Keyword research: a step-by-step method with SEOZoom, Ahrefs and Search Console

The marketing manager of a company that makes biomass boilers sends us an Excel file. It contains 1,400 keywords exported from an SEO tool, sorted by volume. At the top is “boiler”, with tens of thousands of searches a month. Further down, among the rows with 30 or 40 searches, are “wood chip boiler for farms” and “Conto Termico biomass boiler”. The question is: “Where do we start?”. The file doesn’t say, because a list of keywords isn’t yet research.

The short answer: keyword research is for deciding which pages to write and in what order, not for compiling long lists. The method we use has six steps: seed keywords taken from customers’ language, expansion with SEOZoom and Ahrefs, comparison with real Search Console data, clean-up, grouping by intent and prioritisation on three criteria, namely volume, difficulty and value to the business. At the end, each group of searches corresponds to a page, either new or to be rewritten. In the boiler example, the two searches with 30 and 40 a month were probably the first ones to cover.

We work mainly on the Italian market, so the example searches in this article are Italian ones, translated into English.

This article is about the method. What search intent is and how to recognise it is explained in our article on search intent and its four types: here we take it as read and use it as a tool.

What is keyword research and what should it produce?

It’s the work of finding out how people search on Google for what a company sells or knows how to do, and turning that into a plan of pages. The end point isn’t a list: it’s a table in which each group of searches has its intent, the page that covers it (existing or to be created), its priority and who’s responsible for it.

Good keyword research answers four questions:

1. How do people search for us? The words customers use, which often don’t match the ones in the catalogue.
2. What do they want to find? An explanation, a comparison, a price, a supplier.
3. What do we already have? Which pages on the site already get impressions for those searches, and in what position.
4. Where do we start? Which groups are worth most to the business relative to the effort they take.

The first two questions are answered with market tools, the third with Search Console, the fourth with a conversation with the people who sell. Skipping the last one is the most common mistake: you end up writing for the biggest searches, which are also the most contested and often the least useful.

Which tools do you need for keyword research?

Three are enough, plus a free one for checks. Each answers a different question, and it’s worth knowing which before comparing the numbers.

Tool	What it’s for	Where the data comes from	Cost
Google Search Console	The searches your site already appears for, with clicks, impressions and position	Google’s data on your site	Free
SEOZoom	Volumes, competitors and topics in the Italian market, intent analysis	Estimates based on a database of Italian keywords	Paid
Ahrefs	Difficulty, traffic potential, parent topic, searches in other languages	Estimates based on an international database	Paid; limited free version for verified sites
Google Ads Keyword Planner	Ideas and trends, useful as a second opinion on volumes	Google Ads data	Free with a Google Ads account

We use SEOZoom as our reference for Italy, because its database is built on Italian searches. Ahrefs is for difficulty, traffic potential and cases where the client also sells abroad. Search Console is the only one of the three that reports real data about your site: for pages you already have, when the numbers don’t match, Search Console wins. It’s the same rule we follow in an SEO audit.

Google’s Keyword Planner was built for advertisers. It’s useful for ideas and for seeing seasonality, but for accounts that spend little it often shows volumes in ranges (for example 100-1,000) rather than a precise number. For SEO we use it as a check, not as the main source.

Step 1: which seed keywords do you start from?

Seed keywords are the starting terms, the ones from which the tools generate variants. If they’re wrong, everything that follows will be tidy but useless.

We take them from four sources, in this order:

• The people who sell. Half an hour with a salesperson or whoever answers the phone is worth more than any database. We ask: what questions do customers ask before requesting a quote? What words do they use to describe the problem?
• Emails, website enquiries and chat. The enquiries that have come through the contact form in recent months are an archive of real language.
• Search Console. The queries already bringing impressions to the site, even in low positions.
• Catalogue and price list. Products, services, models, applications, industries served.

