Expand one seed into useful question and modifier patterns
Generate repeatable question, preposition, comparison, and alphabetic prompts for a content brief without inventing demand metrics.
Keyword Idea Generator
Free, no signup. Build questions, prepositions, comparisons, and alphabetic expansions from a seed. Add a bounded public suggestion source when it is available.
Checks run from our server; we fetch the URL you enter and don't keep the results. The optional live pass sends only the seed to the chosen public suggestion source through a cached, rate-limited proxy. Anonymous run-level outcome counters may be used for aggregate research; URLs, domains, IPs, and identifiers are never included, and no statistic is released below 100 runs.
Site passport Local context for this saved site
Local data
Saved targets, named lists, and recent check summaries remain only in this browser.
Rate this tool
How to use it
- Enter a focused seed topic, product, or problem.
- Choose the public suggestion corpus that best matches the search context.
- Select Generate ideas. The heuristic list appears immediately while one live request runs.
- Review ideas by bucket, then copy the CSV with keyword, source, bucket, inferred intent, cluster, and observation state.
Example output Example data
For technical seo audit, the deterministic engine produces 44 planning rows before any live source responds. A few exact examples are:
KeywordBucketIntentState/source
how to technical seo auditquestionsquestionheuristic / heuristic expansion
technical seo audit foralphabeticalinformationalheuristic / heuristic expansion
technical seo audit alternativescomparisonscomparisonheuristic / heuristic expansion
technical seo audit aalphabeticalinformationalheuristic / heuristic expansion
A live row would be labeled observed and name the selected source. No example above claims that status.
What you get
Heuristic means the phrase was assembled by a fixed rule. Observed means a public suggestion endpoint returned it; cached responses are disclosed in the status. Bucket describes the phrase shape, intent is a rule-based label, and cluster is a lightly stemmed planning key. None of these fields measures demand.
How it works
The local pass normalizes the seed and generates eight question prefixes, five preposition suffixes, five comparison suffixes, and 26 alphabetic suffixes. The optional proxy sends only the seed to a fixed upstream host, caches by source and normalized query, keeps up to 12 clean suggestions containing the seed, and returns explicit unavailable or rate-limit errors. Observed duplicates replace equivalent heuristic rows.
Features
- Useful deterministic fallback even when every network source fails.
- Five available public suggestion corpora.
- Clear observed-versus-heuristic provenance on every exported row.
- CSV copy with buckets, inferred intent, and transparent clusters.
Limitations
- No search volume, CPC, trend, difficulty, rankings, or conversion evidence.
- Public suggestion endpoints can change, throttle, localize, personalize, or disappear.
- Question grammar, intent rules, and stemming are simple English heuristics.
- Alphabetic expansions are prompts, not evidence that the source suggests those phrases.
Frequently asked questions
Where do the keyword ideas come from?
The tool always creates deterministic question, preposition, comparison, and alphabetic expansions. It then makes one bounded request to the selected public suggestion source and labels returned strings as observed.
Does an observed suggestion have search volume?
No. Observed means the selected source returned the string for this request or a cached equivalent. It does not provide volume, trend, competition, or conversion value.
What happens when the suggestion source is unavailable?
The live request reports the failure honestly and leaves the full deterministic idea set available. TikTok is explicitly unavailable because there is no stable public endpoint in this implementation.
How many live suggestions can be returned?
The proxy normalizes, deduplicates, and caps source-returned suggestions at 12. It keeps only strings containing the supplied seed.
How are intent and clusters assigned?
Simple English-language rules classify questions, comparisons, transactional terms, and informational terms. Clusters remove a small list of modifiers and apply basic stemming; they are planning labels, not SERP-derived intent.
Feature requests for Keyword Idea Generator
Upvote what you want most. New ideas can be submitted from the floating Feedback menu; requests appear here once approved, and the most-wanted rise to the top.
You won't be emailed about that request anymore.
Loading…
➕ Request a feature
New requests are reviewed before they appear here.
Where this tool helps
Common use cases
Compare ideas by phrase shape and inferred intent
Review transparent buckets and simple rule-based intent labels before deciding which phrases deserve deeper SERP or audience research.
Add bounded suggestions from a public source
Combine up to 12 source-returned strings with the local idea set while keeping every observed row distinct from heuristic expansions.
Keep planning work available when a live source fails
Continue with the complete deterministic idea set when a suggestion endpoint is unavailable, throttled, or returns an unsupported response.
Export a provenance-aware keyword handoff
Copy CSV rows with keyword, source, bucket, inferred intent, cluster, and observation state for review and follow-up evidence gathering.
Watch the full workflow
Keyword Idea Generator walkthrough
Read the transcript
Keyword Idea Generator
This beginner walkthrough explains how Keyword Idea Generator expands one seed into transparent planning ideas. We will cover practical use cases, the interface, a complete fictional run, question and comparison buckets, heuristic versus observed provenance, inferred intent, clusters, C-S-V handoff, source failures, features, limitations, and the research you should do next.
Step 1
Keyword Idea Generator starts with one focused topic, product, or problem. It creates repeatable question, preposition, comparison, and alphabetic phrases locally. An optional public source can add strings it returned. The tool never invents search volume, and it keeps planning rules separate from source observations so beginners can see where every row came from.
