- 29Issue types we actually check
Each one states the evidence behind it, and when it would be wrong to treat as a defect.
- 9Search Console lists, ranked by impact
Four from Search Console alone, five more once the crawl can say why.
- 28 daysBefore anything is called a result
Movement before the window closes is shown, always labelled “so far”.
- 4Answer engines, asked every week
ChatGPT, Perplexity, Gemini and Google’s AI Overviews.
- $0For one site, permanently
Not a trial. A ledger takes about six weeks to return its first verdict.
Root-cause grouping
Hundreds of findings. A handful of decisions.
Every audit tool is generous with findings and silent on priority. Four hundred missing meta descriptions is one templating bug, not four hundred tickets, and a severity label invented in the abstract will not tell you which three are worth a Saturday.
Rankelle groups by issue type and URL pattern, states the confidence (400 of 400 is a template; 6 of 14 is a question), and ranks the groups by the Search Console impressions sitting behind them. The top row is the one to spend the evening on.
What to do next
It reads your Search Console, then tells you where to spend the evening.
The crawl knows what your pages contain. Search Console knows what people search and where you rank. Held apart, each gives you a list; crossed, they give you a reason — the query with real volume whose landing page never targets it, the page earning impressions that nothing links to, the URL ranking despite the noindex you forgot about.
Four lists come from Search Console alone and are ready the moment you connect it: queries stuck on page two, positions whose click-through is below what they should earn, pages losing traffic, and two of your own URLs fighting over one query. Five more arrive with the first crawl, each naming the reason and the fix.
The ledger
Then it tells you whether fixing it did anything.
You ship a title rewrite. Traffic moves a little. Was that you, or seasonality, or a core update that landed the same week? Nothing else you own can answer that, so SEO gets done on faith or quietly dropped on a hunch.
When the defect disappears from the watched pages, we freeze a set of comparable URLs you did not touch and compare the two over the following month. Improved, no effect, regressed, or inconclusive — and there is no setting to hide the ones you would rather not have seen.
AI visibility
And whether the assistants recommend you.
Someone looking for what you sell asks ChatGPT and gets a paragraph naming four products. You are in it or you are not, and until now you had no way of knowing which — or how often, or who was named instead.
Every week we ask the questions a buyer would ask. You see whether you were named, who was named instead, what the answer cited, and the answer itself, word for word. A run that failed is a dash, never a zero.
How a fix becomes a verdict
- CrawlYour public site, the way a search engine would. robots.txt obeyed, rate-limited, identified as Rankelle-Audit. Page text is hashed and discarded.
- GroupFindings collapse into root causes by issue type and URL pattern, ranked by the impressions behind the affected pages.
- Notice the shipWatched pages are re-checked. When the defect is gone from all of them, the cause is marked landed. Nobody has to tell us the developer shipped.
- Freeze a controlComparable untouched URLs, chosen at that moment and never recomputed. A moving control is not a control.
- VerdictTwenty-eight days later: worked, no effect, regressed, or inconclusive. Nothing is decided while the window is still open.
Who it is for
Founders doing their own SEO
You are the developer, the writer, and the person who has to decide whether any of this is worth another Saturday.
SEO agencies on retainer
You are judged monthly on work whose results arrive quarterly. The report that closes that gap, including the misses.
Freelance consultants
One person, several retainers, no time to hand-assemble the deck at eleven on a Sunday.
The report shows the work that failed
No-effect and regressed verdicts sit in the record beside the wins. A tool that only reports wins is a horoscope — you would learn to read it as one, and you would be right to. Take the failures out and you have removed the reason to believe the wins, including from yourself.
Free for one site, permanently. Not a trial — the first verdict takes about six weeks to arrive, and a fortnight would only show you our screenshots.
Questions
I only have one site. Is this overkill?
One site is the case it was built for. The crawl and the grouping pay off immediately, and the verdicts matter more when you are the one who spent the weekend, because nobody else is going to tell you it was wasted.
How long until the first verdict?
About six weeks after a change ships: a 28-day measurement window plus Search Console’s reporting lag. Movement before that is shown, always labelled “so far”, and never as a result.
Do you need access to my site or my repository?
No. We crawl the public site the way a search engine does and read Search Console. There is no write token anywhere in the product, so we cannot change anything even by accident.
I run an agency with twenty clients. Does it scale?
Yes — that is the other half of who uses it. Clients are the pricing unit, there is a portfolio view, and the monthly report is branded as yours. See the pages under who it is for.
Does this replace Ahrefs or Screaming Frog?
No, and we would rather say so. Research platforms find and size the work; this ranks it, watches for the deploy and measures the outcome. The overlap is the crawl, which is the cheap part of both.
Try it on your own site, free
No card, and the free tier does not expire — the first verdict takes about six weeks to arrive.