Optimization reports
The optimization reports turn Search Console data into short lists of places to act. This page maps the nine reports, gives the page-poaching gain formula and explains the decay, CTR and cannibalization settings.
✓ Checked in the workspace code · 7 October 2026
What the optimization reports are for
The optimization reports turn Search Console data into short lists of places to act: queries close to the top, pages losing traffic, pages competing with each other, snippets with a weak CTR. The hub page, Optimization Reports, links to nine of them. Each one reads the same data and the same periods as the dashboard (see dates, comparisons and data delay).
A list is a set of candidates, not a verdict. Open the page and the live results before you change anything.
The nine reports
| Report | Question it answers |
|---|---|
| Content Groups | how does a set of pages perform (groups and clusters) |
| Topic Clusters | how does a theme of queries perform |
| On-Page SEO Check | do the top queries appear in the title, description and H1 |
| Content Decay Map | which pages keep losing clicks |
| Query Decay Map | which queries keep losing clicks |
| Keyword Cannibalization | which queries are split across several of your URLs |
| Page Poaching Opportunities | which queries are just outside the top and worth lifting |
| CTR Benchmark | which queries get fewer clicks than their position suggests |
| Long-Tail URL Clustering | how is traffic split between head, middle and long tail |
Page Poaching: queries just outside the top
The report lists queries by impressions with their current position and the page recommended for each. The mode selector narrows the range: All (positions 4–20), Top-10, position 10–15 or position 10–20. A target position, 3 by default, sets the goal.
Estimated gain = max(0, impressions × expected CTR at the target ÷ 100 − current clicks)
The expected CTR comes from the property’s own CTR curve, not from a public table, so the estimate reflects your site. It is the number of clicks implied if the query reached the target, not a forecast. A second table lists CTR opportunities: queries that already reached the target but whose CTR is below the expected one, where the snippet is the thing to test. Counts are based on the rows Search Console returned, so the long tail may be cut (dimensions).
CTR Benchmark
Press “Run analysis” after choosing filters. The table lists each query with its position, CTR, expected CTR, the difference against expected, clicks and impressions. A chart shows the CTR of your filtered set by position against the site average. Reports can be saved and shared by link. The aim is to find queries where the title or description fails to earn clicks; the playbook on improving organic CTR shows how to test a change.
Content Decay and Query Decay
Both reports compare two equal periods and separate a drop on one page or query from the overall movement of the site. The column “over trend” shows the loss beyond the site’s own trend, which is the part a content refresh can plausibly win back.
- Content Decay: periods of 3, 6, 12 or 16 months; a minimum number of impressions; a threshold (4 percent by default); filters by URL text or content group; metrics clicks, impressions, average position and CTR; views by month or week and by path or full URL; a heatmap you load with “Load heatmap”.
- Query Decay: periods from 7 days to 12 months, a minimum number of impressions and a query filter, loaded with “Load data”.
See also how to find and fix content decay.
Keyword Cannibalization
- Window: 7, 14 or 31 days.
- Brand terms: exclude branded queries, where several results are normal; the property’s brand terms can be reused.
- Minimum share: a URL must have at least the chosen share (5 to 25 percent) of the leading URL’s impressions to count as competing.
- Date file: an optional CSV or TXT with URL, datePublished and dateModified to show publication dates beside the URLs.
- Folder filter: limit the analysis to one section.
Two URLs for one query do not prove a problem. Read how to find and fix cannibalization before merging pages.
How to work through a list
- Pick one report for one question. Do not open all nine at once.
- Set a minimum impressions floor so tiny rows do not dominate.
- Open the page and the live result for the top three rows.
- Change one thing, annotate it and compare equal periods later.
Limits to keep in mind
- Estimates, not promises. A gain is the clicks implied by an expected CTR.
- Capped exports. Search Console returns the most important rows.
- Preliminary days. The newest days can change.
If something looks wrong
| Symptom | Cause and fix |
|---|---|
| “Report not loaded yet” | press Run analysis, Load heatmap or Load data after setting filters |
| “Could not load data” | use Retry, or connect Search Console again if your session expired |
| Explanations are in Russian | the help blocks of several reports are Russian-only for now; the data is not affected |
Optimization reports, answered
How is the page-poaching gain calculated?
Impressions times the expected CTR at the target position, minus current clicks, never below zero.
What does “over trend” mean?
The loss beyond the site’s overall movement over the same periods.
Where does the expected CTR come from?
From the property’s own CTR curve.
Continue
Next, see rankings and rank changes or the playbook on striking-distance keywords.