Keyword research without Ahrefs or Semrush
Paid keyword tools sell you two things: a volume number and a difficulty score. Both are estimates. Here is how to get better answers for free, using sources that are closer to the truth than either.
A subscription to a big SEO suite starts around $99 a month. For an established agency that is nothing. For one person with a site making $40 a month in ads, it is the whole business.
The good news is that the two headline numbers you are paying for are the two least reliable things in the product.
What you are actually buying
Search volume in every tool traces back to Google Keyword Planner data, which is bucketed, averaged over twelve months and rounded. Two tools will confidently show you 1,900 and 3,600 for the same keyword. Neither is wrong, exactly; both are modelling.
Keyword difficulty is a made-up score. Each tool invents its own, mostly from the backlink counts of the current top 10 in that tool's own index. That is why the same keyword is "34" in one tool and "61" in another. It is a useful rough sort, not a measurement.
What the paid tools genuinely give you that is hard to replace is scale — a hundred thousand keyword ideas and a competitor's whole ranking profile in one click. If you are choosing between ten thousand keywords, pay for the tool. If you are deciding what to write next on a 60-page site, you do not need it.
Where to get keyword ideas free
In rough order of how useful they are for a small site:
1. Your own Search Console
This is the best keyword tool in the world for your site, and it is free. Performance → Search results → Queries shows every search that already showed your site, with real impressions and your real position. Two goldmines in there:
- Queries with impressions and zero clicks. Demand you are visible for and not capturing.
- Queries you rank for by accident. Things you never wrote about but Google associates you with. Each one is a page waiting to be written by someone Google already half-trusts on the topic.
2. Google autocomplete, mined properly
Type your seed and read the suggestions — then do it again with a letter appended ("seo a", "seo b"…), and with question words in front ("how to seo", "why seo"). Autocomplete reflects what people actually type, which is exactly the thing volume estimates are trying to approximate.
3. The results page itself
Search your seed and harvest three blocks: People Also Ask (expand a few — it keeps generating), Related searches at the bottom, and the autocomplete for each of those. Ten minutes gives you a hundred real queries.
4. Reddit, Quora and niche forums
Where people ask in their own words, before they learn the jargon. Search site:reddit.com your topic and read the question titles. These make excellent long-tail targets because the phrasing is human and the competition is usually a thread, not an optimised page.
5. Your competitors' subheadings
Open the three pages ranking for a term you care about and list their H2s. Each is a sub-topic someone decided was worth covering, and several will be standalone posts.
6. Google Trends
Free and genuinely good for the one thing estimates are worst at: direction. A keyword with modest volume that has doubled in six months beats a bigger one that has been sliding for two years.
Judging volume without a volume number
You rarely need the absolute number. You need to know whether a keyword is worth a day's work, and you can get that three ways:
- Relative volume from your own data. If a similar keyword on your site gets 2,000 impressions a quarter at position 9, a comparable one will land in the same order of magnitude. Your own site is the best calibration you have.
- Autocomplete position. The order of suggestions correlates with popularity. The first suggestion is searched a lot more than the eighth.
- Google Trends comparison. You cannot get an absolute number, but you can compare two keywords against each other, and that is usually the actual decision.
Stop chasing volume anyway. On a small site, ten keywords at 200 searches a month that you can rank #2 for beat one keyword at 10,000 you will never crack. The first set is roughly 400 clicks a month. The second is zero.
Judging difficulty from the results page
This is where free beats paid, because you are looking at the actual result set rather than a score derived from someone's crawl of it. Search the keyword in an incognito window and score what you see:
| What you see on page one | Read |
|---|---|
| Wikipedia, Amazon, YouTube, government or big-brand sites filling the page | Skip it |
| Every result is a huge site in your niche | Hard — park it for later |
| At least two results from sites your size | Winnable |
| Reddit or Quora threads ranking on page one | Very winnable — nobody has written the proper page yet |
| Results that do not quite answer the query | Best case — demand with no good supply |
| Mostly ads and shopping results | Commercial; organic clicks will be thin |
Then look closer at the weakest result on page one — the one at position 9 or 10. That is your real competition, because that is the slot you are taking first. If you can clearly write something better than that page, the keyword is worth doing.
The check that matters more than either
Before writing anything, answer: what kind of page is Google rewarding for this query?
Look at what the top 10 actually are. All blog posts? A list of tools? Product pages? A forum thread? Videos?
If the results are product pages and you write a blog post, you will not rank — not because your post is bad, but because you answered a different question than the one being asked. This single check saves more wasted work than any difficulty score, and it costs one search.
A workflow you can run weekly
- Export Search Console (last 3 months, Queries). Ten minutes, once a month is enough.
- Pull out the striking-distance rows — position 4 to 20, decent impressions. These are existing pages to improve, and they always come before new ones.
- Pull out the zero-click, high-impression rows. Either the snippet is failing or the page does not really cover that query.
- Take one seed keyword for something you want to own and mine autocomplete, People Also Ask and related searches for fifteen minutes.
- Search your top three candidates and score page one with the table above. Keep the winnable ones.
- Check intent on the survivor, and write that page type.
- Write it down. The decision and the date. In a month, compare against Search Console and you will start to learn which of your judgement calls are actually right.
That last step is the one people skip, and it is the one that compounds. Paid tools give you other people's estimates about your market. Running this loop for six months gives you real data about your own — which is the thing no subscription sells.
I run a content site, got tired of paying four subscriptions to find out what to write, and built this instead. Everything here is tested on my own traffic first.
Want this done for you?
Fix My SERP does everything in this guide automatically — finds the pages, ranks them by what you would gain, and gives you the list in order. Free, and it runs on your own hosting.
Start free What it doesRead next
A free SEO audit checklist you can run in one afternoon
Paid audits hand you 600 issues sorted by a severity score somebody else invented. This one has 24 checks, in the order that actually affects traffic, and every tool it uses is free.
Should you self-host your SEO tools? An honest answer
Running your own SEO tooling on cheap hosting sounds like the frugal, principled choice. I built one and ran it for months. Here is what it actually gains you, what it costs you, and why I moved mine to a server I maintain.
Google Site Kit: what it is actually good for (and what it is not)
Site Kit connects Search Console, Analytics and AdSense to WordPress with one sign-in. That is genuinely useful. The dashboard it puts in your admin is mostly not. Here is how to set it up and what to ignore.