Score and History
The findings list is a snapshot. It does not answer the question that months of work are measured by: is it getting better? For that, every audited listing gets an SEO Score and a state row, and every run over several listings leaves a line in the history.
How to get there: Via the SEO Score figure on the dashboard (opens the full runs), via the Listings audited figure and the Listing health card there, via the Audited Listings and Audit Runs actions on the dashboard, or via Tell me Audited Listings or SEO Audit Runs.
The scale
Every listing starts at 100 points. Every finding deducts, depending on its severity:
| Severity | Deduction per finding |
|---|---|
| Error | 25 points |
| Warning | 10 points |
| Info | 3 points |
The score never falls below zero. A listing without findings scores 100, a listing with one error and two warnings 55, an empty listing with five errors 0 — where it belongs. Which rule has which severity is listed in the rule set.
The colour of the score is the same everywhere: green from 80, amber from 50, red below. A listing does not change colour on its way from the dashboard into the list.
The weights are deliberately not a setting
Everything else about the audit is configurable, the score is not. It exists to be compared with last month's score — a scale that moves with a setting makes every such comparison worthless: you would never know whether the curve rose or only the yardstick. Whoever changes the rule limits changes what counts as a finding — that is honest, the rule changed. The exchange rate between a warning and an error stays put.
What a finding costs
The SEO Audit Finding card — click the Severity in the findings list or in the SEO Findings FactBox — shows three figures under What it costs:
| Field | Meaning |
|---|---|
| Weight of this severity | What one finding of this kind deducts: 25, 10 or 3 points |
| Score of this listing | What the last audit gave |
| Score once this is fixed | What the listing would score with this one finding gone and the others standing |
Two scores rather than the weight alone, because the scale stops at 0. A listing with five errors is at 0, and fixing one of them leaves it at 0 — "25 points" would be a promise the scale cannot keep there. When both figures are equal, the card says so in one sentence: the errors are what brings the listing off the floor.
The state row
A listing without findings gets its row too, in Audited Listings. It is the only thing that tells "audited and clean" from "never looked at": an empty SEO Findings FactBox on the marketplace item card can mean either, a state row with a score of 100 means only one. The coverage figure on the dashboard rests on it.
The row is rewritten by every audit of the listing — in the scheduled run, in a run over a filtered list, with Audit SEO on the marketplace item card, with Audit This Listing Again, and after applying a content proposal, which audits the listing again on the spot.
The list is sorted by score, worst first — it is the work queue you take from the top.
| Column | Meaning |
|---|---|
| Marketplace Code / Sub-Account Code / Item No. | The listing. Variant Code is an additional column you can show |
| Item Description / Title | Item and listing title, for recognition |
| SEO Score | 0 to 100, coloured as above |
| Errors / Warnings / Hints | The number of findings per severity. A click opens exactly those findings |
| Findings | All findings together |
| Last Audited At | When the listing last went through an audit |
| Waiting for Automatic Draft | Whether the scheduled draft run has the listing in its queue (see Copilot Drafts) |
| Action | Effect |
|---|---|
| Audit Again | Audits the selected listing right away and replaces its findings |
| Draft Content | Asks Copilot for new texts; the draft becomes a content proposal |
| Release for Redraft | Puts the selected listings back into the automatic draft queue |
| Findings | Opens every finding of the selected listing |
| Marketplace Item | Opens the marketplace item card |
Listings of blocked items
When the item or item variant of a listing is blocked or blocked for sales, the engine does not audit the listing. Its row and its findings are removed by the next scheduled run, by a full run started by hand, or right away by Audit This Listing Again.
Where you see the score
| Place | What it shows |
|---|---|
| Dashboard · SEO Score figure | The average of all listings with a state row, the sub line with the direction, and the curve of the full runs. A dash instead of a zero while nothing has been audited. A click opens the full runs |
| Dashboard · Listings audited figure | How many listings have a state row, out of how many in scope — the coverage. The sub line names how many are still waiting for their first run |
| Dashboard · Listing health card | Clean (no finding), Hints only, With warnings (no error), Blocked (at least one error) — and below, how many listings in scope have never been audited |
| Audited Listings | Score and counts per listing, worst first |
| SEO Audit Finding card | The listing's score and its score once this one finding is fixed |
| SEO Audit Runs | The average score per run, the history |
"0 errors" on the dashboard means either "clean" or "never looked at". The coverage next to it says which.
Average and curve
The SEO Score figure shows the average of all listings that have a state row — as it is now. Because applying a proposal and auditing a single listing rewrite that listing's state row, the figure moves between two runs as well.
The sub line therefore compares like this:
- When the average has moved since the last full run (by at least 0.1 points), it compares today's state with that run: "Up x points since the last run" or "Down x points since the last run".
- Right after a run, when both are the same, it compares the last full run with the one before: "Up x points since the run before", "Down …" or "Unchanged since the run before".
- As long as there is nothing to compare: "No second full run to compare with yet".
The curve next to it shows the averages of the last full runs, oldest first — and, once today's state has moved away from the last run, today's state as its last point. The curve thus ends where the figure stands.
Only full runs count for the curve
A run over eleven filtered listings and the nightly run over the whole catalogue must never share a curve — the filtered one would look like a collapse the next morning although nobody made anything worse. Every run is recorded; only the full ones go into the comparison and the curve.
The history
SEO Audit Runs — the dashboard action to it is called Audit Runs — keeps one line per run over several listings, newest first. Read the Average Score column downwards to see whether the catalogue is improving.
| Column | Meaning |
|---|---|
| Run At | When the run started |
| Average Score | The average of the listings this run checked — not of the whole catalogue |
| Full Run | Whether the run checked the whole scope. Only these runs are comparable with each other |
| Listings Audited / Clean Listings | How many listings the run checked and how many of them were clean |
| Errors / Warnings / Hints | The run's findings per severity |
| Duration (ms) / Started By | Run time and user |
Show Full Runs Only and Show All Runs switch the view. The SEO Score figure on the dashboard opens the list filtered to the full runs — the runs the comparison and the curve are read from; the Audit Runs action opens it with all runs.
| Run | Full? |
|---|---|
| Scheduled run over the whole catalogue, including the day set in Full Run On | yes |
| Run SEO Audit on the dashboard or in the findings list | yes |
| Scheduled run with Scheduled Run Checks Changes Only | no |
| Run SEO Audit in the filtered Marketplace Items list | no |
| Audit SEO, Audit This Listing Again, Audit Again, applying a proposal | writes no line into the history — only the listing's state row |
A run that audited not a single listing leaves no line.
See also
- Findings — the rule set and the severities
- Dashboard
- SEO Engine Setup — rule limits, scheduled run, full run
- Content Proposals