What is the Keyword Research Tool?
| What it answers | Real Google autocomplete expansions plus Search Console opportunity mining. |
|---|---|
| How the answer is produced | The first half of the tool uses Google's own autocomplete — the suggestions that appear under the search box as millions of people type. |
| What you need to enter | Start with autocomplete. |
| Where it stops being reliable | Autocomplete shows that people search a phrase, not how many. |
| Cost and sign-up | Free, runs in your browser, no account and no stored inputs. |
Where this keyword data actually comes from?
The first half of the tool uses Google's own autocomplete — the suggestions that appear under the search box as millions of people type. Autocomplete is not a guess or a model's opinion; it is a ranked list of searches real users have performed often enough for Google to predict them. We query it eight different ways for your seed keyword: the bare phrase, question forms (how, what is, why), commercial forms (best, cost), comparisons (versus) and an audience form (for beginners). The results are deduplicated and grouped so you can see the intent behind each cluster of phrases at a glance.
The second half connects to Google Search Console, which is the only source of keyword data that is measured rather than estimated. Search volume tools from third parties extrapolate from clickstream panels and Keyword Planner ranges; Search Console reports the actual number of times a site appeared for each query, the actual average position, and the actual clicks. That makes it the most honest keyword research instrument available — with one structural limitation, which is that it can only report on queries where the site already gets impressions.
The Search Console section therefore runs a fixed analysis rather than a generic report. It pulls every query the site appeared for over the last 28 days and sorts them into three buckets that map to the three highest-leverage actions in SEO: keywords ranking on page one below the fold (quick wins), keywords on pages two and three with real impression volume (striking distance), and keywords ranking in the top five that almost nobody clicks (a title and description problem, not a content problem).
How do you use the Keyword Research Tool?
- 1.Start with autocomplete. Enter the broadest honest phrase for your topic — 'ai side hustles', not 'ai side hustles for graphic designers in 2026'. Broad seeds give autocomplete room to show you the full neighbourhood of real searches.
- 2.Read the intent groups, not just the phrases. A keyword that only surfaces under 'Questions' wants an informational page; one that surfaces under 'Commercial' wants a page with pricing, comparisons and a decision. Matching the page type to the intent group is most of what ranking is.
- 3.Copy the group that matches the page you plan to build, and use the exact phrasing. Autocomplete phrases are written in the words searchers use, which are frequently not the words professionals use — that gap is the opportunity.
- 4.Then load the Search Console section. Any striking-distance keyword with triple-digit impressions is a page you should build or expand this month, because Google has already told you it considers the site relevant.
- 5.For CTR-upside keywords, rewrite only the title and meta description first. Changing the body of a page that already ranks top five risks losing the position; changing the snippet does not.
What can this tool not tell you?
- Autocomplete shows that people search a phrase, not how many. A suggestion with ten times the volume of another appears identical in the list — use it for phrasing and intent discovery, and validate volumes elsewhere before committing a content budget.
- Search Console data describes this site's past, not the market's present. It cannot tell you the volume of a keyword the site has never appeared for, and positions are averages that can hide wide variation between locations and devices.
- Very new sites will see thin or empty Search Console results. Google typically needs several weeks of impressions before the query report is meaningful, and data arrives with a delay of roughly two days.
How to turn a keyword list into rankings?
The most common failure in keyword research is not picking the wrong keywords — it is treating the list as the deliverable. A list becomes rankings only after three more decisions: which page each keyword belongs to, what type of page that is, and what the page must contain that the current results lack. The intent groups this tool produces are the first of those decisions made for you. Question phrases cluster onto informational pages, 'best' and 'cost' phrases onto commercial pages, and 'vs' phrases onto comparisons. Splitting one article across all three intents is the reliable way to rank for none of them.
The second decision is consolidation. Twenty autocomplete phrases are rarely twenty pages; they are usually three or four pages with sections. A practical test: if two phrases could be answered by the same paragraph, they belong on the same page. If answering them well requires different examples, different data or a different promise in the title, they deserve separate pages. Cannibalisation — five of your own pages competing for one query — does more damage to small sites than weak content does.
The third decision is where the Search Console buckets earn their keep. Quick wins and CTR-upside keywords are editing tasks, measured in minutes: retitle, rewrite the first paragraph, add the internal links the page should already have. Striking-distance keywords are building tasks, measured in hours: a new page, or a major expansion of the ranking one. Doing the editing tasks first is not just easier — the clicks they recover arrive in days, which funds the patience the building tasks require.
Finally, close the loop. The queries this tool shows you today are the baseline; the same report in a month tells you whether the pages you built moved. Keyword research done once is a guess. Keyword research re-run monthly against your own impressions is a feedback system, and feedback systems are what separate sites that grow from sites that publish.
What do worked examples look like?
From seed to content plan in ten minutes
A site about home coffee enters 'espresso machine'. Autocomplete returns question phrases ('why is my espresso machine leaking'), commercial phrases ('best espresso machine under 500', 'espresso machine black friday') and comparisons ('espresso machine vs nespresso'). The questions become a troubleshooting guide, the commercial phrases a buying guide, the comparison a head-to-head page — three pages, three intents, zero overlap, all phrased exactly as searchers phrase them.
Acting on a striking-distance keyword
Search Console shows 'ai meeting notes cost' at position 14 with 300 impressions and four clicks. Google already trusts the site on the topic but no page answers the cost question directly. The fix is a dedicated page with real pricing numbers in the first hundred words and the phrase in the title. Because relevance is pre-established, pages like this frequently reach page one within a month — no links required.
What do people ask most about this tool?
Is autocomplete data reliable enough to plan content around?
For phrasing and intent, yes — it is a direct sample of real searches, which is more than most paid tools can claim. For volume, no. Treat it as a map of what people ask, then size the opportunity with Search Console or a volume tool before writing ten articles.
What is a striking-distance keyword?
A query where your page ranks between positions 11 and 30 — the bottom of page two to the middle of page three. These matter because Google already associates your site with the topic, so a dedicated, deeper page often reaches page one within weeks, while a brand-new topic can take months.
Why does a top-five ranking get almost no clicks?
Usually because the title and description do not match what the searcher hoped to find, or a rich result, AI Overview or ad block is absorbing the clicks above you. The diagnostic this tool applies — top five with under two percent click-through — isolates exactly the pages where a snippet rewrite is the fix.
How fresh is the Search Console data?
The tool queries the most recent 28 days, ending three days before today to allow for Google's processing lag, and caches the result for six hours to stay comfortably inside API quotas. It is the freshest complete window Search Console offers.
Should I still use a paid keyword tool?
If you are choosing between topics you rank for nothing on, yes — that is the gap Search Console cannot fill. For deciding what to write next for a site that already gets impressions, this data is better, because it is measured on your actual rankings rather than modelled from someone else's panel.
Which related tools should you try next?
Written and reviewed by Jim Vernon, Editor, AI Intelligence International. Published by AI Answer Engine, a service of AI Intelligence International, and checked against our editorial standards.