How to Do Keyword Research for a Niche Site Without Paid Tools
Paid keyword tools sell two things: a list of ideas and a pair of numbers, search volume and difficulty, next to each one. The list you can get for free, from the same place the tools get it. The numbers are estimates, and for a new niche site the free alternative, reading the actual search results, is more accurate about what you can rank for.
This is the method: six free sources, a grid that turns a handful of product nouns into hundreds of keywords, how to judge difficulty by looking at the top ten, and how to turn the result into a list you can publish from for months.
The short answer
List the products and problems in your niche. Run each through Google autocomplete with question words and modifiers, collect People Also Ask and related searches, add Amazon's search suggestions and your competitors' sitemaps.
Then search each keyword and keep the ones where the top ten is beatable: small sites, forums, thin pages, or titles that don't quite match. Sort by intent, cluster by topic, publish.
Start with nouns, not keywords
Every keyword in a niche is a noun with something attached. Best + noun. Noun + for beginners. How to clean + noun.
So the fastest way to a big list is to write the nouns first: every product type, accessory, problem and activity in the niche. A camping niche might produce forty in ten minutes: tents, sleeping bags, stoves, water filters, headlamps, condensation, bears, cold nights.
Then attach the modifiers. A short list of each does most of the work:
- Commercial: best, top, review, vs, alternative, under $50 / $100 / $200, for beginners, for [use], budget, lightweight, small.
- Informational: how to, what is, why does, can you, should I, how long, how often, difference between.
Forty nouns times fifteen modifiers is six hundred candidate keywords. Most won't be searched; the next step finds out which are.
Six free sources
1. Google autocomplete
Type a noun into Google and read the suggestions; those are real queries, ranked by popularity. Then add a letter: tent a, tent b, through to z, and prefix with question words: how tent, why tent, can tent. Each variation surfaces a different set. Use a private window so your history doesn't shape the results.
2. People Also Ask
Search a keyword and expand the "People also ask" box. Each question you open loads more questions, and they're grouped by what Google thinks the searcher wants next. Ten minutes on one seed keyword gives you thirty informational article ideas already phrased as searchers phrase them.
3. Related searches
At the bottom of every results page. These are the queries people run after the one you searched, which makes them the natural next articles in a cluster: the reader of best backpacking tent goes on to search backpacking tent weight and one person vs two person tent.
4. Amazon's search box
For an affiliate site, Amazon's autocomplete is the commercial-intent version of Google's. Type a noun and you get how shoppers describe products: tent for 2 people waterproof, tent with vestibule. Those are the modifiers buyers use, and the category filters on the left of any results page list every product type in the niche.
5. Competitor sitemaps
Every competitor publishes their keyword list as a sitemap; each article URL is a keyword they decided was worth writing. Pull three competitors' sitemaps and you have hundreds of validated ideas plus a view of which clusters they've built out. The full method is in how to find a competitor's keywords from their sitemap.
6. Search Console, once you're live
After a few months, Google Search Console becomes the best keyword tool you have, because it's real data about your site. The Performance report lists every query you appear for, with impressions and average position. Filter for positions 11 to 30: those are keywords Google already thinks you're relevant for, one good article or update away from page one.
Judging difficulty without a score
A paid difficulty score is a guess built from the backlinks of the pages ranking. You can read the same thing more precisely by searching the keyword and looking at the top ten. The signs a new site can rank:
- Small sites in the top ten. If a site with a few hundred pages and no brand ranks there, so can yours.
- Forum threads and Q&A pages. Reddit or Quora on page one means Google couldn't find a good dedicated article. Write it.
- Titles that don't match. If few results contain the phrase in their title, nobody is targeting it directly. Google is ranking approximate answers, and an exact one beats them.
- Old or thin pages. A 2019 article with no updates, or a 400-word page, is holding a spot it can lose.
The signs to move on: a top ten of major brands and big review sites, every title an exact match, and long, current, well-structured pages. Not never, but not in year one. For a new site, the sweet spot is keywords specific enough that the big sites haven't bothered, which is most of the long tail.
Sorting the list
A few hundred keywords that pass the difficulty check need two labels before they're a plan. First, intent: commercial or informational, which decides the article format; the distinction and the ratio to aim for are in informational vs commercial keywords. Second, cluster: the product or topic each one belongs to.
Then the plan writes itself. Each cluster is one commercial page plus its three or four informational articles. Start with the cluster where you found the most beatable keywords, publish it whole, and move to the next. The order of clusters matters more than the order of keywords within them.
What you give up, and what you don't
Without a paid tool you have no volume number. That matters less than it seems: for a new site the question isn't "how many people search this" but "can I rank for it", and a keyword Google autocompletes has enough demand to be worth an article at long-tail scale. Once you're live, Search Console gives you real impressions, which no estimate matches.
What you don't give up is anything that decides whether the site works. The ideas, the intent, the clusters and the difficulty judgment are all here, and they're the parts a paid tool can't do for you anyway.
Doing it faster with AffSERP
Two of the tools in AffSERP exist to shorten this. The Keyword Explorer runs a seed through Google's autocomplete, with the a-to-z and question-word expansions above, for your chosen country and language, and returns the list grouped and exportable, with a Write button that sends any keyword to the article tool.
The Keyword Grouping tool takes a pasted list, from any of the six sources, and clusters it by intent with a suggested pillar article per cluster.
The difficulty judgment stays with you; it's the part that needs a human looking at the results. Everything around it, the gathering, the sorting and then the writing, doesn't.
FAQ
Can you do keyword research without paid tools?
Yes. Autocomplete, People Also Ask, related searches, Amazon's search box, competitor sitemaps and Search Console produce more ideas than a new site can write. The paid extras are volume and difficulty estimates, both replaceable by reading the results.
How do I know the volume?
You don't get a number. Autocomplete and People Also Ask show demand exists; Search Console shows real impressions once you're live.
How do I judge difficulty?
Search it. Small sites, forums, thin or old pages and non-matching titles in the top ten mean you can rank; a wall of big brands with exact-match titles means not yet.
How many keywords do I need?
Fifty to a hundred for the first three to six months, in clusters of one commercial keyword and three or four informational ones.
Keyword Explorer and Keyword Grouping gather and sort; the writing tools turn each keyword into a published article with your affiliate links.
Start with free wordsKeep reading
