Keyword research is the process of finding the words and questions your buyers use, working out what they actually want when they search, and deciding which page on your site should answer each need. Good research ends in a keyword-to-page map with a clear intent for every URL, not a long spreadsheet of search volumes.
Most keyword lists fail for a simple reason: they are sorted by volume and difficulty, but nobody has asked what the searcher is trying to get done, whether your business can genuinely help, or which existing page should carry the topic. This guide walks through a practical framework, from collecting real buyer language to grouping queries by intent, choosing page types and checking your work in Search Console. It is written for in-house marketers, founders and SEO leads who need a plan they can defend, not a tool export.
Key takeaways
- Intent beats volume. A low-volume query from a buyer comparing providers can be worth more than a high-volume definition query that never converts.
- Start with your own data. Sales calls, support tickets, site search and Search Console queries reveal language that keyword tools often miss.
- Group, then map. Cluster queries that share the same intent and assign each cluster to one page, so pages do not compete with each other.
- Volume figures are estimates. Tools such as Keyword Planner describe their numbers as estimates; treat them as relative signals, not forecasts of traffic.
- Research is never finished. Review the map every quarter as new queries appear in Search Console and as your offer changes.
What is search intent, and why does it decide everything?
Search intent is the goal behind a query. Two searches can share most of their words and still want completely different things. Someone typing “CRM implementation” may want a definition, a step-by-step plan, a list of vendors or a partner who will do it for them. If you send all four people to the same service page, three of them leave.
The classic four labels are still a useful starting point:
- Informational: the searcher wants to understand something (“what is technical SEO”, “how does lead scoring work”).
- Commercial investigation: they are comparing options before choosing (“HubSpot vs Zoho for small teams”, “best WooCommerce hosting India”).
- Transactional: they are ready to act (“hire SEO consultant Delhi”, “book website audit”).
- Navigational: they want a specific brand or page (“Bargaon contact”, “Search Console login”).
In practice these labels are too blunt for planning content. A more useful question is: what job does the searcher need this page to do? Explain a concept, help them choose, show them how, prove you can do it, or let them get in touch. The job tells you the page type, the depth, the evidence needed and the next step to offer.
How to read intent from the results page
The quickest test is to search the query yourself, in a private window, from the location your buyers are in. Look at what Google already rewards:
- If the first results are guides and definitions, the intent is mainly informational.
- If they are comparison articles, review sites and “best of” lists, the searcher is still choosing.
- If they are service pages, local listings and ads, the searcher is close to buying.
- If you see AI Overviews, People Also Ask boxes, videos or product carousels, note them. They tell you which formats Google thinks satisfy the query.
Do this for every important cluster. It takes minutes and prevents the most expensive mistake in SEO: writing the wrong kind of page for a query.
Where should keyword ideas come from?
Keyword tools are useful, but they only show what many people type. Your best opportunities often come from what your buyers type, which is visible in your own data first.
| Source | What it gives you | Strength | Limit |
|---|---|---|---|
| Google Search Console | Queries that already show your pages, with clicks, impressions, CTR and average position | Real data for your site; reveals page-2 opportunities | Only shows queries where you already appear; some queries are hidden |
| Sales and discovery calls | The exact words buyers use to describe problems, objections and comparisons | Closest to commercial intent | Small sample; needs someone to record and tag it |
| Support tickets and site search | Questions existing customers and visitors cannot answer on the site | Shows content gaps on pages that already get traffic | Skews towards existing customers |
| Keyword Planner and other tools | Related terms and estimated monthly searches | Broad discovery and relative demand | Figures are estimates; low-volume B2B terms often show little or no data |
| People Also Ask and AI Overviews | Follow-up questions and the angle Google expects | Good for FAQs and answer sections | Changes often; varies by location and device |
| Forums, Reddit, LinkedIn comments | Unfiltered language, frustrations and edge cases | Finds problems competitors ignore | Noisy; not always your buyer |
Google’s Search Console help on the Performance report explains the core metrics you will use: clicks, impressions, click-through rate and the average position of your topmost result. Google Ads’ Keyword Planner documentation describes its monthly search figures as estimates. That word matters. Use tool volumes to compare topics with each other, not to promise a number of visitors.
The Bargaon intent map: a five-step framework
We use the same five steps on every keyword project. The output is a single sheet that sales, content and development can all read.
