How we verify a claim before we publish it

This page carries one non-negotiable rule: no number ships without a source. That sounds obvious until you notice how much acquisition-industry content doesn't follow it, quoting a stat from a stat from a screenshot, three or four steps removed from anything you could actually check. We built our editorial process to be the opposite of that.

Two kinds of claims get published here, and each gets a different treatment. Public claims, the kind that circulate on social media or in industry newsletters, get recomputed from the primary dataset before we repeat them, never taken secondhand. Our own claims, drawn from OA's proprietary listing database, get pulled from a live query against production data at write time, timestamped, and cited with enough detail that anyone could reproduce the same result.

The mature-firm closure verification elsewhere on this page is the concrete example of the first kind. A stat about small-business closures was circulating widely. Rather than repeat it, or dismiss it, we downloaded the actual Census Bureau microdata tables it was supposedly built from and recomputed every figure from scratch. The headline number held. The mature-firm number held. Even the one place the original claim oversimplified, a record-high framing that missed a documented data artifact in a single outlier year, we caught and corrected, and the underlying claim still held once we accounted for it. That is the standard: not just is the number right, but is the framing around it right too.

The second kind, our own market-pulse pieces, follows a parallel discipline. When we say a given industry's median multiple sits at a specific figure, that number came from a query against our production listings table run on the date stated, filtered and aggregated the same way every time, not eyeballed off a dashboard or remembered from last quarter. If the underlying data changes, the next piece using it gets a fresh pull, not a stale number carried forward out of convenience.

We also try to be honest about what a number does and doesn't prove. Census tracks that a firm closed, not why. Our own database tracks what's listed and at what price, not what actually closes at that price versus a negotiated discount. Every piece on this page tries to state the limitation alongside the finding, rather than let a clean number imply more certainty than the underlying data supports.

This is also why the page exists in the first place. A marketplace that gets its own numbers wrong has no business asking anyone to trust the deal data behind it. Holding ourselves to the same bar we'd apply to a competitor's claim is the whole point.