Keyword tracking follows where your pages sit in the search results for the terms you care about. Positions refresh once a day, per country and per device, and every change keeps its history.
This is the oldest measurement in SEO and it still works, because people still click results. What changed is that it is no longer the whole picture. See rank tracking against AI visibility for why you need both.
What to track
The terms you actually compete for, not the terms you would like to own.
A keyword you rank 80th for is not a measurement, it is a wish. It will sit at 80 for months and tell you nothing. Track what you rank for now plus a realistic set you are working toward. Performance shows the queries you already get impressions for, which is where the next terms worth tracking usually come from.
Skip your brand name. You rank first for it. It will move when something is badly wrong and never otherwise, which makes it a poor early-warning signal and a permanent flattering line on your chart.
Adding keywords
Add keywords by hand with Add keywords: one per line, with the country and the devices to check. The first check runs in the next daily update, and the keyword's history builds from there.
Each keyword is tracked once per country and device. Adding one you already follow changes nothing. When you publish an article, its main keyword is added to tracking on its own.
Country and device
Both matter more than people expect.
Results differ between countries even for the same language, and mobile and desktop diverge on exactly the queries where the commercial stakes are highest. If your customers are in one country and on phones, tracking desktop results in another country produces a number that is real and irrelevant.
Reading a move
A single day's change is usually noise. Results fluctuate for reasons that have nothing to do with you.
What is worth reacting to:
- A sustained drop over several days on a term that was stable
- Several related terms moving together, which usually means a page rather than a keyword
- A drop that coincides with something you shipped
For the third case, Logs in your organization settings records when articles were published or deleted, which is what turns "rankings fell in March" into "rankings fell the day we published that rewrite".
When positions and AI visibility disagree
Ranking well while absent from generated answers is common and is not a contradiction. The two measure different things. AI visibility covers what to do about it.
Where it connects
Positions sit next to Performance, which shows the impressions and clicks those positions produced, and the pair is more useful than either alone: a position that improved without impressions following is a sign the term has less demand than you thought.
Need help?
Write to us and a person answers. See the frequently asked questions first if your question is a common one, or get in touch for anything else.








