What is a conversion signal, and are there really 200?
Semrush publishes 140 checks, Ahrefs 170, Screaming Frog 300, Baymard 700. Revslip's 200 is ordinary, and a count is scope rather than quality.
"Two hundred signals" reads like a number someone chose. A conversion signal is one named thing on a page with a threshold, a mechanism and a cost attached. The count means nothing unless all three hold.
The suspicion is fair. Every tool here leads with a count, no two counts match, and almost none will tell you what one item on the list is. The number does the work a specific claim should be doing.
What is a conversion signal?
A conversion signal is one named condition on a page that can be measured against a threshold and priced. Revslip checks roughly 200 of them across five groups: copywriting and messaging, offer and pricing, trust, UX and design, and technical and tracking. Each names a thing, a limit, and what crossing it costs you.
Signal 36 is "more than seven form fields on the primary form". Signal 50 is "body text below a 4.5:1 contrast ratio", the WCAG Level AA minimum for normal text. Signal 59 is "largest contentful paint above 2.5 seconds". Signal 37 is "the form asks for a phone number".
You can check each of those on your own site in a minute, and you can disagree with the threshold. Both are the point. A signal you cannot argue with is a slogan.
Is 200 conversion signals a big number?
No. It sits in the middle of what the category already publishes. Semrush states over 140 checks, Ahrefs 170, Screaming Frog over 300, and Baymard over 700 UX guidelines. Measured against its own field, Revslip's 200 is ordinary, and that is the honest reading of it.
- 140+ website issues checked, Semrush Site Audit
- 170+ SEO issues, Ahrefs Site Audit
- 300+ issues, warnings and opportunities across 23 categories, Screaming Frog
- 700+ UX guidelines from 200,000+ hours of user testing, Baymard Institute
Counts from each vendor's own current documentation: Semrush Site Audit, Ahrefs Site Audit, Screaming Frog, Baymard. All four sell something, which is why the spread is interesting.
Semrush and Baymard differ fivefold because they count different objects. Semrush counts faults a crawler can detect in markup. Baymard counts behaviours watched in moderated testing with real people. Neither is inflating anything. They have different units, which is why a raw count cannot be compared across tools.
Google makes the same case by accident. The Lighthouse performance score is built from five weighted metrics, total blocking time at 30%, largest contentful paint and cumulative layout shift at 25% each. The dozens of Opportunities and Diagnostics in the same report feed the score not at all.
A check count tells you how wide the net is. It says nothing about whether there is anything in it worth pulling up.
Want the arithmetic on your URL? Revslip prices each finding it names.
What do conversion audits check?
Five areas, in the order they usually cost money. Copywriting, whether the page says what you do. Offer and pricing, whether the deal is legible. Trust, whether a stranger believes you. UX, whether the path works at 360 pixels. Technical and tracking, whether the page loads and whether you would know if it did not.
The items inside are specific. Pricing hidden behind "contact us". Testimonials with no name or company. A cookie banner over the primary button. Validation errors that appear only on submit. GA4 with no key events. Each is a thing you can look at.
Baymard shows why thresholds beat opinions: the average checkout runs 5.1 steps and 11.3 form fields, where most sites need eight. Our threshold of seven sits deliberately below that, reasoning in how many form fields is too many. Baymard's own conversion audit page quotes 11.8 fields from another benchmark year, so even a careful figure moves. Use the direction, not the decimal.
What turns a conversion signals list into something usable?
Three properties, and a check missing any of them is padding. It needs a threshold you can test against, a mechanism that explains why crossing it loses sales, and a monthly cost so it can be ranked against everything else. Without the third, you get homework instead of a decision.
- A threshold. Seven fields, 2.5 seconds, 4.5:1. Not "too many", not "slow".
- A mechanism. Named, arguable, traceable to research rather than to a hunch.
- A price. The published estimate, so the list can be sorted by money.
traffic × CVR gap × AOV × mobile weight
Worked through: 6,000 monthly visitors, 1.1% converting now, 1.9% as the target taken from your own best page, €74 average order value, 60% of the loss on mobile. That is 48 extra sales, €3,552, weighted to roughly €2,131 a month. That is the free tier's estimate, bounded by the traffic figure, so rerun it when traffic moves. On a paid plan, tracked impact through the snippet and the integrations replaces it. More in are website audit revenue estimates accurate.
How do you check the count yourself?
Read the list. This blog is the 200 published one signal at a time, 32 posts in, and every post names its threshold, its mechanism and the arithmetic behind its euro figure. If a signal cannot survive its own post, it should not be in the count.
Second check: recompute a euro figure by hand. The formula above is the whole thing, not a simplified version of a hidden model. Three of its four inputs are yours.
Prefer the numbers to the arithmetic? Run a free audit.
Which checks Revslip does not run
Three, named plainly. Revslip is not a technical SEO crawler, not a usability test with real people, and not an accessibility conformance review. Each is a different job with a different tool behind it, and claiming otherwise would undo the argument above faster than any competitor could.
Hreflang, pagination logic and canonical sprawl across 40,000 URLs are Screaming Frog's work, so if that is your real problem, buy Screaming Frog rather than an audit that will barely mention it. Nobody is recruited and nobody thinks aloud, so to learn why one visitor hesitated, watch recordings or run a 5 second test beside it. Rank the suspects by cost first, which is how to read a conversion rate that is dropping. Contrast below 4.5:1 gets flagged because it costs sales, but conformance is a legal standard with an auditor attached.
The part of the number we would not defend A fair share of the 200 are one rule applied to several page types, so 200 checks is not 200 independent insights. Read the count as coverage and judge the top three findings instead. Our frequency figures come from 134+ audited businesses, all sites whose owners already suspected a problem, so treat any "percentage of sites" figure from us as skewed high.
Questions people ask about signal counts
Three come up, and they are one question in different clothes: can I trust a number I did not count myself?
Does a higher check count mean a better audit?
No, and the Lighthouse weighting is the cleanest proof. Five metrics decide the performance score while the long list beside them decides nothing. What matters is whether the tool ranks its findings and attaches a number to each.
Where does the 200 figure come from?
It is the count of distinct conditions in the audit logic across the five categories. Routing by business type changes which subset applies, so an ecommerce site and a SaaS site do not get the same 200.
Is this the same as a conversion signal in Google Ads?
Different thing entirely. In ad platforms a conversion signal is data sent back about a completed action, used to train bidding. Here it means a condition on your page. Same two words, opposite direction of travel.
For the broader map of what tends to break, the conversion leak index ranks the problems by how often they show up.