From buyer language to a page plan
Each step narrows the list. A good map usually ends with far fewer pages than the raw keyword count suggests.
Step 1: Collect real buyer language
Before opening a keyword tool, write down 20 to 40 phrases taken directly from buyers: questions asked on sales calls, problems described in enquiry forms, objections raised in proposals, and terms used in tender documents. Ask sales which questions come up before every deal. These phrases are your seed list. They anchor the research in real demand and stop it drifting into topics that attract the wrong audience.
Step 2: Expand with tools
Put the seeds into Keyword Planner and your preferred SEO tool. Export related terms, questions and modifiers (“cost”, “vs”, “for small business”, “India”, “template”, “checklist”). Then open Search Console and filter queries containing your seed terms over the last 16 months if available. Pay special attention to queries where you rank between positions 8 and 20 with healthy impressions. These are pages Google already associates with the topic, and small improvements can move them onto page one.
Step 3: Group by shared intent
Group queries that a single, well-written page could satisfy. The test is simple: if the same page would genuinely be the best answer for both queries, they belong together. “How to do keyword research” and “keyword research process” belong in one group. “Keyword research tools” usually belongs elsewhere, because the searcher wants a comparison of tools, not a process. Checking whether the same URLs rank for both queries in the results is a quick, reliable way to confirm a grouping.
Step 4: Assign one page per group
Every group gets exactly one primary page, and every page gets a defined job. Existing pages come first: if a page already ranks for some of the group, improve it rather than creating a competitor. New pages are only created where no existing page can sensibly cover the intent. This is the step that prevents keyword cannibalisation, where two of your own pages compete for the same query and both perform worse.
Step 5: Review in Search Console
After publishing or updating, check whether the page is now receiving impressions for the queries in its group, and whether unexpected queries appear. Unexpected queries are not a failure. They are new research. Add them to the map, decide whether the page should cover them, or whether a different page is a better home.
Which page type matches which intent?
Once intent is clear, choosing the page type becomes straightforward. This table is the decision rule we apply when filling in the map.
| Searcher’s job | Example query | Best page type | Next step to offer |
|---|---|---|---|
| Understand a concept | what is answer engine optimization | Guide or glossary entry | Related guide, newsletter, free tool |
| Learn how to do something | how to plan a website migration | Step-by-step guide or checklist | Downloadable checklist, audit offer |
| Compare options | HubSpot vs Zoho CRM small business | Honest comparison guide | Service page, consultation |
| Check cost or fit | SEO audit cost for small business | Service page with pricing logic or FAQ | Enquiry form, booking link |
| Hire someone | technical SEO consultant India | Service page (L2 or L3) | Enquiry form above the fold |
| Use a calculator or check | meta description length checker | Free tool page | Related guide, service page |
Notice that the commercial queries map to service pages, not blog posts. A common mistake is to write an article for “technical SEO services” and hope it converts. Searchers with that intent want to evaluate a provider, so they need scope, process, proof and a way to get in touch.
How do you prioritise a keyword list?
A finished map can still contain a hundred groups. You cannot work on all of them at once, so score each group on four questions and work from the top down:
- Business value: if this page ranked and attracted the right visitor, how close would that visitor be to an enquiry or sale? Score commercial and comparison groups higher.
- Ability to answer: do you have genuine experience, data or examples that make your page better than what already ranks? If not, the page is unlikely to win and may not deserve to.
- Current position: is there already a page with impressions for this group? Improving a page at position 12 is usually faster than launching a new page from zero.
- Effort: does the page need design, development, original research or just a rewrite?
Use simple high, medium and low ratings rather than precise scores. Precise scores create a false sense of accuracy when the inputs are estimates. The aim is to agree an order of work, not to calculate a forecast.
Example: scoring five keyword groups for a B2B service firm
Each group scored 1 to 3 on business value, ability to answer, current position and effort (inverted). Fictional data for illustration only.
Does keyword research still matter with AI search?
Yes, but the unit of research is shifting from single keywords towards questions and topics. AI Overviews, AI Mode, ChatGPT and Perplexity answer longer, more conversational queries, often with follow-up questions. That makes the buyer-language work in Step 1 more valuable, not less, because those conversational questions are exactly what your sales team already hears.
