Methodology

Data source: live Google Autocomplete

Every keyword in the tool is a suggestion returned by Google's public autocomplete service for your query, country, and language. Live mining reads those responses directly in your browser, falling back to a server-side proxy when browser requests are blocked. If neither answers, the tool shows an error with a Retry button — it never invents keywords behind the scenes.

The three tabs are three ways of querying the same source: related modifier expansions (best, free, vs,template, pricing), direct completions including a–z letter probes, and question phrasing built on who / what / how / when / why. Competitor mode instead derives the alternatives, pricing, review, and comparison queries people search around a rival brand.

How intent is classified

Intent is assigned by a rule-based (heuristic) classifier that reads the words in each suggestion, in one ordered rule set: committed-action words (buy, discount, free trial,near me) are transactional; question forms and how-to / what-is resources are informational; product, price, and ranking words (software, pricing, best, vs) are commercial investigation; direct brand or system access is navigational. It is a practical heuristic, not a reading of any individual searcher's mind — close calls exist, and the label is meant as a first pass you can override in your own head.

What we clean before anything renders

Every payload — live or server-rendered — passes one pipeline: each suggestion is checked for structural sanity and junk phrases (debris, broken fragments), de-duplicated by canonical keyword identity within each tab and across all three (Questions wins a cross-tab tie), and its intent is classified by the rule set above. The pipeline is idempotent: re-validating clean rows changes nothing, so the tool page re-runs it on boot with identical results.

Known limitations

• No search volume data. The tool does not display search volume, CPC, PPC, keyword difficulty, or trend — those numbers come from paid data providers, and we do not invent our own.
• Intent is heuristic. Labels are rule-based classifications of the phrase's words. They are usually right and occasionally wrong; they are a prioritization aid, not ground truth.
• Google may change suggestions anytime. Autocomplete is Google's service: it can reorder, expand, withhold, or localize suggestions without notice, and there is no stability guarantee.
• Results vary by country and language. The same seed returns different phrases for the US vs the UK vs India, and for English vs Spanish. Pick the market you actually rank in.
• No offline fallback. When Google Autocomplete cannot be reached, the tool shows an error and a Retry button. It never generates keywords on its own — if Google does not answer, you get no rows, not made-up ones.

What to use the output for

Discovering the phrasing real people use, grouping it by intent, and deciding which pages to write. It is a research and prioritization aid, not a guarantee of traffic or rankings — before committing budget, confirm critical keywords with your own analytics, Google Search Console, or a licensed data source.

For what we store (nothing beyond transient server logs), see theprivacy policy. Questions about this page? Get in touch.