the source laundering problem

Due Diligence Desk · The Local Aim · April 2026

A statistic is making the rounds in marketing right now. You may have seen it in a pitch deck, a vendor email, or a trade publication. It goes something like this: artificial intelligence is cutting irrelevant ads by 40 percent, search queries are doubling, and AI-powered ad spending is on pace to hit $56 billion this year.

These numbers appear in The New York Times. They are attributed to real analysts and real executives. They look like facts.

Some of them are. Some of them are not. And the mechanism that makes it hard to tell the difference is what this desk calls source laundering.

What Source Laundering Is

Source laundering is not fabrication. Nobody is making numbers up. It is something subtler — and in some ways more dangerous because it is harder to catch.

Here is how it works.

A company — Google, Meta, a marketing platform — generates a statistic about its own product. That statistic gets quoted by a journalist in a credible publication. A vendor reads the article and cites "The New York Times says AI is cutting irrelevant ads by 40 percent." A prospect reads the vendor's pitch deck and sees a Times citation. The original source — Google, talking about Google — has disappeared. What remains looks like an independent finding.

It is not an independent finding. It is a company claim that passed through a respectable outlet and came out the other side looking like verified research.

The outlet did not lie. The journalist reported accurately what the executive said. But the statistic was laundered in transit. By the time it reaches a small business owner making a budget decision, the conflict of interest has been scrubbed clean.

A Recent Example

On April 29, 2026, The New York Times published a piece on how AI is reshaping the digital advertising industry. The article cited several statistics that are now circulating in marketing circles.

One of them: irrelevant ads have declined 40 percent due to Google's AI improvements.

The source is Dan Taylor, Google's Vice President of Global Ads, speaking at a Google media roundtable. Google is announcing that Google's product is working. There is no independent measurement. There is no definition of what counts as "irrelevant." There is no methodology. It is a company executive making a claim about his company's product at a company event.

The Times reported it accurately. The Times did not verify it independently. That is how trade journalism works — and that is fine, as far as trade journalism goes. The problem starts when the claim escapes that context and gets cited as industry fact.

Another statistic from the same article: Google search queries have doubled in two years. Source: "the company said." Google is reporting its own search volume. No independent measurement. No definition of what counts as a query. No audit.

These are not lies. They are self-reported claims from parties with a direct financial interest in the conclusion. That distinction matters enormously when you are deciding where to spend money.

Why It Works

Source laundering is effective for three reasons.

First, credible outlets confer credibility. When a statistic appears in The New York Times or The Wall Street Journal, most readers extend trust to the number itself — not just to the reporting of what someone said. The publication becomes a co-signer of the claim, even when the publication is only documenting what a company announced.

Second, the original source gets buried fast. By the third citation, the chain looks like this: vendor cites article, article cites executive, executive cites internal data nobody can see. The vendor pitch says "industry data shows." The industry data is a press release.

Third, the numbers are usually directionally plausible. AI advertising is growing. Search volume is increasing. Irrelevant ads are probably declining somewhat. The laundered statistics are not random — they are inflated or unverifiable versions of things that are probably true in some form. That plausibility is what makes them stick.

The Three Questions That Cut Through It

This desk evaluates every claim — vendor pitch, trade article, industry report — against three questions.

Who funded it? A statistic generated by the company that profits from the conclusion is not independent research. It is marketing. It may be accurate marketing. It is still marketing.

What was the methodology? A percentage with no denominator is not a finding. "Irrelevant ads declined 40 percent" requires a definition of irrelevant, a baseline, a measurement period, and a methodology that someone other than Google can audit. Without those, the number is a talking point.

When was it published? Digital marketing moves fast. A 2023 study on AI search behavior is not a 2026 fact. The desk applies a two-year recency threshold for digital marketing research. Anything older than that gets treated as historical context, not current evidence.

Any claim that cannot answer two of those three questions gets discarded. Not hedged. Discarded.

What This Means for You

If you are a small business owner evaluating a vendor, a platform, or a marketing service, you will encounter laundered statistics constantly. They will appear in sales decks, in emails, in articles your vendor forwards you. They will look authoritative. The citations will be real. The underlying data will be unverifiable.

The question to ask is not whether the publication is credible. The question is whether the original source has a financial interest in the conclusion.

Google saying Google's ads work better is not a finding. It is a press release with a byline.

A marketing agency citing that press release as industry research is not evidence. It is the press release one step removed.

The way to protect yourself is not to distrust everything. It is to ask the simple question: who told you this, and what did they have to gain from telling you?

Most of the time, the answer explains the statistic.

The Desk's Standard

Every claim that runs on this desk — in the Buyer Beware column or anywhere else — is evaluated against a single standard: named source, disclosed methodology, verifiable date. A vendor claim that passes through a major publication does not automatically meet that standard. The publication's credibility is not transferred to the claim.

That standard is not complicated. It is just consistently applied. Which is more than most of what gets forwarded in a sales email can say.

The Local Aim Due Diligence Desk · Orange County, CA · April 2026 Independent. Verified. No Hype.

Have a vendor claim, marketing statistic, or industry report you want evaluated? Submit it to the Due Diligence desk: kirby@thelocalaim.com