Euracle
Marketing

How to Do B2B Keyword Research When Volume Lies

Every B2B keyword research guide will tell you that search volume is the wrong metric. They are right, and almost none of them gives you the metric to use instead. B2B keyword research means finding the search terms your buyers use, tagging each one by what the searcher wants, and pricing each one in expected pipeline rather than expected traffic. The formula is volume times click rate times lead rate times close rate times deal value. Everything else is sourcing and sorting.

TL;DR

  • Price keywords in pipeline, not traffic. Monthly value = volume × click rate × visit-to-lead rate × lead-to-close rate × deal value. Six inputs, all of which you already have.
  • Ahrefs found that 99.2% of keywords triggering AI Overviews are informational, so informational keywords carry a click discount that commercial keywords largely avoid.
  • That discount is real. Ahrefs measured position-one click-through falling 34.5% when an AI Overview is present (April 2025) and 58% in its February 2026 re-run on December 2025 data.
  • In the worked example below, a 50-search commercial keyword outperforms a 5,000-search informational one. Remove the AI Overview discount and the answer reverses, which is why 2026 guidance differs from 2022 guidance.
  • The highest-intent B2B terms often show zero volume in every tool. They come from sales call recordings, your site's own search box and support tickets, not from a keyword tool.

How do you do B2B keyword research?

Collect the words your buyers actually use, tag each one by what the searcher wants, price each one in expected pipeline, then group them into pages. The tool comes fourth, not first.

Three plain-English terms you will meet in any tool:

  • Search volume is the estimated number of times a term is searched per month. It is an estimate, and for B2B terms it is often wrong or shown as zero.
  • Keyword difficulty, usually shown as KD, is a tool's guess at how hard it is to rank on page one, scored 0 to 100. Different tools score the same keyword differently.
  • Search intent is what the person wants. Informational means they want to learn. Commercial means they are comparing options. Transactional means they are ready to buy.

That third one is the load-bearing concept in B2B, and the rest of this article is mostly about it.

Why does search volume mislead B2B teams?

Because in B2B keyword research the people searching high-volume terms are mostly not your buyers. A term like "what is content marketing" attracts students, job seekers, journalists and competitors' employees. A term like "b2b content marketing agency pricing" attracts about forty people a month, and several of them are trying to buy something.

There is a structural reason for this. In B2B, the more specific a search gets, the smaller the audience and the more purchase-ready it is. Specificity works as a proxy for intent. Somebody searching "accounts payable automation software" has already diagnosed their problem and picked a category. Somebody searching "fintech software" has not.

The cost of getting this wrong is not zero traffic. It is worse. You rank, the traffic chart looks excellent, the sales team gets nothing, and the programme gets cut at budget review because nobody can connect it to revenue.

This part of the advice is now standard. Most guides on this topic say it. What almost none of them do is give you a way to prove it on your own keyword list, which is the next three sections.

What has changed in 2026?

Informational keywords now lose clicks in a way commercial keywords mostly do not, and that shifts the arithmetic rather than the principle.

Ahrefs analysed which keywords trigger Google's AI Overviews, the generated summaries that appear above the normal results, and found 99.2% of them are informational in intent. So the discount applies almost entirely to the educational half of your keyword list.

The size of that discount is measurable. Ahrefs compared click-through rates on keywords with and without an AI Overview across 300,000 keywords and found the top-ranking page received 34.5% fewer clicks when an AI Overview was present (April 2025). Re-running the same method on December 2025 data, the figure was 58% (February 2026).

99.2%: the share of AI Overview-triggering keywords that are informational in intent (Ahrefs, 2025)

Two consequences for a B2B keyword list.

Your educational content now converts a smaller share of its impressions into visits. That does not make it worthless. It makes it worth less, and the amount is now something you can put in a spreadsheet.

Your commercial and transactional keywords are largely untouched. "Vendor comparison" and "pricing" style searches rarely trigger the feature, so their click rates hold up. Whether a specific keyword triggers one is checkable rather than assumable, which is step 5 below and is covered in more depth in our guide to ranking in AI Overviews.

How do you price a single keyword?

B2B keyword research reduces to six numbers and one multiplication. You already have five of them.