Google’s documentation on AI features and your website states that there are no additional requirements or special optimisations needed to appear in AI Overviews or AI Mode; a page must be indexed and eligible to show with a snippet. In other words, good keyword research, clear answers and sound technical SEO remain the foundation. Google’s Search Console generative AI performance report, which Google says was rolled out to all websites worldwide by 31 August 2026, now shows impressions from AI Overviews and AI Mode, so you can start to see which pages and topics appear there.
Practical adjustments for AI search:
- Record full questions, not just head terms, in your map.
- Write a direct 40 to 60 word answer near the top of each guide, then the detail.
- Cover the natural follow-up questions on the same page where they share intent.
- Read our guides on making answers easy to quote and how AI assistants choose which sources to cite for the next layer.
Worked example: mapping a service cluster
Imagine a fictional agency that offers CRM setup. Its seed list from sales calls includes “CRM for small business”, “move from Excel to CRM”, “HubSpot or Zoho”, “CRM setup cost” and “how to clean CRM data”. After expanding with tools and Search Console, it has around 180 queries.
Grouping by intent reduces those to 14 groups. Mapping them shows that:
- Four groups belong on the existing CRM service page (setup, cost, partner, India).
- Three belong on the existing HubSpot and Zoho sub-service pages.
- Two need a new comparison guide (“HubSpot vs Zoho”).
- Two need a new how-to guide (“migrating from spreadsheets to a CRM”).
- Three are definitions better handled by a glossary than by full pages.
The result is two new guides and a set of improvements to three existing pages, instead of 14 new articles. That is the outcome to aim for: fewer, stronger pages, each with a clear job. This example is illustrative; it does not describe a real client.
Common keyword research mistakes and their limits
- Treating volume as demand for your business. High-volume terms are often informational and attract people who will never buy. Check intent first.
- One keyword, one page. Creating a page for every variation spreads authority thin and causes cannibalisation. Group variations onto one strong page.
- Ignoring zero-volume terms. Many valuable B2B queries show little or no tool volume. If buyers ask it on calls, it is worth answering.
- Copying competitor keyword lists. Their audience, offer and authority differ from yours. Use them for gaps, not as your plan.
- Never revisiting the map. Language changes, products change and AI search changes how questions are asked. Review quarterly.
Keyword research also has limits. It cannot tell you whether you will rank, how many leads a page will produce or what competitors will publish next month. It can tell you what buyers are looking for and where your site should answer them. That is enough to plan well.
How keyword research connects to the rest of SEO
The intent map feeds every other part of an organic programme. It decides the structure explained in our guide to site architecture and internal linking, the clusters described in planning topical authority, and the priorities in your wider SEO strategy. When you check pages after publishing, the SEO audit checklist helps confirm that technical issues are not hiding them. For measuring what happens after the click, see marketing analytics and attribution.
Frequently asked questions
How many keywords should one page target?
One primary intent, which may include dozens of closely related queries. If the same page is the best answer for all of them, they belong together. If a query needs a different kind of answer, it needs a different page.
Are free tools enough for keyword research?
For most small and mid-size businesses, yes. Search Console, Keyword Planner, the results page itself and your own sales data cover the essentials. Paid tools add speed and competitor data, but they do not replace understanding intent.
What should I do with keywords that show zero search volume?
Keep them if real buyers use them. Tools often show little or no data for specialist B2B terms, yet those queries can bring highly qualified visitors. Add them to the relevant page as questions or sections.
How often should keyword research be updated?
Review the map every quarter, and whenever you launch a new service, change positioning or see new queries in Search Console. A full refresh once a year is usually enough for stable markets.
Do I still need keyword research for AI Overviews and ChatGPT?
Yes. AI answers are built on questions and topics, so understanding the questions your buyers ask is still the starting point. Focus on full questions and clear, direct answers rather than exact-match phrases.
Sources and further reading
- Google Search Console Help: Performance report overview, definitions of clicks, impressions, CTR and position.
- Google Ads Help: Use Keyword Planner, what the tool provides and how search figures are estimated.
- Google Search Central: AI features and your website, eligibility for AI Overviews and AI Mode.
- Search Console Help: Generative AI performance report, AI Overviews and AI Mode impressions.
- Google Search Central: Creating helpful, reliable, people-first content, the quality questions every page should pass.
Want help building your intent map? Bargaon’s SEO, AEO and GEO service starts with a free review of your current pages, queries and gaps, and ends with a keyword-to-page plan your team can use.