Keyword Grouping Tool

Group Keywords by Intent and Topic for SEO

Grouping keywords by intent and topical relevance has long been a beneficial strategy to help webmasters understand and improve SEO performance.

Keyword grouping helps to organise content topics identify specific user pain points and helps you structure content so that:

• it is organised
• you are tackling the biggest, most popular and high value topical areas first
• you’re helping to prevent keyword crossover, known as cannibalisation
• Information retrieval for both your website users and search engines ranking your website, is improved

So, here’s how to quickly group keywords today and why it can benefit more than just your SEO performance.

Benefits of Keyword Grouping

The core benefits of keyword grouping are;

• at a glance understanding of sizing, value and density (from a keyword count perspective) of topics
• control over how keywords are bucketed, using methods like; Semantic and NLP (natural language processing), intent and context of keywords
• Lemma-Based or Morphological grouping is easier, faster and more accurate with AI vs old spreadsheet based methods.

This means accurately sizing keyword lists, even in very large quantities and quickly gaining an understanding of how and where the value is can be accelerated.

While keyword grouping might not give you complete clarity over what ranks in search engines today, it can help you size and understand specific topics by user search within an at a glance framework.

With well structured, highly accurate and clear keyword sets, you’ll be better able to;

• identify and target high value keyword and topics with your content
• understand the specifics at scale of what users want and expect to see on pages
• define a content ‘hierarchy’ for each page and section on your website
• help you to clarify achievable goals to leadership teams and executives based on data you have available
• identify uncommon niche intents or user demand outliers that could support your operation
• plan your content more effectively, feeding grouped and subgroup keyword data into task management tools and content delivery systems
• produce faster and more comprehensive content briefs for colleagues creating pages
• measure actual performance of pages vs “predicted” or “total addressable market” insights contained within your dataset
• prevents issues of keyword cannibalisation, helping you “see” keywords as pages and traffic potential, rather than a disparate list
• optimise internal linking by helping you see topics within a hierarchical framework

What is The Most Accurate Keyword Grouping Method?

There are a few different methods commonly used for grouping batches of keywords and these include;

• Lemma-Based or Morphological grouping
• Semantic keyword grouping
• AI based grouping

Lemma, morphological and semantic methods pre-date the widespread mass-adoption of AI LLMs (large language models) and are still used today. However, they generally require an understanding of python to get clear and consistent results.

Unlike these models, AI grouping can be used by anyone who’s able to input data into an AI LLM chatbot or CLI interface.

AI grouping methods can lose focus, lack clarity, hallucinate intent, groups or data or even more frustratingly, get bored and leave tasks incomplete.

This is why we’ve created our keyword grouping tool on the Keyword Universe website.

Unlike AI LLM chatbots, you can copy and paste a batch of keywords into the tool and define the parent topic and topical sub-groups that you’re interested in. Our tool will then:

1. Chunk keywords into batches of 50, 100 or 150 (based on your selected setting of precise, balanced or fast)

2. Send each “chunk” to the chosen AI LLM within a structured prompt that says: “for every keyword, first check if it’s relevant to the parent topic, assigning N/A if not. If yes, assign the best matching subtopic from the attached list.”

3. generateObject forces the LLM to return valid and structured JSON, ensuring the data returned follows a logical; keyword -> topic + intent format

4. live results accumulate in the table on the page as they complete

Where this beats static, code based rules is that this AI keyword grouping will;

• stick rigidly to the user defined subtopics
• it won’t hallucinate or invent new topics, keywords or data
• it’ll complete jobs because it is set up and designed to work through the list, unlike chatbots which tend to lose focus
• AI LLMs can understand keywords which use different words but have the same base level meaning and intent; “get a loan to buy and home” and “mortgage rates” - could both the categorised within the parent topic; “mortgages” where Lemma-Based or Morphological grouping tools might struggle.

While we’ve found the system incredibly useful, it isn’t totally flawless because;

• defining subtopics still requires you to have some familiarity with the entire keyword set (this can be tricky for larger batches of keywords)
• bulk lists will include outliers where intent and topical relevance won’t always match
• AI LLMs can be inclined to opt for N/A if a keyword is relevant because it isn’t a good fit for a subcategory, this necessitates including a “catch all” sub-topic, which can act as a default group, however this group can then be used frequently.