Expected monthly pipeline = volume × click rate × AI Overview adjustment × visit-to-lead rate × lead-to-close rate × average deal value

Where it comes from
Monthly search volumeYour keyword tool
Click rate at your target positionYour own Search Console data. As a planning default, roughly 9% at positions 3 to 5 and 4% at positions 5 to 10
AI Overview adjustment1.0 if the keyword shows no AI Overview. Roughly 0.42 if it does, using Ahrefs' February 2026 figure of a 58% CTR reduction
Visit-to-lead rateYour analytics, split by page type. Commercial pages convert several times better than blog posts
Lead-to-close rateYour CRM
Average deal valueYour CRM

The click rates above are planning defaults, not measurements of your site. Replace them with your own Search Console figures as soon as you have thirty days of data, because they vary enormously by industry and brand recognition.

Run this on your top thirty keywords before you commission a single article. It takes about an hour and it is the difference between a keyword list you can defend to a CFO and one you cannot.

Worked example: 5,000 searches against 50

Two keywords, one hundred times apart in volume. Both assumed to rank at position 3. Deal value $15,000 and a 5% or 15% close rate depending on lead type. All inputs are illustrative; substitute yours.

Keyword A: "what is content marketing"

Informational. 5,000 searches a month. Triggers an AI Overview.

StepCalculationResult
Volume
5,000
Click rate at position 3× 9%450 clicks
AI Overview adjustment× 0.42189 clicks
Visit-to-lead, blog reader× 0.5%0.95 leads
Lead-to-close, low intent× 5%0.047 deals
Deal value× $15,000$709 per month

Keyword B: "b2b content marketing agency pricing"

Commercial. 50 searches a month. No AI Overview.

StepCalculationResult
Volume
50
Click rate at position 3× 9%4.5 clicks
AI Overview adjustment× 1.04.5 clicks
Visit-to-lead, commercial page× 8%0.36 leads
Lead-to-close, high intent× 15%0.054 deals
Deal value× $15,000$810 per month

The 50-search keyword wins. It has one hundredth of the volume and produces more pipeline.

Now remove the AI Overview discount

Set keyword A's adjustment to 1.0, as it would have been before AI Overviews existed:

With AI Overview discountWithout
Keyword A, 5,000 searches$709$1,688
Keyword B, 50 searches$810$810

Without the discount, keyword A wins by more than two to one. The AI Overview is what flips the result.

This is why keyword research advice written before 2024 pointed at volume and current advice does not. The principle did not change. One of the six inputs did, and it changed enough to reverse the ranking. Anybody telling you low-volume keywords have always been better is telling you something that was not true when they first said it.

The eight-step process

Time required: about four hours to run B2B keyword research across a list of 200 keywords, then an hour a quarter to maintain it. Prerequisites: Google Search Console verified, a keyword tool with a SERP features filter, three recent sales call recordings or transcripts, and your average deal value and close rate from your CRM.

Step 1. Harvest your buyers' actual words, before opening any tool

Listen to three recent sales calls and write down how the prospect described their problem before your team reframed it in your terminology. That language is where the highest-intent terms live, and it is nowhere in a keyword tool because you have never typed it in.

Step 2. Add the sources tools cannot see

Pull your site's internal search queries from GA4, your last hundred support tickets, and the questions your sales team gets asked most. Section 7 covers this in detail.

Step 3. Expand into a seed list

Now open the tool. In Ahrefs, use Keywords Explorer, enter your seed terms, and work through Matching terms and Questions. In Semrush, the equivalent is the Keyword Magic Tool. Aim for 200 to 400 raw keywords. Do not filter yet.

Step 4. Tag every keyword by intent

Three tags: informational, commercial, transactional. Do this by hand or in a spreadsheet column. It is the least glamorous step and the one that determines everything downstream. A keyword you cannot confidently tag is usually one where you should look at the actual results page before deciding.

Step 5. Check which keywords trigger an AI Overview

In Ahrefs, open Site Explorer, go to the Organic keywords report, and set the SERP features filter to AI Overview. For keywords you do not yet rank for, check them in Keywords Explorer, which shows SERP features per keyword. Mark each keyword yes or no. This single column decides which adjustment factor each keyword gets in step 6.

Step 6. Price every keyword

Apply the formula from section 4 in a spreadsheet. One row per keyword, one column per input, one final column for expected monthly pipeline. Sort by that final column, descending.

The list you are now looking at will not resemble the one you would have produced by sorting on volume. That is the point.

Step 7. Group keywords into pages

Keywords that would be answered by the same page belong on the same page. Section 8 covers how to decide.

