In short
Free keyword research works by combining what you already rank for (Search Console), what people actually type (autocomplete, People Also Ask), and what's already winning (competitor titles and headings) — the same raw signals paid tools repackage, just gathered by hand.
Start with what you already rank for
Google Search Console's Performance report shows every query your site already gets impressions for, including ones you never deliberately targeted. Sort by impressions with a low click-through rate — that's a list of queries Google already associates with your content, where a better title, heading, or expanded section could win more clicks without writing anything new. This is the single highest-signal, completely free keyword source most sites never check.
Mine autocomplete for real phrasing
Type your seed topic into Google's search box and read the autocomplete suggestions — they're generated from actual query volume, not a tool's estimate. Do this with the seed term alone, then with "how," "why," "best," and "vs" prepended or appended. Five minutes of this produces a more accurate picture of real phrasing than most keyword tools' "suggested keywords" feature.
Read People Also Ask, all the way down
Search your target query and open every question in the "People Also Ask" box — each one you expand reveals more related questions. This is Google directly telling you what real searchers ask next, in their own words, which makes it a ready-made outline for FAQ sections and internal topic coverage.
Check what's already ranking
Search your target keyword and read the exact titles and H1s of the top five results. Note the pattern — are they all "complete guides"? All comparison posts? All short definitions? This tells you the format Google currently rewards for that query, and whether there's a gap (a format nobody's done well yet) worth filling.
Use your own site search and support inbox
If your site has internal search, the query log is a direct feed of what visitors are looking for and not finding easily. Your support inbox or sales call notes work the same way — the exact phrases customers use to describe their problem are often better keyword targets than anything a tool suggests, because they're phrased the way your actual buyers think.
Group by intent, not just topic
Once you have a list, sort it by what the searcher actually wants: informational ("what is X"), comparative ("X vs Y"), transactional ("X pricing," "hire X"), or navigational (looking for a specific brand). Writing one page that tries to serve all four intents usually serves none of them well — match content type to intent.
When it's worth paying for a tool
Free methods run out of steam at scale — if you need search volume estimates across thousands of terms, or want to track ranking movement over time, a paid tool earns its cost. But for finding what to write about next, the free signals above are usually enough.
Related on Troiana: The Case Against Chasing Every Keyword · How to Add FAQ Schema That Wins Rich Results.
Common questions
Is free keyword research as good as paid tools?
For finding what to write about, yes — the free signals (Search Console, autocomplete, People Also Ask, competitor pages) are the same raw data paid tools repackage. Paid tools add convenience and scale, not fundamentally different information.
How often should I redo keyword research?
Revisit your Search Console impressions-without-clicks list monthly — it changes as your content and rankings shift, and it's the fastest source of low-effort, high-impact fixes.
Should I target exact-match keywords in my headings?
Match the intent and phrasing naturally rather than forcing an exact string — modern search engines and AI answer engines both reward clear, natural language over keyword-stuffed headings.