Keyword Research for a Website: What It Is, How to Do It and What It Costs
A keyword map is the list of search queries your customers use to find what you sell, assigned to specific pages of your site. Everything else depends on it: site structure, copy, meta tags and even which pages are worth creating at all. Below - how to build one in 7 steps, how many keywords different types of sites need, what it costs and the mistakes that make finished keyword research useless.
What keyword research actually produces
Say you sell kitchen range hoods. One person searches for "range hood for kitchen", another for "buy built-in range hood 60 cm", a third for "how to choose a range hood for a gas stove". These are three different people at different stages of buying, and each should land on a different page: a category, a filtered listing or an article.
Proper keyword research gives you the full set of such queries for your niche - cleaned of junk, grouped by meaning and tied to specific pages. Not a 10,000-row spreadsheet nobody opens, but a working map: which page is responsible for which queries.
The core rule. If two pages target the same queries, they compete with each other and both rank worse. If a group of queries has no page, you simply don't capture that demand.
Why bother, if you can just write content
- Site structure. Keyword research shows which sections, categories and filters you need. This is often where you find that customers search for "service + city" or "product + feature" and the site has no page for it;
- Copy and meta tags. The writer gets a specific keyword group for each page instead of "write something about range hoods". Title, H1 and text are written for real demand;
- Priorities. Search volume and competition show where to start to get traffic sooner and what can wait;
- Forecast. Total search volume shows how much traffic the niche can bring at all and whether it's worth investing in.
Types of search queries
| Query type | Example | Where to send it |
|---|---|---|
| Commercial | "buy range hood 60 cm" | Category, product page, service page |
| Informational | "how to choose a kitchen range hood" | Blog article, guide |
| Branded | "ventolux range hood" | Home page, brand page |
| Local | "iPhone repair near me" | Service page with address, Google Business Profile |
Queries are also split by volume into high, medium and low (long-tail). The boundaries depend on the niche: in phone repair a "high-volume" query means thousands of searches a month, in industrial equipment - a few dozen. A new site usually does better starting with medium and long-tail queries: they are more specific, less competitive and often convert better.
How to do keyword research: 7 steps
Seed keywords
List the basic names of everything on the site: product categories, services, brands, locations. For a range hood store that's "range hood", "built-in range hood", "island range hood", "Ventolux range hood" and so on. The site menu, the price list and a talk with the owner help here: how do customers actually name the product? Often not the way the catalog does.
Expansion
Each seed is expanded into hundreds or thousands of variations. The sources I use:
- Ahrefs and Semrush - keyword databases with volume and difficulty;
- SE Ranking - convenient for smaller markets and rank tracking;
- Google Keyword Planner - volume straight from Google, especially useful for commercial queries;
- Google Search Console - queries the site already gets impressions for. For a live site this is the most valuable source;
- Google autocomplete and "People also ask" - real phrasings and ready-made questions for articles.
Competitor keywords
Take 5-10 sites ranking in the top for your seed keywords and export the queries they rank for and you don't (Content Gap in Ahrefs, Keyword Gap in Semrush). It's the fastest way to find whole topics you hadn't thought of.
Cleanup
After cleanup, usually only a smaller part of the raw list remains. Remove:
- irrelevant queries: "free", "DIY", "used", "jobs" - unless you sell that;
- brands you don't carry;
- other regions: if you only serve one city, queries with other cities are junk;
- duplicates, including the same words in a different order;
- zero-volume queries, except clearly commercial ones.
Clustering
Group queries by search results: if Google shows the same sites in the top 10 for two queries, one page can rank for both. If 3-4 or more URLs overlap, it's one cluster. If the results differ, you need separate pages, even when the queries look similar. Clustering by SERP overlap, not by eye, is what turns a word list into a working keyword map.
Mapping to pages
Each cluster is assigned to an existing page or marked as "page to be created". The result is a keyword map: URL, primary keyword, secondary keywords, total volume. This is what the developer (for new pages) and the copywriter (for copy and meta tags) work from.
Prioritization
You can't do everything at once, so sort by what brings results fastest. First in line are usually commercial clusters with moderate competition where the site already sits in positions 11-30: they are the easiest to push into the top 10. Next come new pages for uncovered demand, then informational content.
Need keyword research for your site?
Send me your website - I'll look at the niche and tell you roughly how many queries it has, which pages are missing and what the research will cost.
Message me on TelegramMultilingual site: one keyword set or several
Separate research for every language and every market. People search differently in different languages: other phrasings, other volumes, sometimes different results. Translating a finished keyword list doesn't work - part of the translated queries simply don't exist, and the real phrasings get lost. In Ukraine, for example, almost every site needs separate Ukrainian and Russian keyword sets, even for the same audience.
The same applies when you expand abroad: each country gets its own research with its own local search volumes. How I did this for a site operating in 40+ countries - in the article on international SEO.
How many keywords you need
It depends not on wanting "more", but on how many pages the site has and how much demand the niche has. Benchmarks from my practice:
| Site type | Keywords after cleanup |
|---|---|
| Landing page, single service | 50-150 keywords |
| Service website (10-30 pages) | 300-1,500 keywords |
| Corporate site, large service catalog | 1,500-5,000 keywords |
| Online store | from 3,000 to tens of thousands of keywords |
More isn't better. A 20,000-keyword list nobody mapped to pages is less useful than 500 keywords that already have pages and copy behind them.
How much keyword research costs
| Scope | Price | What's included |
|---|---|---|
| Landing page or small service site | from $100 | Collection, cleanup, clustering, keyword map |
| Service or corporate website | from $200 | The same + competitor analysis and structure recommendations |
| Online store | from $350 | The same + keywords for categories, filters and product pages |
Timing: 3-7 business days for a service site, 2-4 weeks for an online store. With ongoing SEO from $450/month, keyword research and its regular expansion are already part of the work - no separate fee. These are my rates as a Ukraine-based specialist; I bill in USD or EUR. More on budgets - in the article how much SEO costs.
Five mistakes that make keyword research useless
- Collected but never mapped. A spreadsheet with thousands of rows sits in a folder, and nobody created pages for it. Research without a keyword map is wasted budget;
- Grouped by eye. Similar-looking queries ended up on one page, although Google shows different sites for them. As a result the page ranks for none of them;
- Two pages for one cluster. A category and a blog post compete for the same query. That's cannibalization: both rank lower than they could;
- No cleanup. "Free", "DIY" and other brands stayed in the list, copy gets written for them and brings irrelevant traffic;
- Done once and forgotten. Demand changes: new products, phrasings and questions appear. Expand the research at least once a quarter - the easiest way is through new queries in Search Console.
Checklist: 10 signs your keyword research is done right
- Seed keywords cover all categories and services
- Keywords are expanded from several sources, not one tool
- Keywords of 5-10 top-ranking competitors are analyzed
- Irrelevant queries, other brands, other regions and duplicates are removed
- Keywords are grouped by top-10 SERP overlap
- Each cluster is assigned to exactly one page
- Pages that need to be created are marked
- A multilingual site has separate research for each language
- Priorities are set: what we start with
- Regular updates are scheduled
When to do it and who should do it
Ideally before launch or before a redesign: then the structure can be built around demand instead of being patched later. More on that in the article on SEO for a new website.
Basic research for a small site can be done in-house following this plan. But for an online store or a competitive niche a clustering mistake costs months: pages built for the wrong groups will have to be redone. If you already have keyword research and no results, start with an SEO audit: often the problem isn't the keywords but how they are mapped.
I'll do keyword research for your niche
Describe your site and goal - I'll suggest the scope, timing and price. The first consultation is free.
Send a request