front of: what fraction of campaigns on this platform lose money? Nobody publishes that, anywhere, and the reason is not technical. Every platform could produce a histogram of return by account decile in an afternoon. The histogram would show a long left tail, and no sales deck survives contact with a long left tail.

Publishing the mean and withholding the median is not lying. It is lighting.

Reason 2: price discovery hurts the house

This is the deepest of the three, and the one platforms will never say aloud.

If every buyer could see what comparable placements actually cleared for, and what those placements actually delivered, negotiation would collapse toward the data. That has happened in every market that gained transparency — equities after the tape, airfares after the aggregators, freight after the load boards. The spread narrows and it narrows permanently. Opacity is not a side effect of complexity; it is the margin.

You can see the shape of this in which transparency initiatives the industry actually adopts. Brand safety gets standards, certifications and third-party verification, because advertisers demand it and it costs the platform nothing competitive. Comparative performance — what the same dollar bought you here versus there — gets nothing, ever, because that is where the money is.

The one time the ledgers were forced open, the result was instructive. The ISBA programmatic supply chain study, run with PwC, followed real advertiser money through real supply paths and found that roughly half of what advertisers spent reached publishers at all, with a substantial slice untraceable — not stolen necessarily, just unaccounted for, because no participant in the chain published enough to reconcile it. Auditors with contractual access found that. No dashboard would have.

The related failure mode is what happens to money that leaves the traceable chain entirely. Juniper Research puts advertising spend lost to fraud at $68 billion a year. That number is only possible in a market where the buyer cannot check the delivery.

Reason 3: advertisers are complicit

The quiet third reason, and the one that makes the first two durable.

Buyers do not want their numbers public either. A failed campaign is nobody's press release. Marketing careers are built on dashboards shown selectively, and a platform that offers privacy is giving a great many of its customers exactly what they would ask for if asked. When a platform says "we can't share that, it's advertiser-confidential", it is frequently true, and the advertiser is grateful.

This is why transparency campaigns that frame the platform as the villain tend to stall. The coalition is smaller than it looks. Most of the people who would benefit from an open market have something in the drawer they would rather not open.

What "transparent" usually means in a pitch

Four phrases worth translating, because they all appear in decks marked transparency:

• "Full funnel visibility" — you can see your own numbers in more detail. Not anyone else's. Not the distribution you sit in.
• "Third-party verified" — an outside vendor confirms the ad was served to a plausible human. That is delivery, not performance.
• "Open auction" — the auction is open. The fees taken between your bid and the publisher's receipt usually are not; that is the gap the supply-chain audits found.
• "Benchmark data" — means, dressed as expectations.

None of these is a scam. Each answers a narrower question than the word suggests, and the narrowing is always in the same direction.

What a buyer can actually do

Four things, in rough order of how much they are worth.

Prefer platforms that publish something real about other people. It is a short list — here is the list, and the ad rate comparison puts their published numbers side by side — and the test is simple: can you see what somebody else paid and what they got, before you spend anything?

Demand placement-level receipts rather than summaries. A summary is a number the seller chose. A receipt is a row per placement with a price and a count on it. If a platform cannot produce rows, it has decided not to.

Convert every quote to cost per visit yourself. CPM, CPC and flat fees are not comparable until you do, and the unit a seller quotes in is chosen to flatter the number. The arithmetic is here, and it is two lines long.

At the test stage, weight verifiable small channels above unverifiable large ones. Certainty about a small number beats a large number you cannot check, and at test volumes the verifiable option is usually cheaper anyway. The full case for that trade is in why fast tests need honest numbers, and the counter-argument — that a small channel cannot tell you about scale — is real and worth holding onto.

The counter-argument, stated properly

There is a serious defence of opacity and it is not "we like margin". A platform that published every advertiser's results would leak commercial information the advertiser owns: launch timing, product mix, what a competitor's acquisition costs look like. Aggregate data can be de-anonymised at small n. And a public leaderboard of performance would, predictably, be gamed the moment it existed.

The honest version of the position is that some opacity protects buyers and the rest protects the seller, and no platform has ever been asked to say which is which. That is the question worth putting to one.

Our own incentives, stated

BidSurvivor publishes every brand's card opens, click-throughs and clearing prices, and the live totals sit on the front page and on each brand's own page. That is not virtue. At our size transparency is a competitive weapon rather than a cost: we cannot out-reach anybody, so we out-verify everybody, and the numbers are currently small enough that publishing them costs us nothing we were going to get anyway.

If the board were a thousand times bigger, keeping that promise would hurt, because the published numbers would then include failures at scale. We are saying so now so you can hold us to it later. The rules the board runs on, including the ones that cut against us, are in how BidSurvivor works. The day the numbers on the front page and the brand pages get vaguer is the day we have become the thing this article is about.