How to measure an SEO change in Google Search Console
To measure an SEO change, note the date it went live, then compare the page's daily clicks for at least 14 days before and 14 days after with the rest of the site's clicks over the same days. Leave out the days before Google recrawled the page. If the page rose clearly more than the rest of the site, the change probably worked; if the range of likely effects crosses zero, it is too early to tell. Search Console's Compare view alone misleads, because seasons and Google updates move the whole site.
Why most before and after comparisons are wrong
The usual method is to open Search Console, compare the 28 days after a change with the 28 days before, and look at the page's clicks. That mixes three things: the effect of the change, everything else that moved the whole site in those weeks, and ordinary day-to-day noise.
Seasons, holidays, news and Google's own updates move most pages on a site at once. If the whole site rose 15%, a page that rose 15% did not gain anything from its change. If the site fell, a change that helped can look like a loss. The fix is to measure the page against something that was not changed.
The method in six steps
The method compares the changed page with pages you did not change, over the same days.
- Write down the change and the day it went live. One change per page at a time, or you cannot tell which one worked.
- Pick the metric that the change should move (see the table below). Clicks for most changes; impressions for pages that get few clicks.
- Wait for Google to recrawl the page. URL Inspection shows the last crawl date; the after window starts then.
- Set the windows: at least 14 days after, and the same number of days or more before. 28 days on each side is better.
- Export the page's daily numbers and the rest of the site's for the same days, and compare their ratio before and after.
- Read the range, not the single number: a win only when the whole likely range is above zero.
Which number to measure for each kind of change
Measure the number the change is meant to move. Clicks are the goal, but some changes move impressions or position first.
| Change | Measure | Why |
|---|---|---|
| Title or meta description | Clicks and CTR on the page's main searches | A title changes whether people click, not usually where the page ranks. |
| Rewritten or expanded content | Clicks, then impressions | Better content can rank for more searches, which shows first as impressions. |
| New internal links to the page | Impressions and average position | Links help Google find and weigh the page; clicks follow later. |
| Fixing an indexing problem | Indexed status, then impressions | Until the page is indexed, no other number can move. |
| Refreshing a decaying page | Clicks against the same period last year | Decay is slow; a short window misses it. |
Getting the numbers out of Search Console
Use the page filter for both exports. Search Console counts data a little differently when it is grouped by page than for the whole property, so two page-filtered exports compare like with like.
- Performance, Search results, set the dates to cover both windows.
- Add a Page filter, Exact URL, with the page. Export and open the Dates tab: the page's daily clicks and impressions.
- Change the filter to URLs not containing the same URL and export again: the rest of the site, day by day.
- Drop the change day, the days until the recrawl, and the latest two or three days, which are often still incomplete.
Reading the result
The 10% line is a choice: below it, a change is too small to be worth the work of finding it. A page with few clicks has wide ranges, so it can only show big changes. On such pages, measure impressions or position instead, or measure a group of similar pages together.
| Result | What it means | What to do |
|---|---|---|
| Win | The whole likely range is above zero. | Keep the change, and try it on similar pages. |
| Loss | The whole likely range is below zero. | Undo it or try a different change. |
| No change | The whole range sits within 10% either way. | The change did not matter. Move on to a bigger one. |
| Not clear yet | The range crosses zero and reaches past 10%. | Wait for more days. If it is still unclear at 28 days, the change was too small to see. |
A worked example
A page averages about 20 clicks a day for the 28 days before a title change, while the rest of the site averages about 400. In the 28 days after the recrawl the page averages about 27 a day and the rest of the site about 410. The simple comparison says the page rose 40%. Against the rest of the site it rose about a third, with a likely range of about 21% to 45%: a win, but smaller than the simple comparison suggested.
These are the numbers behind the calculator's example button, so you can load them and see the same result.
Measuring AI answers is different
ChatGPT, Perplexity and Gemini answers change every time you ask, so there are no daily clicks to compare. Instead, ask the same questions several times before and after the change, count how often the answers cite you, and compare that rate with questions you did not try to change. The number of answers decides how small a change you can see.
Let RankEcho run the measurement
SEO/GEO/AEO Intel does this for every change you ship. It freezes the page's numbers when you plan the change, waits for the change to go live, compares the page with the site's unchanged pages, and reads the result at 14 and 28 days after a two-day wait for Search Console data. AI answers are re-asked on the same questions at 3, 7, 14 and 28 days. It is $29/month with a 3-day trial on one site.
Sources reviewed
Search and publication principles below were checked against primary documentation. The worked examples are fictional; these sources do not establish a traffic outcome or a guaranteed ranking.
2 claim-level source records
| Claim reviewed | Official source | Review record |
|---|---|---|
| Search Console's Performance report shows clicks, impressions, CTR and average position over time and can be filtered by page and compared across date ranges. | Google Search Console: Performance report | Checked 2026-11-02 · Google public documentation, reviewed 2 November 2026 · Defines the numbers used here. The comparison against the rest of the site is RankEcho's method, not Google's. · Confidence: High |
| URL Inspection shows information about Google's indexed version of a page, including when it was last crawled. | Google Search Console: URL Inspection tool | Checked 2026-11-02 · Google public documentation, reviewed 2 November 2026 · The crawl date shows when Google fetched the page, not when the change started to affect rankings. · Confidence: High |
Frequently asked questions
At least 14 days after Google recrawls the page, and 28 days is better. Small sites and pages with few clicks need longer windows to see anything.
As a first look, yes. But it compares the page with itself, so anything that moved the whole site, such as a season or a Google update, looks like the effect of your change.
Measure them as a group against the pages you did not change. You will learn whether the batch worked, but not which page or change did it.
Either it had no effect, or the page has too few clicks to show an effect of that size. Check the smallest change the calculator says your windows could show.