Brand
What we got wrong the first time
Most teams write a 'limitations' section and bury it in a footnote. We put 'What we got wrong' above 'Where Lighthouse loses' on the evals page. Same visual weight as the headline numbers. Admitting the dumb thing is the part that earns the right to publish the wins.
The decision was about where the section sits, not what the section says.
On the Lighthouse evals page, the headline numbers are at the top. Below them is a section titled What we got wrong the first time. Below that, Where Lighthouse loses. The autopsy sits between the wins and the losses, in the same typography, at the same indentation, with the same heading weight. The reader cannot scroll past the wins without seeing the corrections.
That placement is the post.
The conventional shape
The conventional shape is a footnote. An asterisk on the third paragraph of a methodology appendix. A collapsed Limitations accordion below the fold. A separate page called Notes that nobody links to from the main report. A PDF.
All of these shapes share a property: a buyer skimming the page on a Tuesday afternoon will not encounter the limitations. The page reads as a list of wins. The limitations are technically published and technically reachable, and none of that matters because the reader does not work that hard.
The conventional shape is built for a reader who is auditing. The audience for a product page is a reader who is skimming. Treating them as the same person is how a Limitations section becomes ornamental. The page picked the auditor and lost the skimmer. The skimmer is the buyer.
Our shape
The autopsy goes between the wins and the losses, at the same altitude as both. It names the bug — empty-string responses scored as neutral instead of dropped, which inflated the headline numbers on the first publication. It names the fix. It names the post-cleanup numbers, including the ones that moved against us: Llama dropped from neutral to −2.6.
Same typography. Same heading weight. Same vertical rhythm. No grey-out, no collapse, no accordion. The autopsy is not a footnote on the wins. It is a sibling.
The reader who skim-reads the page sees three blocks at the same scale: what worked, what we got wrong, what still doesn't work. The page makes no attempt to hide the second block behind the first.
What the placement does
A sanitised report is a marketing asset. An honest report is an evidence asset. The two surfaces look identical except for one section, and that section is what distinguishes them.
The autopsy section is doing the same emotional work as the losses section. It is proving the page is not sanitised. The wins read differently after the reader has seen the autopsy, because the reader now knows the page is willing to publish numbers that hurt. The wins acquire a property they did not have before: they were not selected for.
This is not a courage move. It is a positioning move. The wins are only credible because the corrections are visible. Remove the autopsy and the page becomes a brochure — same numbers, but the reader cannot tell whether the numbers survived a cleanup or whether the cleanup ever happened. The autopsy is the receipt.
The cost
The cost is real. A buyer who skim-reads sees Llama got worse with Lighthouse and closes the tab. That happens. We do not try to recover them, because the recovery move is to soften the autopsy, and softening it removes the property that earned us the rest of the page.
The buyers who do not close the tab are the buyers we want. They have been burned by a vendor whose evals page turned out to be a brochure. They recognise the shape of an honest page because they have been looking for one. Losing the skimmer-who-leaves is the price of being legible to the buyer-who-stays.
This is the same trade as empty versus empty, where the bug autopsy lives in full, and the same discipline as no broken links the second time, where we refused to sweep redirect failures under a 200. The pattern is the same across the surface. Things that went wrong stay visible. We do not get to compound trust on a page that hides its mistakes.
Which sections belong above the fold
Wins belong above the fold because the reader has to know what the page claims.
Limitations belong above the fold because the reader has to know what the claim costs.
Methodology belongs below the fold, because by the time the reader cares about methodology, they have already decided to read further. The reader who reaches methodology has self-selected; the reader who sees the headline has not.
Most pages get this inverted. They put the wins above and the limitations below, on the assumption that limitations are an asterisk on the wins. They are not. The limitations are what makes the wins legible as evidence rather than as marketing. They belong at the same altitude as the thing they qualify. The reader sees them first, and the trust compounds from there.