Short version: type one broad seed keyword, pick the country database, and Semrush returns every related term it holds, sorted into folders by the phrases they share. You then filter by volume, keyword difficulty, intent and SERP features, tick the terms worth keeping, send them to a keyword list, and export. The research part takes about ten minutes. The part that takes the rest of the afternoon is deciding which of those terms belong on the same page, and the tool does not do that for you.
This walkthrough covers the workflow that actually produces a usable plan, the filters worth setting, the export limits per plan straight from Semrush own documentation, and the step you are left holding at the end.
How do you use the Semrush Keyword Magic Tool?
Open the tool, enter a single seed keyword, select your target country database, and hit search. Semrush returns the full related keyword set grouped into subfolders by shared phrase, with volume, keyword difficulty, intent and cost per click on every row. Apply filters to narrow it, select the keepers, and export or send them to a list.
That is the whole mechanic. What separates a useful session from a wasted one is the seed you start with and the filters you set, so both are worth getting right.
Start with a broad seed, not your product name
The most common mistake is seeding the tool with something specific, like a brand or a long phrase you already decided you want to rank for. The tool expands outward from what you give it, so a narrow seed returns a narrow set and you never see the topics you did not think of.
Give it the two or three word head term for the category instead. If you sell scheduling software, seed it with "appointment scheduling" rather than "best appointment scheduling software for dentists". The long phrase will appear in the results anyway if it has volume, along with the forty adjacent phrases you would have missed. Run three or four broad seeds per site rather than thirty narrow ones.
One practical limit here: Semrush caps seed keywords at 100 an hour on every paid tier. That is generous for broad seeding and restrictive if you were planning to fire hundreds of narrow terms at it, which is another reason to work broad.
What the phrase match folders actually do
The left hand panel groups results into folders by the words they contain. Seed "appointment scheduling" and you get folders for "software", "app", "system", "online", "free" and so on, each holding the terms containing that word.
This is genuinely useful for scanning, and it is worth being clear about what it is not. Folders group by shared words, not by shared intent. "Appointment scheduling software" and "appointment scheduling system" sit in different folders and are the same search intent, satisfied by the same page. Meanwhile "free appointment scheduling app" and "appointment scheduling api" might share a folder and want completely different pages. Treat the folders as a browsing aid, not as your page structure.
The filters that matter
Four filters do most of the work, and the order you apply them matters less than actually applying them:
- Volume. Set a floor, but keep it low. A floor of 10 or 20 keeps the long tail that a newer site can realistically win. Setting it at 1,000 leaves you with only the terms every competitor is already fighting over.
- Keyword difficulty. The most valuable filter for a small site. Cap it well below where you think you can compete, because difficulty scores are optimistic about newer domains.
- Intent. Semrush tags terms informational, navigational, commercial or transactional. Filtering to commercial and transactional gets you the pages that convert; filtering to informational gets you the blog topics. Do both passes separately rather than mixing them.
- Exclude. Strip the terms that will never be your buyer. Competitor brand names, "free", job titles, and anything geographic you do not serve.
The cost per click column is worth a glance even if you never buy an ad, because a term nobody bids on is often a term nobody makes money from. If you are buying that traffic rather than earning it, CPC becomes the main column and the sensible next move is putting the campaign itself on autopilot rather than hand building ad groups from a spreadsheet.
How many keywords can you export from Keyword Magic?
It depends entirely on your plan, and the gap between tiers is large. Semrush publishes the limits, and they are worth reading before you pick a subscription:
| Tier | Results per report | Reports per day | Seed keywords per hour |
|---|---|---|---|
| No SEO Toolkit | 10 in the interface, 0 in export | 10 | 10 |
| Pro, $139.95 a month | 10,000 | 3,000 | 100 |
| Guru, $249.95 a month | 30,000 | 5,000 | 100 |
| Business, $499.95 a month | 50,000 | 10,000 | 100 |
Two things stand out. The free tier exports nothing at all, which is the answer to most of the confusion about whether the tool has a usable free version. And the seed keyword ceiling is 100 an hour whether you pay $139.95 or $499.95, so the expensive tiers buy depth per report rather than breadth across topics. The full per-plan breakdown sits on the Semrush pricing page.
Is the Semrush Keyword Magic Tool free?
No. Without an SEO Toolkit subscription the tool shows 10 results per report in the interface and exports zero rows, with 10 reports a day. You can see the shape of the data and take none of it away. Paid access starts at Pro, $139.95 a month, which is a platform subscription rather than a price for this one module.
That pricing is reasonable if you use rank tracking, site audits and paid search data in the same login. It is expensive if Keyword Magic is the only tab you ever open, which is the situation that sends most people looking at a Semrush Keyword Magic Tool alternative.
Read the volume numbers as estimates
Every figure in the export is modelled, not measured. Semrush, Ahrefs and everyone else build volume from a blend of Google data and clickstream panels, which is why two tools can report numbers that differ by a factor of three for the same term. Use the numbers to rank keywords against each other inside one tool, and never to forecast traffic in absolute terms or to compare across tools. There is more on where the figures come from in how accurate keyword search volume is.
The step Keyword Magic leaves you to do
You now have a CSV with somewhere between four hundred and ten thousand rows. Nothing about it tells you how many pages to write. That decision is grouping the terms by what the searcher actually wants, and it is manual work: read the list, spot that eleven phrasings all want one page, notice that two terms in the same folder need separate pages, decide which cluster your site can win this year, and put the whole thing in a publishing order.
For a small site that is an afternoon. For a client site with several thousand rows it is a couple of days, and it is the step people quietly skip, which is how you end up with six thin posts competing with each other for one query. A bigger export does not help. It makes the sorting longer.
This is the part we built Keywordpilot to remove. You give it one seed, and instead of rows it returns the terms already grouped into intent clusters, with a difficulty band per cluster and a publishing order attached, so the output is a plan rather than a spreadsheet. If you want to see the grouping logic on its own, the keyword clustering tool does that step, and the keyword difficulty checker covers the winnability read. Run a seed through the demo at the top of this page and compare it against your last Keyword Magic export.
Put it into practice
The fastest way to apply this article: run your own niche through the free Keywordpilot demo. One seed keyword, twenty seconds, a clustered mini content plan. No account.