The result is a list of 15-40 seed keywords, no more. For a biomass boiler manufacturer they might be “pellet boiler”, “wood chip boiler”, “wood-burning boiler”, “biomass district heating”, “biomass boiler incentives”, “pellet boiler maintenance”. These are examples made up for illustration, but they show one thing: half come from the catalogue, the other half from customers’ questions.

One detail that makes a difference: internal jargon. In many companies people talk about “heat generators” or “dosing systems”, while customers search for “boiler” and “doser”. Seed keywords should be written the way customers write them.

Step 2: how do you expand the list with SEOZoom and Ahrefs?

From the seed keywords you generate the variants: longer searches, questions, synonyms, combinations with places and industries. Here the tools do most of the work.

With SEOZoom

In its keyword research section, SEOZoom suggests related searches, questions and topics sharing the same intent for each seed keyword. For the Italian market it’s the most reliable starting point on volumes. There are two features we always use:

• Competitors’ keywords. By entering the domain of two or three direct competitors you can see which searches they rank for. It’s the quickest way to discover topics you hadn’t thought of.
• Domain comparison. It shows the searches where competitors appear and you don’t. These are the first candidates for new pages.

With Ahrefs

Ahrefs adds two metrics that help you avoid the wrong choice. Traffic potential is the estimated organic traffic received by the top-ranking page from all the searches it ranks for, not just the main one. Parent Topic indicates the search that brings that page the most traffic. The definitions are in Ahrefs’ guide to its metrics.

They help you understand one thing: if “pellet boiler prices” and “how much does a pellet boiler cost” share the same parent topic, Google satisfies them with the same page. Two separate articles would compete with each other.

How many keywords do you collect?

It depends on the site. For a company with ten or so services or product families, the raw list often reaches a few thousand rows. That’s not a problem: in step 4 it shrinks dramatically. At this stage it’s better to collect a lot and filter later.

Step 3: what does Search Console data add?

Market tools estimate how often a term is searched for. Search Console tells you what your site actually appears for, with how many clicks and in what position. It’s the step that turns generic research into research about your site.

In the Performance report we look at four things:

1. Queries with lots of impressions and few clicks. If a page appears often but is rarely clicked, the problem may be the title, the position or an intent different from the one the page covers.
2. Queries in positions 8 to 20. These are searches for which Google already considers the site relevant, but not relevant enough. Often improving an existing page is enough, rather than writing a new one.
3. Several pages for the same query. By filtering for a query and opening the Pages tab you can see whether Google is alternating between two URLs. That’s a sign of cannibalisation to resolve before writing anything else.
4. Branded and non-branded searches. The branded queries filter, introduced by Google in November 2025, separates people already searching for your name from people searching for the problem. For keyword research, it’s mainly the second group that counts.

For sites with lots of queries, Search Console Insights also shows AI-generated query groups: it gathers the variants of the same question and shows which topics are growing or declining. Google points out that they’re only available for properties with a high volume of queries.

The limitations to know about

Search Console keeps 16 months of data. The table in the interface shows a maximum of 1,000 rows: for larger sites it’s better to filter by folder or export via the API. Some queries don’t appear for privacy reasons, so the sum of the rows doesn’t match the total in the chart; Google explains this in its deep dive into filters and data limitations. These are known limitations, not errors.

Step 4: how do you clean up the list?

This is where a list of a few thousand rows comes down to a few hundred. The exclusions are almost always the same:

• Off-target searches. People searching for “used pellet boiler” or “DIY pellet boiler” aren’t customers of a manufacturer. They come out, unless there’s a specific reason to keep them.
• Competitors’ brand names. They come out of the main list. If it makes sense to create content around them, that’s assessed separately, with care.
• Duplicates and minor variants. Singular and plural, the same words in a different order, typos. Keep the form with the most volume and note the others.
• Areas you don’t serve. A search including the name of a city where you don’t work brings visits that don’t turn into leads.
• Zero volumes, or volumes estimated as such. They don’t always need removing: for a B2B company, a search with 10 searches a month can be worth a contract. Keep those with a clear commercial value.

