Nothing here is a black box.

Two sources of evidence

SearchDone looks at two things: what Google tells us about your site (which searches showed your pages, where they ranked, who clicked) and what your live pages actually contain (titles, descriptions, links, images, tags). The first tells us what is worth fixing; the second tells us what is wrong and lets us fix it.

Rows below 50 impressions — fewer than 50 people seeing your page for a search — are discarded before anything else happens. Below that, click rates are statistically meaningless and a single click swings them wildly.

What a position is worth

Two checks need to know how many clicks a Google position normally earns. SearchDone uses a blended desktop-and-mobile baseline, filling in between whole positions because an average position is rarely a whole number:

This is deliberately a rule of thumb, not a claim of precision. Its job is to rank fixes against each other consistently, not to predict exact visits. Your real curve depends on your industry, your brand and what else is on the results page — so treat the estimates as ordering, not forecasts.

So close to page one

Flags a page whose average position is between 3 and 20 for a search with real volume. The value shown is the extra clicks the page would earn if it reached position 3, using the curve above against its current impressions.

Position three rather than one is deliberate. Modelling every opportunity as if it will reach the top spot produces flattering numbers and a badly ordered list.

Nobody's clicking

Flags a page ranking at position 10 or better whose actual click rate is at least 50% below what the curve expects for that position. The value is the clicks recovered by reaching the expected rate.

These are usually the best findings on a site, because the ranking is already earned — the listing is simply failing to convert it. A new title and description is cheap, reversible, and something SearchDone can write and apply for you.

Competing with yourself

Flags a search where two or more of your pages each hold at least 10% of that search's impressions. The floor matters: without it, every search with an incidental second page would be reported as a conflict.

The finding lists each competing page with its clicks, impressions and position, so the decision about which page should win is made on evidence rather than instinct.

Quietly slipping

Splits the period in half and compares them. Flags a page whose clicks fell by at least 30% from the first half to the second, provided it had at least 10 clicks in the first half so that very small pages cannot dominate the list. This check also runs on its own every week across your whole site and emails you when a new page starts to slip.

Detection is not diagnosis. SearchDone tells you a page is declining and by how much; establishing why — Google changed, a competitor improved, the page got old — is a human judgement, and the rewrite is yours to approve.

Checks on your live pages

SearchDone fetches a bounded set of your pages — the ones earning impressions, your homepage, and your sitemap — and reads what is actually there. It follows no links beyond that set, so the cost is known before a scan starts. Pages Google is told not to index, and pages that don't load, are excluded. On the pages it reads, it checks for:

• Titles and descriptions that are missing, duplicated across pages, far too long, or that ignore the search the page actually ranks for. Severity scales with how many people see the page.
• Address tags (the canonical tag) that are missing or point somewhere other than the page itself — a small mistake that can hide a page from Google.
• Missing main headings.
• Images with no description, one finding per page with examples — decorative images and images already described are correctly left alone.
• Broken links to pages on your own site that no longer exist, found by actually fetching the link targets.
• Orphan pages nothing else on your site links to, judged within the pages we read.
• Structured data that is missing or broken — the behind-the-scenes code that tells Google what a page is about.

Ranking and severity

Findings from your Google data are pooled and sorted by estimated click value, highest first. Severity comes from the same number: high at 100 or more estimated clicks, medium at 25 or more, otherwise low. Findings on your live pages carry a severity that scales with the page's impressions.

What the AI does and doesn't decide

Whether a finding exists is determined entirely by the rules above — running the same data through twice gives the same answer both times. AI is used to explain findings, to answer your questions from figures already computed, and to write drafts: titles, descriptions, image descriptions and page rewrites. Every draft is checked against its sources and reviewed for quality before you see it, and nothing an AI wrote reaches your site without your approval. See how the writing works.