We’ve helped you to overcome these minor drawbacks through daily usage and testing of the grouping tool, implementing;

• the “Auto Topics” mode; where 200 keywords are sampled by the AI LLM and used to define a list of 5 - 15 distinct topical subcategories automatically, which accurately reflect the contents of the list
• user customisable, each auto topics recommendation can be updated to best suit your requirements, prior to the run, helping to maximise accuracy and relevance
• Running keywords in batches, especially if you’re working through a very large list of keywords, is beneficial. The process we’ve used is;
• define a set of consistent and versatile sub topics based on keyword sets (using either auto topics or your own topics or a combination of both)
• running keywords through our grouping tool in batches
• retrieving all keyword groups and identifying all keywords with “N/A” grouping after the final batch is tested
• re-running these “N/A” keywords back through the tool with updated and refined subtopics

This method can help you to;

• compile an initial batch of mappings and groupings for your keyword sets
• refine, review and re-examine keywords which are relevant but may not have fit logically or clearly within existing buckets

Approaching keyword grouping in this ‘batched method’ has helped us to group many thousands of keywords, into clear, logical, organised and well structured groups.

Keyword Grouping Methodologies Compared NLP vs Lemmatisation vs Manual Methods vs AI Tools

Comparing the different methodologies for comparing keywords can help you identify the best methods for your needs. Here are the pros and cons of each method, specific use cases and drawbacks to help you quickly reference them.

Each of these methods can help your business benefit from benefits that contextual keyword targeting presents. Plus, you’ll be able to infer potential value too. Picking a specific methodology, type of keyword grouping and overall approach will also help reduce the differences which may exist within cross functional teams in your business.

Consistency in Approach Across Teams

This can be incredibly beneficial for;

• defining a clear, consistent replicable value proposition for each data set
• ensuring each team within your business follows a clear and consistent content production method, helping with things like tone of voice etc.
• understanding relevant searches and keyword usage with similar or the same intent, which use different words and would not have been captured using a string or word matching method of grouping.

If multiple teams in your organisation are writing content on different angles around the same topic, you can “port” a grouping structure, alongside sub-topics, to help each team achieve a clear and consistent message for each content piece.

Also, because you’re able to define your own topics, you’re able to extract, define and group keywords that have a highly specific;

• demographic fit
• user intent
• opinion or idea
• disambiguation clarity, where abbreviations woven into keywords create uncertainty, this tool will help you to understand specific user meanings, mistyped errors and alternative interpretation. e.g. MVP might mean very different things to sports fans vs software developers, so contextual understanding of these types of ambiguous keywords is greatly accelerated.

With this clarity of understanding, you can ‘programme’ this tool to hone your targeting down to the sets of keywords most valuable to your website and your pages. Coupled with tools like the query fan out tool, you’re able to also factor in specific user concerns, worries or preconceptions, helping you better answer user queries on your pages and through your content.

Accuracy and potential limitations of different grouping methods

While the tool is incredibly useful, there are quirks which an understanding of upfront, will help you get the most out of it.

It doesn't monitor rankings, tell you what is ranking now or how to rank for specific keyword groups

While the tool is useful for categorising and organising your keywords, it isn’t a substitute for full keyword clustering, where you can understand which pages rank for specific keywords. The clustering technique is designed to help you group and understand which topics are required to rank for a specific keyword set or keyword group. Keyword grouping is the step prior to this, where you define the groupings, based on a large set of queries.

There’s No Live Ongoing Monitoring

Keywords, search behaviours and intents change over time. The dawn of AI search interfaces has seen query lengths increase drastically;

Our query fan out tool page, ranks for extremely long search terms, which we’d have unlikely ever seen before AI search, e.g.;

• Which query fanout providers have the best reputation for customer service? and
• what are the best mid-market aeo tools for understanding query fan out for prompts?

This tool operates on a static list, which will require re-runs and recategorisation as search intents evolve.

There’s No Search Engine Scrape Check

There’s nothing here to align these keywords with value from search engines, unless you provide the data.

Assuming that your keywords are;

• searched for
• valuable
• something your customers would input

Is very often the first mistake made by marketers trying to get into search marketing. Ensure that your source search queries from verified tools to get at least some idea of value.

Drift

While the keyword tool is accurate and useful, the batching method used to group very large lists can cause ‘drift’.

This means, as batches of keywords are processed, memory between batch 1 and batch 5 is not stored.

Some accurate and relevant queries in batch 1 might be included, while equally accurate and relevant queries in batch 5 aren’t because the AI’s contextual understanding has ‘switched’.

Variance

Rerunning keywords multiple times through the tool, with the same groupings and subtopics can yield different results.

While this isn’t necessarily an issue, it does arguably mean results are inconsistent.

If a keyword was grouped as “user looking to apply for a credit card” in the first instance, but grouped as, “user looking to transfer an existing balance” in batch 2, then a human decision is likely needed to verify the best group.

This is an issue with AI LLMs and how they're fundamentally designed, rather than our tool.

Every word within AI LLM datasets is weighed against each other word, helping the each model decide the "likelihood" of the next word should be. Meaning results can vary, even if data is the same.

We always recommend including a human review step to assess and regroup AI keyword groupings to guard against issues.