The rule: every row that remains must be able, sooner or later, to become a lead, a sale or a useful answer for a customer. If you can’t imagine who’s typing it, it comes out.

Step 5: how do you group keywords by intent?

This is the step that turns the list into a plan. Each group gathers the searches answered by the same page, and each page covers a single group.

The criterion we use is practical: two searches go in the same group if Google shows largely the same results on page one. If the top ten positions for “wood chip boiler price” and “how much does a wood chip boiler cost” overlap substantially, the two searches want the same page. If the results are different, you need two pages.

Then each group is assigned an intent and a page type. The four types of intent are explained in our glossary entry on search intent; what matters here is how they translate into pages.

Group (example)	Searches included	Intent	Page type
Cost of a wood chip boiler	wood chip boiler price, how much does a wood chip boiler cost, cost of a wood chip system	Commercial	Cost guide with price bands and factors
Wood chip boilers for farms	wood chip boiler for farms, greenhouse heating with wood chips	Commercial/transactional	Industry solution page
Incentives	Conto Termico biomass boiler, tax deductions for pellet boilers	Informational	Article kept up to date with official sources
Maintenance	cleaning a pellet boiler, how often to clean a pellet boiler	Informational	Practical guide
Local servicing	biomass boiler servicing Cuneo	Local transactional	Service page with area

The table is made up for illustration (Conto Termico is an Italian incentive scheme for renewable heating). But it shows the two decisions that matter: how many pages are needed, and of what type. A classic mistake is writing an article for a group whose page one shows only product pages, or vice versa.

What if a page already exists?

Before creating a new page, check in Search Console whether an existing page already gets impressions for that group. If it does, it’s almost always better to improve it than to put a duplicate next to it. How to write or rewrite the copy once the group is chosen is explained by Giada in her article on SEO copywriting.

Step 6: how do you decide priorities?

With three criteria together, never just one: volume, difficulty and value to the business.

• Volume. How many searches the group gets, adding up the variants. For B2B sites it’s also worth looking at Ahrefs’ traffic potential, which gives a more realistic idea of possible visits.
• Difficulty. How hard it is to get onto page one. Ahrefs’ Keyword Difficulty runs from 0 to 100 and is based mainly on the domains linking to the pages already ranking at the top. SEOZoom has its own indicators. These are estimates: always compare them with a look at page one. If it’s all big portals and marketplaces, it’ll be tough even with a stated medium difficulty.
• Value. How close the search is to a lead or a sale, and how much that customer is worth. No tool gives you this figure: the people who sell do.

A simple matrix

We give each group a score from 1 to 3 on each criterion, with difficulty reversed (3 = easy). We multiply the three scores and sort. The numbers below are made up for illustration.

Group	Volume	Ease	Value	Score
Wood chip boilers for farms	1	3	3	9
Cost of a wood chip boiler	2	2	3	12
Incentives	3	1	2	6
Maintenance	3	2	1	6
Boiler (generic)	3	1	1	3

The keyword with the most volume, “boiler”, ends up at the bottom. Not because it doesn’t matter, but because today it costs too much relative to what it brings in. The score doesn’t decide on its own: it’s there to put the discussion in order. If the salesperson says the season for farms is January, that page moves up even if its score is lower.

Quick wins and long projects

In the final list we separate two columns. Quick wins are groups for which the site already has pages in positions 8-20: improving them brings results sooner. Long projects are new, contested topics that require new pages, links and months. A healthy plan has both.

What changes with AI Overviews and answer engines?

The method stays the same, but the way you read the data changes. For many informational searches, Google shows an AI-generated overview before the results, and some searchers stop there. A question’s volume no longer translates into clicks in the proportions of a few years ago.