Step 8. Set a review date

Quarterly for most B2B companies. Search behaviour, competitor positioning and AI Overview coverage all move, and a keyword list built once and never revisited is a keyword list describing last year.

Where do you find keywords tools cannot see?

B2B keyword research turns up its most valuable terms in places tools cannot reach. Those terms frequently show zero volume everywhere, because tools round small numbers down and B2B numbers are small. Zero volume does not mean zero searches. It means below the tool's reporting threshold.

Five sources, in order of value:

  1. Sales call recordings. The single best source. You are looking for the words used in the first two minutes, before anyone adopts your vocabulary.
  2. Your site's internal search box. In GA4, if site search is configured, these appear as the search_term parameter under Reports, then Engagement, then Events. Somebody already on your site typing a phrase into your search box is telling you about a gap in your content.
  3. Support tickets. Recurring questions are recurring searches.
  4. Search Console's Queries report, filtered to impressions above zero and clicks at zero. These are terms Google already associates with you where you rank too low to get clicked.
  5. Communities where your buyers ask questions. Read how the question gets phrased, not how the answers get phrased.

Do not discard a zero-volume keyword because the tool shows zero. Price it with a conservative volume estimate of five to ten searches a month and see whether it still clears your threshold. High-intent B2B terms frequently do.

How do you group keywords into pages?

Group by the answer, not by the words. If two keywords would be satisfied by the same page, they belong on the same page.

The practical test: search both terms and compare the first page of results. If the same pages rank for both, Google treats them as one topic and you should too. If the results differ substantially, they need separate pages.

Getting this wrong produces cannibalisation, which is when two of your own pages compete for the same term. Google picks one, usually not the one you wanted, and both underperform. The fix is to merge them, which is cheaper than it sounds and almost always improves both.

One cluster becomes one substantial page plus several supporting pages that link to it. Which keyword gets the main page is decided by the pricing column from step 6, not by which one has the most volume.

What do most teams get wrong?

Sorting by volume and calling it prioritisation. Most B2B keyword research stops here. The most common mistake and the reason for this article.

Tagging intent by guessing instead of by looking. If you are unsure what a searcher wants, search the term and look at what Google is already rewarding. The results page is the answer key.

Ignoring the AI Overview column. Skipping step 5 means every informational keyword on your list is priced as though it were 2022. On the worked example above, that is a 58% error on one input.

Treating tool volume as fact. Volumes are estimates, they differ between tools, and they round small numbers to zero. Use them as a rough size signal, not as a measurement.

Building the list once. A keyword list is a living document. Quarterly review, minimum.

Letting two pages chase one keyword. Check for cannibalisation before publishing, not after rankings drop.

Skipping the sales call step because it is slow. It takes ninety minutes and it produces the terms your competitors' keyword tools will never show them. It is the only step in this process that gives you something nobody else has.

How does Euracle run keyword research?

Euracle runs B2B keyword research so that the pricing column comes before the content calendar, and the calendar is sorted by it.

The Eureka Method, Euracle's discovery sprint, runs four phases here.

Discover starts with sales call recordings rather than a tool, because the language a buyer uses before being corrected is the language that converts.

Design prices every candidate keyword using the client's own conversion and deal-value figures, so the resulting calendar is defensible to whoever signs it off.

Deploy builds the highest-value clusters first rather than the easiest ones.

Scale re-runs the AI Overview column quarterly, because coverage of that feature keeps moving and it sits inside the formula.

The stack is conventional: Ahrefs and SEMrush for volume and difficulty, Google Search Console and GA4 for the click and conversion rates that make the formula specific to the client rather than generic, and the client's CRM for deal value and close rate. The formula is only as good as those last two numbers, which is why the first meeting is usually about the CRM rather than about search.

Two structural commitments come from how Euracle is set up. Senior practitioners only: the people in the pitch do the work. And one contract across six disciplines, so a finding that the real constraint is a missing pricing page rather than a missing blog post does not require a second vendor to fix.

If you want the pricing done on your own numbers before a content calendar is built, that is Euracle's SEO service. The zero-volume problem is sharpest for B2B SaaS companies, where the terms buyers use for a new category often do not exist in any tool yet.

FAQ

Collect the words your buyers actually use from sales calls, your site search and support tickets, expand them in a keyword tool, tag each term as informational, commercial or transactional, check which ones trigger AI Overviews, then price each one as volume times click rate times lead rate times close rate times deal value. Sort by that figure.

