Can You Trust a Site Audit? 5 False Positives Among 11 Findings
Published
This article was first published in Korean by Intrix Lab on August 30, 2026.
Act on site audit results after reproducing the evidence behind them. After Intrix ran its own audit tool on 148 pages of intrix.kr and verified all 11 findings, 5 were false positives and 6 were actual site defects. This was the result of a single measurement on 2026-08-28. Our own site is an extreme sample that frequently covers audit-related topics, so 5/11 cannot be read as the audit tool's overall false positive rate. Rechecking the evidence behind each finding comes before the score.
How did we recheck the 11 findings?
The starting point was our own audit results. For each of the 11 findings, we reopened the underlying source material and checked whether the issue could be reproduced on the actual site.
Tracing the verdicts back to their source material separated the 11 findings into two groups. 6 remained defects to fix on the site. The other 5 belonged to different categories of false positives. This is why the same results screen called for both site fixes and changes to the assessment logic.
Where did the 5 false positives come from?
The false positives involved checks for whether links were still live, schema compliance, popup and advertising language, and platform identification. For example, an external site blocked bot requests but responded normally to browser requests, and this was interpreted as a link that no longer existed. Removing the link based on a single response code could have led us to change a working path unnecessarily.
What caused the five false positives?
The common cause was a failure to distinguish words explaining a topic from evidence that the corresponding feature was actually operating. Pages explaining platforms, popups, or advertising also contain related language.
The body content on intrix.kr frequently covers measurement, platform reviews, and cautions about advertising language. This content mix exposed the weaknesses of word-based assessment in a single audit of our own site. Applying our assessment logic back to our own site became a validation step for identifying categories of false positives.
What should you check first after receiving audit results?
First, check the URL and the location in the source text supporting each finding. The presence of certain wording and the actual operation of a feature are different facts.
Then handle site defects and defects in the assessment logic separately. Fix reproducible issues on the site, like the 6 in this measurement. Fix issues that point to weaknesses in the assessment logic, like the 5 false positives, in the audit system. Intrix's free SEO and GEO audit is a starting point for reviewing verdicts alongside their evidence and deciding what to address first.
How broadly do these numbers apply?
The population was a single run of our own audit tool that crawled 148 pages on intrix.kr as of 2026-08-28. The proportion of 5 false positives among 11 findings cannot be extended to other sites or the audit tool as a whole. Our site has substantial content explaining measurement and platforms, creating conditions in which false positives from word-based detection patterns were more likely to be sampled.
This case shows only that reproducing every audit finding can separate site defects from defects in the assessment logic. Findings for your current site also need their evidence checked through the same process to identify what actually needs fixing.
Frequently asked questions (FAQ)
Q. Should I immediately change my site if its audit score is low?
Before deciding what to fix based on the score alone, check each finding's URL and source evidence. It is safer to act after separating reproducible site defects from false positives in the assessment rules.
Q. Doesn't HTTP 403 mean a link is broken?
Not always. In this case, external sites returned 403 or 999 to bot requests but 200 to browser requests. You need to recheck under different request conditions to determine whether a link is still live.