6 Free Keyword Clustering Tools and Their Limits

Free keyword clustering tools can turn a research export into a manageable set of topics, but their limits matter as much as their grouping method. A 50-keyword planner, a larger English-only semantic tool and a grouper that retains your spreadsheet metrics solve different problems.

The six options below accept keywords you already have. Choose by batch size, output controls and what you need to check afterward. A suggested cluster can help organize work; it does not establish that Google returns the same pages for every term, or that a new article is needed.

Compare the free limits before importing your list

This is a comparison of current official tool pages and documented limits, checked on October 27, 2026. It is not an accuracy benchmark. We selected tools that group an existing list without requiring a paid subscription, and compared their inputs, grouping controls, outputs and restrictions. Paid trials are not counted as permanent free access.

Screpy publishes this article and provides one of the tools. It appears under the same criteria as the other options; the numbering follows different workflow needs, not measured performance.

The paid products in the last column are optional, broader products. Their research, document or article allowances are not the free clusterers' quotas. Likewise, Screpy's seven-day dashboard trial and RankYak's three-day platform trial are separate from their free tools.

Annual equivalents require annual payment. Screpy's pricing page also displayed a 60% welcome offer with a 72-hour window; the table uses standard prices so that a promotion is not mistaken for the normal tariff. Check the current checkout, billing period and eligibility before buying.

1. PEMAVOR: group exports with their metrics

Homepage image: PEMAVOR.

PEMAVOR's keyword grouper is useful when a list arrives with numbers you want to keep. It accepts keyword data with additional metric columns, offers delimiter and header controls, and aggregates supplied metrics into the grouping output. Similarity and minimum cluster size can be adjusted.

For example, an editorial team could prepare this small, fictional Search Console export:

Keyword,Clicks,Impressions
folding bike,12,400
folding bicycle,8,250
folding bike hinge repair,3,90

Select the comma delimiter and indicate that the first row contains headers. After grouping, inspect which terms contributed to each total before exporting CSV. Those totals summarize the input rows; the tool has not discovered new demand or measured additional searches.

The capacity wording needs care: the form advertises 10,000 keywords, but the instructions later mention 50,000. Use the lower figure when planning a job. A large import is not worth organizing around an unresolved claim.

Choose it for metric-aware topic triage. Choose an intent-oriented option when your main concern is distinguishing a buying guide from a repair task, and verify the search results before treating either grouping as a page plan.

2. Zenbrief: control semantic cluster size

Zenbrief's clustering tool suits an English keyword list that needs more control over grouping boundaries. Its FAQ describes a semantics-based method, and the interface exposes similarity, minimum-size and maximum-size settings rather than offering only a single grouping button.

The stated free allowance is up to 5,000 keywords, English only. Although the interface displays more language choices, that does not establish that those languages belong to the free allowance. Search volume and difficulty are paid features; saving sessions requires a free account. The pricing page was unavailable during our check, so there is no verified current paid price to quote.

A useful workflow is to keep the same source list and compare two settings. Inspect a borderline pair such as "folding bike price" and "best folding bike" after each run. Record why you accepted or rejected the grouping; a cleaner-looking cluster is not automatically the right page boundary.

Choose Zenbrief when adjustable semantic groups are the priority. For a smaller list where explicit unassigned terms and original-term tracking matter more, consider Screpy. For metric-rich exports, PEMAVOR addresses a different need. None of these choices removes the need to inspect the intended market's results.

3. NEURONwriter: assign keywords to existing groups

The NEURONwriter grouper, powered by CONTADU, offers two useful starting points: create groups from common subwords, or assign terms to parent groups you provide. The page describes the free method as basic text-distance analysis. Its invitation to try broader semantic and intent analysis is a separate product offer.

Choose the predefined-group mode when the site already has a sensible structure. For a bicycle store, parent labels might be "folding bikes", "repair" and "accessories". Paste the query list, supply those labels and review the unmatched terms. A query rejected by a strict setting is a review item, not evidence that nobody searches for it.

The interface provides matching strictness controls and an option to try grouping every keyword. It warns that broader matching can reduce consistency. Do not force complete assignment just to remove an untidy remainder. The report preview includes a print/PDF control; the page does not disclose a precise current numerical batch ceiling.

This is a useful sorting aid for a taxonomy you have already planned. It is less suitable when you need the tool to establish whether different queries share ranking pages. A common word or a neat parent label does not demonstrate live SERP overlap.

4. Screpy: keep ambiguous keywords visible

Homepage image: Screpy.

Screpy's AI Keyword Cluster Generator is designed for a small existing list whose terms you want to account for. It accepts 2-100 nonblank input lines, with 2-100 characters per term and at least two unique terms after duplicate handling. A valid email is required, and request limits apply. This is a free utility, separate from the paid Screpy dashboard.