Input Quality

Similarly, where grouping subtopics are very close in their intent for example including; “user looking to apply for a credit card” and “user looking to transfer an existing balance”, you may find keywords aren’t grouped entirely as expected.

This is because with this example, users must specifically apply for a credit card to enable them to transfer a balance to it.

Whether the user is aware of this themselves will affect the query they input to search, so trying to separate these users into 2 groups can prove tricky.

While some queries will have very obvious balance transfer intent, other queries may only infer this intent or mask underlying intent with chosen search phrases. You can help guard against tight keyword groups by;

• reviewing keywords upfront
• specifying exactly what you need from your groupings
• defining clear sub topics and groups
• avoiding overlapping sub topics

When using auto topics, first define a clear and focused parent topic and review and refine your sub topic niches, rather than accepting them without review.

Hyper Niche Ambiguity

Almost infinite numbers of acronyms exist, across different niches, subtopics, industries and disciplines.

• ARG can mean alternate reality game or Argentina, it can also represent mathematical arguments and is known in Persian language regions as a word to describe a citadel or fortress

In cases where little or no additional context is added over and above an abbreviation or acronym, you’re likely to get keywords assigned to N/A buckets most frequently.

Defining common acronyms as part of the parent topic or within subtopics makes sense to help with groupings. However, you should always expect some “N/A” terms.

Keyword Categorisation Methods

There are different methods for keyword categorisation, these include;

ngram and textdistance methods

This tool uses neither of these programmatic methods because it connects directly with an AI model to process groupings. Both of these methods are useful for keyword groupings, but they operate in different ways:

N-gram - exact matches words or sequences of words (e.g. credit card, balance transfer, interest rate). It doesn’t help at all if you’re trying to understand synonyms like; “borrowing money”, “0% borrowing” etc.

Text distance, like Levenshtein and Jaccard measures letters and characters which overlap to understand intersecting patterns. While this is useful for grouping words that are spelt the same and have a very close text distance, it doesn’t help you understand ‘intent’. e.g. “avois balance transfer credit card” and “balance transfer credit cards avoid” would be a programmatic match, but clearly have differing intents.

So, AI grouping does add a layer over traditional, programmatic keywords grouping methods.

Manual Keyword Grouping Methods Using Google Sheets or Excel

Other methods can be used to group keywords, many of these pre-date the commoditisation and mass market adoption of AI, meaning they’re based on spreadsheet formulas, pillar approaches or silo structures.S

We’ve been pioneering these methods for a while and they include;

• Keyword matching, excel, using a simple word list you can create a formula to “group” keywords based on matching words right in excel
• Question term identification, using a simple set of “question” term identifiers to highlight specific user queries from a large list
• Manual methods, SERP tracking and grouping

While these methods still benefit content marketing strategy and content pillar creation, they’re much more outdated now with the advent of AI generated automations.

Using them alongside your AI grouping can help you subdivide your data into clear sub groups and identify key questions to replicate through your pages or as FAQ sections.

Our users have found this tool incredibly useful, because manually grouping thousands of keywords in spreadsheets is extremely time consuming.

How To Group Keywords

Our recommended perfect process for keyword grouping is:

1. define seed keywords, using an AI chatbot or interface to spit out “seed ideas”
2. run seed keywords through keyword tools, like ahrefs, SEMRush answer the public and keywordtool.io
3. define a set of “head keywords” where there’s clear relevance and value to the topics you’re interested in, using keyword matching excel formulas
4. extrapolate your list upwards, identify closely related “shoulder keywords” and check keywords against Google and Bing to find related “suggested queries”
5. ensure every query is backfilled with accurate search volume, CPC, trend and intent data
6. merge all keywords together, deduplicating but maintaining data
7. run deduped keywords through our grouping tool, with a high level “topic” defined
8. If necessary, rerun grouped keywords through our tool again, identifying and defining subgroups for each parent group
9. Run manual keyword grouping to identify common question terms for quicker analysis of your organised data set
10. Identify commonly occurring themes and ideas, content groups and hubs to plan your content strategy around.
11. Define pages based on user intent, so you serve;
    • Pages designed to convert users to high intent searchers
    • pages designed to solve problems and answer questions, providing expertise to those who need more information

What Is The Difference Between Keyword Grouping And Keyword Clustering

The core differences between keyword grouping and clustering are understood as;

• keyword grouping is organising keywords into a static “parent list” of topics and groups related to a ‘parent topic’ you’re interested in
• keyword clustering is generally understood as; understanding the “clusters” of keywords needed to include on a page, to help ensure it is competitive in search results

While very similar in terms of concepts, they can be quite different in terms of how they are sourced, how they are understood and interpreted and how they are used.

