Free keyword analysis tool: how to judge a keyword before you write

The best free keyword analysis tool is a stack, not an app: Keyword Planner for a demand band, a manual SERP scan for intent, Search Console for your own position, and one free difficulty score from KWFinder or Semrush. Together they answer the only question worth asking before you write, which is whether this page can take a click.

That stack costs nothing and takes about twenty minutes per keyword. It won't hand you a traffic forecast, because no free source knows one. What it does give you is a yes or a no on the effort, with a reason attached to each.

What does a free keyword analysis tool actually measure?

A keyword generator answers what people search. An analysis tool answers whether you should care. That's four separate signals, and almost every free plan returns only the first.

• Demand, as a monthly band rather than a count, which is enough to separate a live term from a dead one.
• Intent, which isn't a number a tool computes but a pattern you read off the top ten results.
• Difficulty, as a score from 0 to 100 that only means something inside the tool that produced it.
• Your own footing, which is the one number nobody can sell you a better version of.

Miss the last one and the analysis is generic. A keyword with a few hundred searches a month and a difficulty of 45 is a bad target in the abstract and a good one if your page already sits at position 12 for a close variant.

The four checks, in the order that saves you work

Run them in the order that fails fastest. Half the candidates should die at step one, before you spend a single lookup.

1. Read the SERP. If the top ten are eight product pages and one forum thread, a blog post loses on click behaviour no matter how good it looks.
2. Confirm demand exists. One Keyword Planner lookup gives you a band. Anything showing zero or an empty band isn't a keyword, it's a wish.
3. Check difficulty once, from one tool. Don't compare it against a score from a different tool, because the scales aren't the same scale.
4. Check your own footing. Search Console, filtered to positions 5 to 25. A keyword where you already appear is worth two or three times one where you don't.

The order is the point. People who check difficulty first end up paying for a score on a keyword nobody searches, then check intent last, when intent is the only check that can invalidate the other three.

How do you read intent from a free SERP?

Open the query in a private window so your history doesn't colour the results, then count what type of page holds the top ten.

• Commercial investigation: pricing pages, comparison pages, vendor landing pages. The searcher is choosing, not learning, so you need a comparison or pricing angle to get clicked.
• Informational list: the best X posts and listicles. A well-built list fits, and so does one deep review.
• Community: forums and Reddit threads. The searcher wants a person rather than a vendor, and a brand blog usually loses that fight.
• Reference: docs, definitions and official help pages. Skip it unless you own the entity being asked about.

If eight or more of the top ten are one type, that type is the answer format, and matching it isn't optional. If the ten are split, intent is unsettled, which is the one case where an unusual format can win.

Add one row to the scan in 2026: is there an AI Overview above the results? If there is, and it answers the query in a sentence, you're no longer competing for a click, you're competing for a citation. That changes what the page has to be.

Where free keyword data lies to you

Free data is directionally honest and precisely wrong in four places. Knowing them stops you building a plan on a number that was never a number.

Volume bands. Google Keyword Planner gives exact search volumes to accounts with active ad spend, and a band to everyone else. The same keyword can read as 1K to 10K with no way to tell 1,100 from 9,900. Analytify's Keyword Planner walkthrough, checked in October 2026, documents exactly this:

Google Keyword Planner shows a range when your Google Ads account has no active ad spend. Exact search volume numbers are reserved for advertisers who are actively spending. Analytify, October 2026.

A band is a fine filter and a useless forecast. Use it to decide whether demand exists, never to estimate traffic.

Difficulty scales. MetricHQ's June 2026 benchmark puts keyword difficulty on a 0 to 100 scale, suggests sites with low domain authority target 30 or below, and notes the scores aren't directly comparable across Moz, Ahrefs and Semrush. A 28 from KWFinder and a 28 from Ahrefs are two different claims built on two different link indexes.

Round numbers that never moved. Moz compared six keyword difficulty tools against where the same set of posts actually ranked, in a study published 14 August 2018, and the tools did not agree with each other. Treat a difficulty score as a filter that removes obviously hard terms, not as a prediction of page one.

Your own blind spot. Search Console only reports queries with enough volume behind them, so the long tail is partly invisible. The terms it hides are often the exact ones a small site can win.

Which free tools return real analysis fields?

Most free lists compare lookup counts. What matters for an analysis is which fields each one actually returns.

