How to measure what an SEO change did

Start with a baseline

A number after a change means nothing without a number before it. The baseline is the period just before the change, long enough to smooth out the difference between a Monday and a Sunday: several weeks, not a few days.

Compare like with like. Clicks per day before against clicks per day after is a fair comparison; total clicks over a 14-day window against total clicks over a 28-day window is not. Leave the day of the change out of both periods, because half of it belongs to neither.

Pick the metric that fits the change

Clicks are the outcome you can bank: they are what reaches your site. Impressions and average position are context. They tell you whether the page was shown more or ranked differently, which helps you read a change in clicks, but they are not a result by themselves.

Match the metric to the change. A rewritten title or meta description changes how often a shown result gets clicked, so look at clicks and click-through rate for that page. A change that moves a page up a few places shows first in position and impressions, and in clicks later.

• Measure per page when you changed one page; per site only for site-wide work.
• Use the same data source before and after. Search Console for clicks and impressions, analytics for visits.
• Do not mix a branded query with a non-branded one in the same comparison.

Give the change time to show

A search engine has to fetch the changed page, process it and update what it shows, and the data in Search Console arrives a few days late. A comparison made the next day measures nothing yet.

Different changes need different windows. A title or description change can show within weeks; a change to internal links or to the content of a page often needs longer. Choose the window before you look at the result, not after, or you will be tempted to stop when the number looks good.

Look for other explanations

Clicks move for reasons that have nothing to do with your edit. Before you credit a change, check whether one of these also happened in the same period:

• Another change on the same page, by you or by a colleague.
• A seasonal pattern: a shop in December, a tax guide in March. Compare with the same weeks last year when you can.
• A move across the whole site, which points at something other than this page.
• A change in the results page itself, such as an AI answer or a new feature that takes clicks while your ranking holds.
• A search engine update that landed in the window.

What to write down

A result you can check next quarter is worth more than one you remember. For each change, record:

• The page, what changed and the date.
• The baseline period and the window after, and the metric.
• The numbers before and after, and how much data they rest on.
• Other things that happened in the window.
• Your conclusion, worded as a measurement, a likely reason, or no conclusion.