• Keyword grouping = planning phase, user research, understanding how and what users search for, planning content topics, understanding user concerns and their needs, helping you understand performance estimates and total addressable markets as well as giving clarity on defining your expertise.
• Keyword clustering = clarifying how to win in search, assessing content already in existence, understanding how to structure your website based on what is working in search results now.

The differences between keyword grouping and keyword clustering are slight. Often used interchangeably there can be impactful differences depending how you decide to group keywords.

Simply understood:

• Keyword grouping is achieved by assessing lists of keywords to understand closely related topics, keywords with similar intent, similar meaning
• Keyword clustering is based on defining which of the same pages rank across search queries (historically in Google search results) then using this ranking data to inform how keywords are grouped.

With Google homogenising results with AI overviews, AI LLM query fan out data becoming readily available and large language models able to better synthesise big groups of keywords, gaining clear and accurate understandings of how search phrases and keywords group semantically is simpler.

While ranking in Google is still incredibly valuable to most websites, there’s a whole new layer of keyword and search data, like prompts, now available to marketing professionals trying to define content strategies that earn ROI.

Keyword grouping is also critical for specific phases of projects, including;

• research and planning
• value estimation
• forecasting
• quick segmentation for reporting
• rank tracking

Keyword clustering still has uses, especially if your website is focused on traditional SEO outcomes and your aim is Google rankings.

However, ensuring that your website is both:

• agent ready
• employing tactics like query fan out
• clearly displays topical authority through content

You’ll be better placed overall, with content more likely to be referenced in AI LLMs, synthesised to build AI overviews all while earning rankings too. Clustering can help you avoid producing multiple pages which end up targeting the same keyword (reducing what is known as cannibalization) in Google rankings, help you rank one page in search engines for multiple keywords, accurately match search intent of users, and improve your website structure.

Ultimately, both are beneficial.

Keyword grouping for understanding the wider search landscape, planning and understanding opportunity sizes.

and

Keyword clustering for understanding what users really want to see when they arrive at a page, what’s working now and how to beat it.

How often Should I Review And Revisit My Keyword Groups?

Returning to review keyword groups is a tricky topic.

In reality, you’re never really going to feel like this is a valuable or additive task.

However, with search changing faster than ever, there’s never been a better time to examine when, how and the frequency of keyword grouping updates, to improve your success.

Here’s what we recommend;

• Revisit ranking pages and their keywords in Google Search Console frequently

  • Even if you only consider keywords as a user research facility, then you’re missing out on an important data set by not understanding what search engines recognise as a “good potential” match for your pages and content.
  • leveraging what people input when your pages are surfaced in search results can help you improve products, alter web design to improve conversion and grow.

• Updating your keyword list with these queries can help you;

  • Understand growing topics, as trends change, products and services become established, new opportunities arise.
  • if you want to compete in these areas, ensuring that you’re part of that trend can help you convert and sell more
  • Ensuring that your products and services reflect updated language, demands and technology entering your space.
  • keeping up to date on the latest trends should be something you do anyway, but understanding how customers translate and understand search trends is only going to be beneficial when defining new packages, products and services.

Depending on your project, progress towards a live state, content publishing cycle strategy and frequency and your topical niche, we’d recommend returning to refresh and update your keyword groupings;

• once every 3 months for time sensitive or areas such as news
• once every 6 months for more “evergreen” topics or websites; SaaS sites, affiliates
• at least once per year for anything else, simple lead generation websites, specific business sites or annual event cycle businesses

How This Tool Works

Manual Topics mode:

1. User defines a parent topic (“bathroom paint”) + subtopics (“Ceiling paint”, “Wall paint”, etc.)
2. Keywords are chunked into batches of 50/100/150 (your Precise/Balanced/Fast setting)
3. Each chunk is sent to the chosen LLM with a structured prompt that says: “for every keyword, first check if it’s relevant to the parent topic — if not, assign n/a. If yes, assign the best matching subtopic from this exact list.”
4. generateObject with a Zod schema forces the LLM to return valid structured JSON — keyword → topic + intent (navigational/informational/commercial/transactional)
5. Up to 4 batches run concurrently, results accumulate live in the table as they complete

The LLM is explicitly told: don’t invent new topics, only use exact strings from the provided list. That’s the key constraint that makes it a classifier, not a clustering tool.

Auto Topics mode (two-steps):

• Step 1 — discoverTopics(): sends a sample of up to 200 keywords to the LLM and asks it to infer 5–15 distinct topic categories covering the list. Returns those as a suggested subtopic list. User can review/edit the suggested topics

• Step 2 — same as manual mode, using the auto-discovered topics as the subtopic list

We’ve found this tool to be incredibly effective when we’ve grown a keyword list into the thousands and we want a quick way to fully understand how to organise, group, interpret and leverage them.

Which then makes decisions like;

• defining specific pages we’ll need to build simpler and easier

Keyword Grouping Tool FAQs