• Google Keyword Planner returns a demand band and an advertiser competition column. It does not return organic difficulty, and the competition figure describes bidders rather than ranking.
• Google Search Console returns your real positions, impressions and click-through, but only for your own site and only for the queries Google decides to report.
• Google Trends returns relative interest on a 0 to 100 scale. Useful for seasonality, useless for volume.
• KWFinder free returns volume, a difficulty score and related terms on five lookups every 24 hours.
• Semrush's free tier returns intent and difficulty on a few lookups a day, which is enough for a shortlist.
• Bing Webmaster Tools returns Bing volumes plus the top ten pages for a keyword, the cheapest way to see SERP composition without scraping.

No free tool returns all four signals. The signal that decides most outcomes, your own near-miss positions, is free and already sitting in your account.

If the real question is which free plan gives you the most lookups before a cap hits, that's a different comparison and worth reading before you sign up for anything: free keyword search tools, plan by plan.

The scoring sheet, and how to use it in twenty minutes

Score each signal from 0 to 3. Twelve is the maximum, and the total only exists to force a decision you would otherwise postpone.

• Intent, 0 to 3: a 3 if you can match the dominant format, a 0 if the top ten are all something you won't build.
• Demand, 0 to 3: a 3 for a band comfortably above your site's median, a 0 for zero or an unreadable band.
• Difficulty, 0 to 3: a 3 for a score at or under your authority band, a 1 for borderline, a 0 for above it.
• Footing, 0 to 3: a 3 if you already rank 5 to 20 for the term or a close variant, a 2 if another page of yours ranks, a 0 if you have no presence at all.

Write at 8 or above. Hold at 5 to 7 and go looking for a narrower variant. Kill anything under 5. The thresholds matter less than applying them the same way every time, because the whole point is to stop relitigating the same keyword every quarter.

One worked pass on a made-up keyword, to show how the inputs move the answer. The top ten are nine pricing pages, so intent scores 2 because you'll build the table but not the definitive guide. The band is mid range, so demand scores 2. The free difficulty score lands about 12 points above what your winners usually score, so difficulty gets 1. And suppose your HubSpot comparison page already sits at 14 for a close phrase, so footing gets 3. That's 8, and you write it. Flip footing to 0 because the close page targets a different buyer, and you land on 5, which means you hunt for the narrower variant instead.

Is free keyword data good enough to decide what to write?

Yes for the decision, no for the forecast. You can decide whether to invest a day of writing from free data with reasonable confidence, because intent, format and your own footing don't depend on precise volume. You can't forecast clicks, because the one input you would need, an exact volume, is the input free tools refuse to give you.

That split is also why this analysis is a good job for an agent and a poor job for a browser tab. Pulling Search Console rows for positions 5 to 25, filtering by impressions and grouping them by the page that already ranks is a loop over an API, not a series of lookups.

The Search Console API is free, uncapped for your own property and structured, which makes it the only analysis feed you can hand to a script without a paywall appearing halfway through. Everything else in the stack is a manual check you repeat twenty times a month, which is also why free tiers feel generous until you count them.

Once the page is live, the same API tells you whether the analysis was right, and a tracker turns that into a monthly diff. If the tracking layer is still undecided, start with rank tracking software compared by real cost, and the wider buy is covered in the ranking of SEO platforms.

What to do when the score says no

A no isn't the end of the keyword. It's a note about format or scope, and there are four honest moves from there.

1. Change the format to match the SERP. Same keyword, a comparison table instead of a guide.
2. Go one notch narrower, to the variant with the modifier your buyers actually type and weaker pages on the first page.
3. Track it instead of writing it, and revisit once your domain has moved.
4. Kill it. A killed keyword costs twenty minutes. A written one costs a day, plus the internal links you could have spent somewhere else.

The fourth move saves the most and gets used the least, which is the reason this post exists. It's also the reason a cheap tool is worth auditing before you trust its numbers, as the Mangools alternatives breakdown does plan by plan.

Run the pass before you draft, not after

Twenty minutes, in this order: scan the SERP and name the dominant format, pull one demand band, take one difficulty score, then check Search Console for a near-miss position. Score it, write the score next to the keyword, and move on.

The free stack won't tell you how much traffic a keyword will bring. It tells you whether the next four hours of writing have a reason to exist, which is the cheaper question to get wrong.