To tell if an SEO change worked, write down before you make it what you expect to improve, by how much, and how you'll measure it; then compare the changed pages against a fair baseline, such as their own trend and similar pages you didn't touch, over enough time for Google to recrawl them. If the result is smaller than the noise, or something else changed at the same time, the honest answer is "can't tell", and that's a real result, not a failure.
"Traffic went up after we changed the titles" is how most SEO results get reported. It's also how most of them get misread. Traffic moves for many reasons at once: seasons, Google updates, a newsletter, a competitor's outage. This guide gives you a method a small team can run in a spreadsheet, the same one Dottpace uses, without statistics jargon.
Why isn't "traffic went up" proof?
Because the change you made is rarely the only thing that changed. Common look-alikes:
- Seasons and calendars. A store's traffic rises in November whatever its titles say. B2B traffic dips in late December.
- Google updates. Core updates roll out over weeks and lift or drop whole groups of sites.
- Other changes on your site. A new menu, a faster theme, a price change, a migration.
- Outside events. Press coverage, an ad campaign, a viral post, a competitor's site going down.
- How results look. An AI Overview appearing on a query can cut clicks while your ranking stays the same.
Google itself warns that some changes take effect in hours and others take several months. If you check once, a week later, you're mostly measuring noise.
What should you write down before you ship?
Write these five things in one dated note before the change goes live. Writing them first is the whole trick: it stops you from picking the flattering number afterwards.
- The change: exactly what, on which pages, on what date.
- The expected effect: which number should move, and in which direction. For example, "search clicks to these 4 category pages".
- The smallest result worth keeping: the gain below which you'd call it no win. For example, "+8%".
- The comparison: what you'll compare against (next section).
- The window: how long you'll wait before judging, and from when.
What should you compare against?
Pick the fairest comparison you have. In order of strength:
- Similar pages you didn't change. Change half of your 20 product category titles and leave the other half. If both groups rise equally, the rise wasn't your titles. This is the closest a small site gets to a controlled test.
- The same pages' own trend. Compare the window after the change with the same-length window before it, adjusted for the season (for example, the same weeks last year, if you have the data).
- The whole site. If the changed pages grew 20% while the site grew 18%, the change explains about 2 points at most.
Measure the right number, too. A title change should move click-through rate and clicks at similar positions; a speed fix should move conversions on phones; an internal-linking fix should move impressions and positions of the linked pages. Search Console's Performance report gives clicks, impressions, click-through rate and position by page and query.
How long should you wait?
- Until Google has re-read the pages. Use URL Inspection to see the last crawl date. The clock starts after that, not on the day you shipped.
- Then at least four weeks for most on-page changes, so each weekday appears several times and one odd day can't decide the result.
- Longer for small numbers. A page with 30 clicks a month needs months to show a 10% change with any confidence. Group similar pages together to get enough data.
Decide the window in advance and stick to it. Looking every day and stopping when it looks good is the most common way to fool yourself.
How do you call the result: worked, didn't work or can't tell?
When the window ends, compare the result with what you wrote down:
- Worked: the number moved in the expected direction, by at least the smallest result worth keeping, more than the comparison moved, and nothing else that changed explains it.
- Didn't work: no meaningful move, or a move in the wrong direction, with a clean comparison.
- Can't tell: the move is within normal week-to-week swings, or a Google update, a campaign or another change landed in the same window.
"Can't tell" isn't a failure. It's information: either the effect is too small to matter, or you need a cleaner test. Reporting it honestly is what makes your "worked" results believable to a client or a boss. Note also that applied isn't worked: a fix that went live and was checked only tells you the change happened.
What does a worked example look like?
This example is illustrative; it's the same one we use on our how we measure page.
The fix: a site redesign removed the links to four category pages; the fix restores them. Search clicks to those pages had dropped about a fifth after the redesign.
Written down before the fix: search clicks to the four pages over the 28 days after Google re-fetches them, against the pace before the fix, season-adjusted. It counts as Worked only from +8%.
Result, 28 days later: clicks up 31%, past the +8% threshold, and no Google update or other change in the window. Verdict: Worked.
If instead clicks had risen 5%, or a core update had landed in the same weeks, the verdict would have been Can't tell, and said so.
As dottpace.com itself gets Search Console data, we'll add our own dated before-and-after examples to this page.
What if traffic is dropping instead?
Run the same method backwards. List everything that changed in the days before the drop: your releases, plugin updates, Google updates, competitor moves, AI Overviews appearing on your queries. Then check which pages and queries fell. A drop limited to a few pages after a template change points to the template; a broad drop on the day of a Google update points to the update. Our guide on why a website stops showing on Google covers the technical causes to rule out first.
What tools do you need?
- Google Search Console for clicks, impressions, click-through rate and position per page and query.
- Your analytics (such as Google Analytics) for what visitors did after they arrived.
- A change log: a spreadsheet with the date, change, pages, expected effect, threshold, comparison and window. This is the part most teams skip.
Dottpace Changes and verdicts keeps that log for you, records the expected effect and method when you approve a fix, and returns the verdict when the window ends. It's built for freelancers and consultants who have to show clients what their work did.
Sources
- Google, SEO Starter Guide (how long changes take).
- Google, URL Inspection tool.