Short answer: for SEO, the Performance report is where I start. It shows which searches find you, which pages they land on, and the day things changed. Read it in four passes: totals, searches, pages, then the timeline. Each pass should end in a ticket, not a chart. That's what an SEO report is for.
Google Search Console is Google's free tool for site owners. It shows how your site appears in Google Search. The Performance report is the part that counts clicks and impressions.
What information can I find in the Google Search Console Performance report?
Four numbers, split in different ways. Google's help defines them:
- Clicks: how many times someone clicked your site from Google Search results.
- Impressions: how many times your site appeared in Search results.
- CTR: click-through rate, clicks divided by impressions.
- Average position: the average position of your topmost result.
You can split them by Queries (the searches), Pages, Countries, Devices, Search appearance and Dates. The export gives you each of those as its own sheet.
How accurate is Google Search Console?
It's Google's own data, so it's the closest you'll get to what Google saw. But it has limits, and Google lists them in its data discrepancies help:
- The table stops at 1,000 rows. Rare and long-tail rows can be left out of the table but still count in the chart totals.
- Some searches are hidden. Rare (anonymised) queries are left out of the Queries table for privacy, but they're in the totals. So the Queries rows won't add up to the total.
- The newest data lags. Google says collected data is usually available in 2 to 3 days. The newest data can also be preliminary, shown as a dotted line, and may still change.
- It won't match Google Analytics. The two count differently, so don't expect the numbers to line up.
And three more I always keep in mind:
- There was a reporting error. Google's data anomalies page says impressions weren't reported accurately from 13 May 2025 until 27 April 2026. Impressions, CTR and average position were affected. Clicks weren't. If your comparison spans that window, trust the clicks.
- Position is an average. It's the average of your topmost result across every time you showed. A page at "8.4" may have been 3 for some searches and 20 for others.
- Clicks aren't sales. Search Console stops at the click. For what happened next, you need your analytics.
How to use Google Search Console for SEO, by hand
Open the Performance report and use its export button, which downloads both the chart and the table data. Pick a range like the last 3 months, compared with the 3 before. Then go through it in this order.
- Totals. Clicks and impressions, now against then. Did clicks fall because impressions fell (fewer people saw you), or because CTR fell (they saw you and didn't click)? Those are different problems.
- Searches (Queries). Split brand searches from the rest. Brand demand is people looking for you. Non-brand is where SEO work shows. Look for searches with lots of impressions and almost no clicks. Google is showing you and people are choosing someone else. That's usually the title, the snippet or the wrong page for the search.
- Pages. Which pages earn most of the clicks? Those are the money pages. Check them first, before you change anything. Then look for groups of pages that get impressions but almost no clicks. Many on the same path can point to thin or near-duplicate pages.
- Timeline (Dates). Look for one day where clicks dropped sharply, not a slow slide. A sharp drop on one day usually means something changed on the site or in Google that day. A slow slide is more often demand or competition.
- Write the tickets. One problem, one owner, one check. "Clicks fell" isn't a ticket. "These 12 searches show us but nobody clicks, rewrite these 4 titles" is.
What good looks like: you can say in one line why clicks moved, and every finding has a next check. What bad looks like: a report full of charts and no decision.
What is in an SEO report?
For me, an SEO report answers three questions, in this order:
- What changed? Clicks, impressions and CTR against the last period, with the date it changed.
- Why, as far as the data shows? The searches and pages behind the change. Be honest about what the data can't show.
- What do we do next? A short list of fixes, in order, each with a way to check it worked.
Everything else is supporting detail. A report that's mostly screenshots of charts usually skips the third question.
How to view the Search generative AI performance reports
Google announced Search generative AI performance reports in Search Console on 3 June 2026, with separate reports for Search and Discover. They show your impressions inside generative AI features such as AI Overviews and AI Mode. Google's post says that as of 31 August 2026 they're rolled out to all websites worldwide. That data is also still counted in the main Performance report. I haven't confirmed the exact menu path, so check Google's post for the latest. STORY reads the standard Performance export, not these reports.
How to check Google performance for one page
Filter the Performance report to the page (Page, then the URL), and look at the Queries for that page. Then use the URL Inspection tool to check Google can still crawl and index it. Performance tells you what happened. Inspection tells you whether the page is healthy now.
How STORY makes it quicker
STORY is the first mode in the GSC companion. It reads a Performance export and does the four passes for you, then writes the tickets.
STORY sits inside the GSC companion, which asks for an email sign-in (a link sent to your inbox, no password). It doesn't connect to your Search Console account. You bring the export.
What you give it: the Performance export as .xlsx, .xls or .csv. The multi-sheet export works as it is (Chart, Queries, Pages, Devices, Countries). You can also paste a sheet. If a manual action or thin-content review is going on, tick that box to add the Appeal readiness chapter.
How to read the result, chapter by chapter:
- Headline: totals for clicks, impressions, CTR and average position.
- Demand: the searches. Total clicks and impressions, and how much the top 5 searches take. Searches with 100 or more impressions and a CTR under 1% are flagged as weak.
- Pages: the same for pages, with a short note on each page, such as "money page - protect it before changing anything" or "high impressions / low CTR - inspect intent before rewriting".
- Cliff / timeline: the biggest day-on-day click drop of 20% or more, with average position before and after. It needs at least 4 dated rows. With 14 days or more it also compares the last 7 days with the 7 before.
- Fix board: up to 7 tickets, P0 to P2 (P0 means do it first). For example "Explain the [date] cliff before anything else", "Rewrite titles for N visible-but-unclicked queries", "Live-check your top page before touching it" and "N pages collect impressions and almost no clicks". Each ticket opens the right mode, such as DRIFT, THINN, LOCRA or KNOT. If nothing is urgent, it says so: "Nothing in this export demands a P0", then "Bank this run as a baseline".
- Appeal readiness: only when you ticked the box. A readiness order, starting with "Contain the junk first" and ending with "Then verify live". It's an order of work, not an appeal. STORY doesn't submit anything and can't see your Search Console messages.
Example (made-up numbers)
A site exports 3 months of Performance data, all sheets.
- Headline: 21,000 clicks, 640,000 impressions, 3.3% CTR.
- Cliff / timeline: clicks fell from 310 to 190 on one day, a 39% drop. Average position barely moved.
- Demand: 14 non-brand searches have 100 or more impressions and under 1% CTR.
- Fix board: P0 "Explain the cliff before anything else", which opens DRIFT. P1 "Rewrite titles for 14 visible-but-unclicked queries", which opens THINN.
So the first job is to find what changed that day, with redirects, robots.txt and templates checked first. The title rewrites can wait a week.
What to do next
- Work the Fix board from the top. Each ticket opens the mode that checks it.
- Save the export and re-run STORY next month on the same range length, so you're comparing like with like.
- For the CTR side, build your own CTR curve.
- Not sure where the drop came from? Read how to check for a Google penalty.
- For the rest of the companion, see GSC companion: how it works.