The output includes suggested groups, representative keywords, likely intent and explanations. Ambiguous or unrelated terms can remain unassigned. CSV export includes those terms; Markdown copy is also available. Optional website or market context can help clarify ambiguous wording, while output-language selection changes labels rather than translating the original keywords.

For example, include "java" in a bicycle keyword list only if its source makes it relevant. Review an unassigned explanation before removing the term, or add context if it refers to a genuine product. The useful result is an accountable list, not a claim that every word belongs somewhere.

The tool uses semantic inference, not live Google results. It supplies no search volume, difficulty, SERP overlap or cannibalization analysis. Its suggested primary keyword is a representative, not a verified demand leader.

Choose it for small batches where preserving original terms and reviewing exceptions are important. Use another option for larger inputs. Form values and email are recorded with usage, so avoid confidential research or customer information.

5. RankYak: label a small content-planning batch

Homepage image: RankYak.

RankYak's free keyword clustering tool accepts up to 50 keywords per run and groups them with suggested primary keywords and intent labels. It provides copy and CSV output. Its explanation distinguishes meaning-based grouping from clustering based on live SERP overlap.

Choose it when a short shortlist needs readable page-topic candidates. For a list containing "folding bicycle", "best folding bike" and "folding bike hinge repair", inspect the proposed groups and ask what a reader would expect from each destination. These are example review questions, not reported tool results.

The primary label is useful for naming a working topic. It does not establish search volume, ranking potential or which project deserves attention first. Keep any research metrics attached to their original terms rather than assigning the group's label an invented combined demand.

For larger work, the 50-term allowance is a real workflow constraint. Separate batches can produce similar topics that need another comparison before you create pages. If that manual joining becomes substantial, a larger-list tool may be more convenient.

RankYak's paid publishing platform and its three-day trial are separate offerings. You do not need to interpret the platform's article allowance as the capacity of the free clusterer.

6. KeySearch: start with a plain keyword list

KeySearch's free cluster tool offers a simple starting point: paste a list and organize it into related topics. Its FAQ describes the method as semantic relevance.

Choose it for an initial sort when you do not need extensive settings. Keep a separate sheet with the original keywords, suggested topic and your decision. That lets you review a grouping without making the tool's presentation your only record.

The official page does not specify a numerical free batch allowance or document the export format. Verify those requirements before moving a large project into it; "not stated" does not mean unlimited. The separate KeySearch suite's research allowance should not be treated as a limit for this free utility.

For more explicit batch sizing, consider Zenbrief, Screpy or RankYak. If the decision depends on actual Google result overlap, collect that evidence separately rather than assuming that semantic relevance establishes it.

Check the clusters before assigning pages

Use a small, deliberately awkward list before importing a whole project. Include wording variants, related but different tasks and an ambiguous term. Here is an illustrative input, not a benchmark dataset or an observed output from any tool:

folding bike
folding bicycle
best folding bike
folding bike price
folding bike hinge repair
java

A useful review might look like this:

The point is not to force a particular number of clusters. A broad semantic relationship can be useful for research without making the terms suitable for one article.

Check that the keywords survived

Keep the original input alongside the export. Count unique terms, then inspect for missing keywords, invented additions and duplicates across groups. If you remove duplicate input, retain a way to trace each original row. Terms left unassigned belong in the review record, not in a silent discard pile.

When a list must be split into batches, compare the outputs afterward. A term can receive different neighboring terms simply because the rest of the input changed. Do not create separate pages for similarly named groups from different batches without another review.

Check the reader task before choosing a URL

Search the important terms in the same country, language and device context. Record the date, dominant page types and relevant ranking URLs. Compare the actual pages rather than relying on a tool's intent label. Several shared results can support a same-page hypothesis, but there is no universal overlap threshold in this article.

Then compare your existing pages and drafts. A cluster may call for updating a current guide, adding a section or doing nothing. Use a keyword mapping template for your site type once you have made that decision.

Keep the reasoning with the export

This copyable CSV is an example review record. The decisions are illustrative and remain pending until you examine real results and existing content:

keyword,provisional_topic,action,evidence_needed
folding bike,General folding bikes,Review existing page,Market and page-type check
folding bicycle,General folding bikes,Compare with folding bike,Same-task and result overlap
best folding bike,Buying comparison,Keep separate for review,Comparison-page evidence
folding bike price,Price information,Keep separate for review,Category versus pricing-guide intent
folding bike hinge repair,Hinge repair,Find existing support page,Repair coverage
java,Unassigned,Clarify source,Meaning and relevance

Keep source metrics in separate columns. Summed impressions are not unique searchers, and repeated source rows can inflate a total. A tool that names a primary keyword has not necessarily measured its demand.

For a wider content plan, the next step is planning topic clusters around distinct reader questions. Choose the free utility that fits your input and review needs, then make the page decision from evidence rather than the appearance of a tidy group.