Step 2
Use it to expand a seed into question and modifier patterns, compare phrase shapes and inferred intent, add bounded suggestions from a public source, keep planning work available when that source fails, or export a provenance-aware keyword handoff. Each use case produces research leads, not a keyword-volume forecast.
Step 3
Enter a focused seed, choose Google, Bing, YouTube, Amazon, or App Store as the optional suggestion context, then select Generate ideas. The local engine runs first. The live pass sends only the seed through a cached, rate-limited proxy. Choosing a source changes the corpus being observed; it does not add volume or ranking data.
Step 4
The workflow has four steps. Enter one focused seed. Choose the public suggestion corpus that matches the search context. Generate the ideas while the heuristic list appears and one live request runs. Review the buckets, then copy C-S-V rows that preserve keyword, source, bucket, inferred intent, cluster, and observation state.
Step 5
The example table shows four of the forty-four deterministic rows created for technical S-E-O audit. Each row has a keyword, phrase-shape bucket, rule-based intent, and state with source. Every example is labeled heuristic expansion. The note beneath the table is explicit: a live row would say observed and name its source; none of these example rows makes that claim.
Step 6
For the full workflow, enter technical S-E-O audit and keep Google selected. The six source-returned strings in this capture are fabricated planning fixtures, not real Google suggestions. The page processes the forty-four local ideas with the production engine, fulfills the suggestion endpoint locally, and blocks every external response.
Step 7
Choose Generate ideas. The local set appears immediately while the optional source request completes. The finished status reports six observed suggestions plus forty-four deterministic ideas. Copy C-S-V also becomes available. That count describes rows in this run; it does not say how often anyone searches for them.
Step 8
Start with the status line: six observed suggestions and forty-four deterministic ideas. Observed means the selected source returned the string for this request. Deterministic means a fixed local rule assembled it. A cached live response would be disclosed in this same status. Neither label measures volume, trend, competition, ranking, or conversion value.
Step 9
The question bucket begins with how to do a technical S-E-O audit, labeled question, observed, and Google. The rows below it are fixed question-prefix expansions labeled heuristic. This is the important distinction: the first fixture string represents a source observation; the others are repeatable prompts that help a researcher consider common question shapes.
Step 10
Now select comparisons. The list updates without rerunning the source request because every bucket already belongs to the same result set. Tabs organize phrase shapes; they do not rank opportunities. Moving between them helps you review questions, prepositions, alphabetic prompts, and comparisons separately before deciding what needs evidence.
Step 11
The comparison bucket starts with technical S-E-O audit versus site audit, labeled observed and Google in this fixture. The remaining versus, alternatives, competitors, and comparison rows are heuristic expansions. Similar-looking phrases may belong in one topic brief. Keep separate pages only when query behavior, S-E-R-P results, or another strong source supports a distinct need.
Step 12
The What you get section defines every field. Heuristic means fixed rule. Observed means returned by a public endpoint. Bucket describes phrase shape. Intent is a rule-based label. Cluster is a lightly stemmed planning key that removes a small modifier list. These labels make review easier, but none of them measures demand or proves that two phrases share one S-E-R-P.
Step 13
The local pass creates eight question prefixes, five preposition suffixes, five comparison suffixes, and twenty-six alphabetic suffixes. The optional proxy keeps up to twelve clean suggestions that contain the seed, discloses unavailable or rate-limit errors, and replaces an equivalent heuristic row when the source returns the exact same phrase.
Step 14
The issue cards keep planning honest. A heuristic idea is unverified, so validate it with first-party Search Console, paid-search, or keyword-demand evidence. If many observed rows are near-duplicate structural variants, consolidate them into one topic brief and keep separate variants only when query or S-E-R-P evidence supports a distinct page.
Step 15
The main features are a useful deterministic fallback when network sources fail, five available public suggestion corpora, observed-versus-heuristic provenance on every exported row, and C-S-V copy with buckets, inferred intent, and transparent clusters. The result remains usable even when a live endpoint is unavailable or throttled.
Step 16
There is no search volume, cost per click, trend, difficulty, ranking, or conversion evidence. Public endpoints can change, throttle, localize, personalize, or disappear. Question grammar, intent, and stemming are simple English heuristics. Alphabetic expansions are prompts, not evidence that a source suggests those phrases. The methodology note repeats the rule: suggestion strings are not demand data.
Step 17
Copy the C-S-V, preserve every source and state column, and group obvious variants before prioritization. Validate promising phrases with first-party behavior, paid-search data, audience interviews, or direct S-E-R-P review. Use the generator to expand the questions worth researching; use stronger evidence to decide what people actually need and what content should be created.
Use the ideas to ask better questions—not to invent demand.
Copy the provenance-aware rows into your planning sheet, group obvious variants, and mark the phrases that need validation. Then use first-party Search Console or paid-search data, audience research, and direct S-E-R-P review to decide whether a phrase deserves content. Keep heuristic and observed state attached so no one mistakes a useful prompt for measured demand.
Common issues & how to fix them
- Information Keyword idea is heuristic and unverified Fix: Validate the suggested terms in first-party Search Console, paid-search, or keyword-volume data before prioritizing them.
- Information Idea set has high near-duplicate overlap Fix: Consolidate ideas in the repeated planning cluster into one topic brief, then keep separate variants only when query or SERP evidence supports a distinct page.