Measurement
A calm way to measure whether a website change worked
Record the baseline, name the expected signal, allow the system time to respond, and resist attributing every change in inquiries to one edit.
A title was rewritten. A specialty page went live. A Business Profile category was corrected. The next question is reasonable: did it work?
The answer is rarely available the next morning, and it is rarely captured by one metric. Search systems need time to recrawl pages. AI assistants may reflect sources on a different schedule. Inquiry volume can change because of seasonality, availability, referral activity, insurance, pricing, or follow-up. Good measurement begins by deciding what the change was expected to influence.
Write down the baseline before editing
Capture the current page title, relevant rankings, map presence, AI responses, cited sources, directory details, and aggregate inquiry-path signals that are available. A screenshot and date are often more useful than a vague memory that the practice was 'not showing up.'
Do not collect client details or protected health information for marketing measurement. Public visibility and aggregate website behavior are enough for this kind of review.
Name the expected signal
A change should have a plausible target. Correcting a Business Profile category might influence local relevance. Rewriting the homepage may improve comprehension and the path to a service page. Adding a specialty page may create a new owned result for a specific search over time.
If the expected signal cannot be named, the change may be too broad to evaluate.
Use different review windows for different systems
A broken link can be verified immediately. A form change can be tested the same day. Search movement may require several crawls. AI citations may lag behind the page that was updated. Treating every signal as instant encourages unnecessary reversals.
Record the launch date and decide in advance when you will review. Continue watching during the interval, but do not rewrite the strategy every time one sampled answer changes.
Separate what changed from why it changed
You can observe that a page moved from one position to another, that an AI assistant began citing the website, or that more visitors reached the scheduler. You usually cannot prove that a single edit caused every downstream business outcome.
Report the observation first. Then label the causal explanation as an inference, especially when several pages, campaigns, referral relationships, or availability details changed during the same period.
A useful report can say, 'the ranking moved after the title change.' It should be more careful before saying, 'the title change created three new clients.'
Choose the next smallest test
If the signal moved in the expected direction, preserve the change and continue monitoring. If nothing moved, inspect whether the page was indexed, the query was realistic, the local information was consistent, and the review window was long enough.
Then choose one next fix. A sequence of documented changes teaches the practice more than a large batch that makes cause and effect impossible to discuss.
Earshot keeps last week's recommendation beside this week's public evidence, so the practice can see what changed without being handed a new twelve-item task list. One fix, one review window, one next decision.