Questions we get asked
Short answers to the questions that reach us most: page counts, firewalls, crawl load, why the score differs from another tool, where the AI gets its answers, what happens at a limit, and where to write.
The reading found far fewer pages than my site has. Why?
In order of likelihood: a firewall answered our reader with "forbidden", and the overview says so with the count and the addresses to allow. robots.txt keeps the reader out of part of the site, which the reading honours. The plan’s page cap was reached. Or the default filters skipped login, cart, sort and filter URLs on purpose, and filter URLs were left unread once their shape was known, which the coverage notes on the overview spell out.
Will the reading slow my site down?
The reader fetches one page at a time per site, reads robots.txt first and honours it, and slows down on its own when the server pushes back or starts answering with errors. It identifies as an ordinary browser, so if you allow it through a firewall, allow the addresses listed on the documentation page rather than a user agent.
Why is my score different from another tool’s?
Because it weighs different things. The score is built from what is costing traffic, the leaks, on the pages that are indexable and canonical, and technical problems are counted once per template rather than once per page. A tool that counts every row of every page will give a different number and a longer list; the two are not measuring the same thing.
Where does the AI get its answers, and does it invent numbers?
The chat, the written briefing and the drafted fixes all read from your own reading, not from general knowledge about SEO. Drafts are checked by code before you see them, and a number in a draft must appear on the page itself. In SEOLift, clicks, positions, forecasts and confidence are arithmetic over your Search Console data; a model only ever drafts text, and code checks it.
The written briefing, the chat messages and the drafted fixes are metered per plan. Copy prompt for your agent is not: it is built without a model and works on every plan.
I fixed something. Why is it still on the list?
The list is the latest reading, and a fix made since then has not been read yet. Press Mark fixed on the card: a live check runs on the affected pages straight away, and the next reading confirms it for good. If the state is deliberate, a noindex or a robots.txt block you meant, use This is intended instead, and the next reading stops raising it.
Do you store my server logs or my visitors?
No. Log Intelligence discards human traffic while the file is being parsed and writes only verified crawler activity, so there is no visitor IP address in our database. The public Citability scan fetches the address you give it and its robots.txt once each and keeps neither.
Can I read a site I do not own?
Any public site can be read, and the reading respects its robots.txt. Search Console data, SEOLift and the Slack note need the site to be yours, because they need its property or its channel. If a site blocks the reader, the owner’s firewall is the place to fix that, not ours.
What happens when I hit a limit?
The sidebar meters show each allowance and how much is used, and reset each billing period. When one is full, the panel says which and what the next plan changes. A reading past the allowance waits rather than being lost, and nothing already read is removed. Upgrading takes effect at once.
How do I delete a site or my account?
Delete site is on the site overview. The slot it used stays in use until the end of the billing cycle, which the sites list states. Readings are kept for the plan’s retention period and then removed. For the account itself, or anything this page does not answer, write to hello@silentstork.com.