Finding keywords with Google Search Console

Google Search Console is the most reliable free source of keywords there is, because it doesn't guess anything: it shows the queries that have already generated impressions or clicks toward your site. Unlike a suggestion tool, you're working from real behavioral data, not hypotheses.

Why start with Search Console rather than a third-party tool

Paid keyword research tools (Semrush, Ahrefs, and the like) estimate search volumes from panels and statistical models. Search Console gives you the exact queries typed by real visitors who saw or clicked on one of your pages. The information is free, exhaustive for your own domain, and directly actionable: every row in the report corresponds to an opportunity that's already started, not a guess. No sign-up beyond verifying your site is required, which makes it the natural first stop before spending on anything else.

The limitation is symmetrical: Search Console only shows queries for which you already appear, even weakly. It's ideal for developing an existing topic, less so for discovering a theme completely absent from your site — for that, free tools like Autocomplete or Google Trends take over.

Getting to the right report

In Search Console, open the “Performance” tab, then “Search results”. By default the tool shows clicks from the last three months: widen the range to 16 months (the maximum available) to capture seasonal variation. Then check the “Queries” box in the table below: that's the raw list of everything people typed to land — or almost land — on your site.

The “high impressions, low clicks” method

Sort the table by descending impressions, then look for queries that combine a high number of impressions with few clicks, at an average position between 5 and 15. These are topics Google already considers you relevant for, but not relevant enough to convince people to click, or not yet ranking high enough to be seen.

Two levers apply here: improve the existing content (structure, a more direct answer in the first lines, examples), or create a dedicated page if the topic deserves better than a paragraph buried inside a broader article.

A concrete example: a query with 4,000 monthly impressions, 40 clicks, and an average position of 8 has a click-through rate of about 1% — well below the average expected at that position (often 3 to 5%). That's a clear signal the title or meta description isn't compelling enough, or that the page doesn't fully answer the query once clicked.

Filtering by page to consolidate what already works

The other useful view: filter by URL instead of by query. Pick a page you consider strategic and look at every query that surfaces it. You'll often find that a page attracts traffic on phrasings you hadn't anticipated when writing it — a strong signal to enrich that page rather than create a new one that would duplicate it. This filter is also the right place to check that none of your strategic pages is missing from results for queries that should logically surface it — a possible sign the content isn't well aligned with visitors' real intent.

Refining with additional filters

Search Console also offers filters by country, device, and search type. If you're targeting an English-speaking market alongside a French one, compare queries by country: the wording, and sometimes the topics themselves, often differ noticeably. The device filter helps spot queries that perform well on mobile but not on desktop — a sign that the expected answer format (quick, consumable on the go) differs from what your current page offers. The search type filter (Web, Image, Video) is worth a separate look too: a topic that draws meaningful Image or Video traffic signals a format preference that a text-only page cannot fully satisfy on its own.

Grouping queries into topics, not isolated keywords

A common mistake is treating every Search Console query as a potential page. In practice, dozens of different phrasings (“how to find keywords”, “keyword research method”, “free keyword finder”) cover the same intent and should feed a single piece of content, not three pages that would cannibalize each other. Export the queries, sort them by theme in a spreadsheet, and keep only one topic per cluster — this logic of grouping by entity rather than by isolated keyword is developed in the lesson on thinking in entities.

Exporting the data for long-term tracking

Search Console only retains a rolling 16 months of data: beyond that, the history disappears. To keep a longer record and compare year over year, export the reports regularly (via the export button or a connection to Google Sheets) and archive them in a dated spreadsheet. This simple habit makes it possible, after two or three years, to spot topics whose interest is structurally growing and those that are stagnating — information no free tool provides directly.

Turning the data into a content plan

Once the groupings are done, sort them by intent (see the lesson on search intent) and by impression volume. High-volume topics with a clear intent become candidates for a pillar page; narrower variants feed spoke pages that enrich the internal linking around that pillar.

This work is best redone every quarter: queries evolve, new phrasings appear, and Search Console remains the best free compass for tracking that movement over time. A recurring calendar reminder is usually enough to keep the habit alive once the first pass is done.