Semrush Keyword Magic Tool: Filters Most People Miss

The Semrush Keyword Magic Tool is a keyword research database that turns one seed term into thousands of related keywords, grouped into topics, with search volume and difficulty for each. Most people type a seed, glance at the list and leave, which wastes the tool’s real power: its filters. The ones people miss most are the Intent filter (to isolate commercial or transactional terms), the Questions filter (for FAQ and People Also Ask ideas), the Keyword Difficulty filter (to hide terms you cannot realistically rank for), and the match type switch between Broad, Phrase, Exact and Related. Used together, these turn an overwhelming list into a short, ranked, winnable shortlist. This guide walks through the tool and the filter recipes we use, as of July 2026.

What the Keyword Magic Tool does

The Keyword Magic Tool sits inside Semrush’s keyword research toolkit and exists to expand a single idea into a full map of what people search. You enter a seed keyword, pick a country database, and it returns every related keyword it holds, organised into topic groups down the left, with volume, keyword difficulty (KD), intent and other metrics for each. The value is not the raw count, it is the filtering: you can slice thousands of keywords down to the few dozen that match your intent, your difficulty ceiling and your format. For a UK business, the single most important first step is setting the database correctly, because the default is often the United States, and US volumes will mislead you badly.

Set the UK database first

Before you read a single number, set the database to United Kingdom. This is the most common mistake we see, and it quietly ruins the whole exercise. Semrush holds separate keyword data per country, and it frequently defaults to the US database, where volumes, competition and even spellings differ from the UK. A term that looks worth chasing at 5,000 US searches may be a fraction of that in Britain, and American spellings such as “optimization” will show instead of the British “optimisation” your customers actually type. The database selector sits next to the search box: change it to the United Kingdom (or the specific country you serve) and confirm it before you export anything. Getting this wrong is the difference between a keyword plan built on your market and one built on someone else’s.

The numbered walkthrough

Here is the workflow we run, start to finish, when researching a new topic.

1. Set the country database to United Kingdom.
2. Enter a broad seed keyword (for example, “garage doors” or “family solicitor”).
3. Switch the match type to Broad to see the widest set first.
4. Open the topic groups on the left to understand how the subject splits.
5. Apply the Keyword Difficulty filter to your ceiling (0 to 29 for a low authority site).
6. Apply the Intent filter to match the page you are planning.
7. Use the Questions filter when you want FAQ and content ideas.
8. Add Include or Exclude keyword rules to focus or clean the list.
9. Select the keywords you want and add them to a keyword list or export to CSV.

Run this once per topic and you finish with a focused, prioritised shortlist rather than a spreadsheet of noise.

Filter recipes most people miss

These are the specific filter settings we reach for most, ready to copy.

• Questions only: turn on the Questions filter to surface “how”, “what”, “why” and “can” phrases. This is the fastest way to find FAQ content and terms that win AI Overview citations.
• Low difficulty shortlist: set Keyword Difficulty to a maximum of 29 (or 19 for a brand new site) so the list only shows terms you can realistically rank for soon.
• Commercial intent: set the Intent filter to Commercial or Transactional to isolate keywords with buying intent, ideal for service and product pages.
• Broad to Exact: start on Broad match for discovery, then switch to Phrase or Exact to tighten in on a specific commercial phrase and its close variants.
• Exclude your brand: add your brand name to the Exclude rule so branded searches do not clutter your non brand research.
• Include a modifier: add “near me”, “uk”, or a town name to the Include rule to focus on local intent.

Match types explained simply

The match type switch changes how broadly the tool interprets your seed, and knowing the difference saves a lot of scrolling. Broad match returns the widest set, including loosely related terms, which is best for discovery when you are mapping a new topic. Phrase match returns keywords containing your seed phrase in order, tightening the list to close variants. Exact match returns only the exact keyword and its very closest forms, useful when you already know the precise phrase and want its metrics. Related shows semantically connected keywords that may not contain your seed word at all, which is where you find angles you had not considered. The practical habit: begin on Broad to see the whole landscape, then move to Phrase or Exact once you know which cluster you are targeting. Our keyword research walkthrough covers the full research process this fits into, and our guide to keyword difficulty explains why the KD scores you filter on differ between tools.

An honest note on the free plan

You do not need a paid Semrush plan to try the Keyword Magic Tool, but the free plan is tightly capped. As of July 2026, a free Semrush account limits you to roughly 10 searches a day across its tools, which is enough to test the Keyword Magic Tool on a handful of seeds but not to run real, sustained research. If you are researching a full site, you will hit that ceiling within minutes. If the caps bite and you are not ready to pay, our comparison of genuinely free keyword research tools covers eight no-cost alternatives and the exact point where each one runs out. The 7 day trial on a paid plan (card required) removes the cap for a week, which is the sensible way to do a proper research sprint before deciding whether to subscribe. If the tool earns its place, our Semrush pricing guide covers which plan makes sense, and our Semrush review scores the keyword modules against the rest of the platform. Browse the wider set of guides in our SEO tools hub.

Frequently asked questions

What is the Semrush Keyword Magic Tool?

It is Semrush’s keyword research database. You enter a seed keyword and it returns thousands of related keywords grouped into topics, each with search volume, keyword difficulty and intent, which you then filter down to a usable shortlist. It is one of the strongest keyword research tools available.

How do I get UK keyword data in Semrush?

Set the country database to United Kingdom using the selector next to the search box, before reading any numbers. Semrush often defaults to the US database, where volumes and spellings differ. Confirming the UK database is the single most important step for British keyword research.

What is the difference between broad, phrase and exact match?

Broad match returns the widest, loosest set of related keywords for discovery. Phrase match returns keywords containing your seed phrase. Exact match returns only the exact term and its closest forms. Start broad to explore, then narrow to phrase or exact once you know your target cluster.

Can I use the Keyword Magic Tool for free?

Yes, but a free Semrush account caps you at roughly 10 searches a day as of July 2026, enough to test it but not to run full research. A 7 day paid trial removes the limit for a week, which is the practical way to do a proper research sprint.

Which filter finds question keywords?

The Questions filter. It isolates phrases beginning with words like how, what, why and can, which are ideal for FAQ sections and for winning AI Overview citations. Pair it with a low difficulty setting to find questions you can realistically rank for.

How is this different from finding competitor keywords?

The Keyword Magic Tool expands your own seed ideas into related terms. Finding competitor keywords starts from a rival’s domain instead. Both are useful: use the Keyword Magic Tool to explore a topic, and our guide to finding competitor keywords to see what rivals already rank for.