Three adjustments we make:

• We note whether the search shows an AI Overview. You can see it by looking at page one. For purely informational questions we expect fewer clicks, and we say so up front, not afterwards.
• We give more weight to commercial and local searches. People comparing suppliers or looking for someone nearby still tend to click.
• We write answers so they can be cited. A question with a clear answer at the start of a section helps both Google and answer engines. It’s the work we describe on our page about GEO.

Searches in answer engines such as ChatGPT are longer and more conversational, and there’s currently no reliable volume data. The questions customers ask the salesperson remain the best source for understanding how they phrase them.

How often should keyword research be redone?

Full research is done at the start of a project, before a website rebuild, and when opening a new market or product line. After that, a quarterly update is enough: look in Search Console for new queries with impressions, check which groups are growing or declining, and update the map.

For companies with strong seasonality (heating, tourism, gifts) it’s worth doing the update two or three months before the season, because new pages need time to be crawled and ranked.

Mistakes to avoid

Sorting everything by volume. The biggest searches are also the most contested and often the most generic. For an SME, small, precise groups bring in customers sooner.

Starting from the catalogue rather than the questions. Customers search for the problem, not the product name. Half an hour with the people who sell saves weeks of content written with the wrong words.

One page per keyword. If Google shows the same results for two searches, one page is enough. Two similar pages compete with each other and neither climbs.

Ignoring Search Console. Market tools estimate; Search Console tells you what happens on your site. Often the page to improve already exists.

Treating the numbers as exact. Volumes and difficulty are estimates, and different tools give different numbers. They’re for comparing groups with each other, not for forecasting traffic.

Stopping at the list. Keyword research without page type, priority and owner remains an Excel file.

Keyword research checklist

• Seed keywords collected from the people who sell, website enquiries, Search Console and the catalogue.
• List expanded with SEOZoom (Italian market, competitors) and Ahrefs (difficulty, traffic potential, parent topic).
• Search Console queries cross-checked against the list: impressions, positions 8-20, alternating pages, branded and non-branded.
• List cleaned of off-target searches, competitors’ brands, duplicates and areas not served.
• Groups formed by comparing Google’s page one; one group, one page.
• For each group: intent, page type, existing page or one to create.
• Priorities decided on volume, difficulty and value, with input from the people who sell.
• Quick wins separated from long projects.
• Quarterly update in the calendar.

Want a map of the searches that bring in customers?

Keyword research is one of the first steps in our SEO work: we start from Search Console and your competitors, talk to the people who sell and deliver a map in which every group of searches has its page and its priority. If you’re wondering what to expect from whoever supports you, you’ll find the questions to ask in our guide to SEO consultancy.

You can book a call with Federico or write to us via the contact form.

Frequently asked questions

How long does keyword research take?

For a site with ten or so services or product families it usually takes a few days of work, including the conversation with the people who sell. An e-commerce site with many categories or a multilingual site takes longer, because there are many more groups to check.

Can you do keyword research with free tools?

Partly, yes. Search Console, the Google Ads Keyword Planner and Google Trends give you real data about your site, ideas and seasonality. What’s missing is competitor data and a reliable estimate of difficulty, which today come almost only from paid tools such as SEOZoom and Ahrefs.

SEOZoom or Ahrefs: which is better for keyword research?

They aren’t alternatives. For the Italian market we use SEOZoom as our reference on volumes and competitors; Ahrefs adds traffic potential, parent topic and an international database that’s useful if you sell abroad. If you have to choose one and work only in Italy, SEOZoom covers most of the work.

Should zero-volume keywords be discarded?

Not always. Tools are poor at estimating very specific searches, and in B2B a search made only a handful of times a month can come from exactly the right person. If the search describes exactly what you sell, it’s worth covering, perhaps within a page that answers a broader group.

What’s the difference between keyword research and search intent?

Keyword research is the complete process: finding the searches, cleaning them up, grouping them and deciding priorities. Search intent is one of the criteria it uses: for each group it indicates what the searcher wants to find and therefore what type of page is needed.