Frequently yes. In the worked example above, a 50-search commercial keyword produced more expected pipeline than a 5,000-search informational one, because it converts several times better and does not carry an AI Overview click discount. Price both before deciding rather than assuming either way.

Zero-volume keywords are terms your tool reports as having no measurable searches, usually because the true figure falls below its reporting threshold rather than because nobody searches them. In B2B these are often the highest-intent terms. Price them with a conservative estimate of five to ten searches a month and see whether they clear your threshold.

They add a discount to informational keywords. Ahrefs found 99.2% of AI Overview-triggering keywords are informational, and measured position-one click-through falling 34.5% when one is present (April 2025), rising to 58% in a February 2026 re-run. Commercial and transactional keywords are largely unaffected, so the gap between the two widens.

About four hours for a list of 200 keywords if you have Search Console access, a keyword tool and your conversion figures to hand. Add ninety minutes if you include the sales call harvesting step, which is worth it. Budget an hour per quarter afterwards to keep the list current.

The main cost is tool access, with major keyword tools running roughly $100 to $500 a month depending on tier and seat count, plus the internal hours above. Agency-delivered keyword research is usually bundled into a strategy engagement rather than sold alone. Figures as of August 2026, and tool pricing changes regularly.

Fewer than most lists contain. A list of 200 to 400 raw keywords usually reduces to 30 to 60 that are worth pages, once priced. Targeting more than that spreads a small content budget across pages that individually never reach a rank worth having.

Cannibalisation is when two of your own pages compete for the same search term, so Google picks one and both underperform. Avoid it by grouping keywords into clusters before writing, using the test of whether the same pages already rank for both terms. If two existing pages already compete, merging them usually improves both.

Conclusion

You can now price a keyword instead of arguing about it. Harvest the language from sales calls, tag every term by intent, check the AI Overview column, run the six-input formula on your own conversion and deal figures, then sort by expected pipeline and build in that order. The list you get will look nothing like a volume-sorted list, and it will be one you can defend to whoever controls the budget. If you would rather have the pricing run on your numbers before a content calendar exists, talk to Euracle about SEO.

Sources

  1. Ahrefs, AI Overviews Reduce Clicks by 34.5%, Ryan Law and Xibeijia Guan, 17 April 2025. Study of 300,000 keywords; the 99.2% informational-intent finding. https://ahrefs.com/blog/ai-overviews-reduce-clicks/
  2. Ahrefs, 2026 update to the above, published February 2026 on December 2025 data, reporting a 58% reduction. https://ahrefs.com/blog/ai-overviews-reduce-clicks-update/
  3. Google Search Central, AI Features and Your Website, on eligibility and how generative features select content. https://developers.google.com/search/docs/appearance/ai-features
Image of Devanshu

Written by

AI-Native Product Manager + GTM Engineer

Keep reading

More from the journal.

Marketing

How to Rank in AI Overviews When the Studies Disagree

Most guides on this topic quote one citation-overlap statistic and build a plan on it. There are two, they disagree by a factor of 4.5, and which one you believe changes what you should do on Monday. To rank in AI Overviews, get the page into Google's top five organic results for an informational query, then put a self-contained 40 to 70 word answer directly beneath a heading that matches the query. Google confirms there is no separate AI Overview optimization: the page must be indexed and eligible to appear with a snippet, and nothing more.

Image of Devanshu
Devanshu Takkar
AI-Native Product Manager + GTM Engineer
Aug 21, 2026
Read article
Marketing

Answer Engine Optimization (AEO): A Working Guide to Getting Cited

Answer engine optimization (AEO) is the practice of structuring content so AI answer systems such as Google AI Overviews, ChatGPT, Perplexity, Gemini and Copilot can retrieve it, quote it accurately, and name your brand as the source. It differs from SEO in its unit of success: a cited passage, not a ranked link.

Image of Devanshu
Devanshu Takkar
AI-Native Product Manager + GTM Engineer
Aug 20, 2026
Read article
Automation

AI Agent vs Chatbot: Who Chooses the Next Step?

For most business processes, the right answer is plain workflow automation, not an AI agent. A chatbot answers questions. An automation runs steps you defined. An agent picks its own steps. If your process already has a known sequence, an agent adds cost and unpredictability and gives you nothing back.

Image of Devanshu
Devanshu Takkar
AI-Native Product Manager + GTM Engineer
Aug 25, 2026
Read article

The breakthrough, delivered

Your breakthrough is one conversation away.

Let's find your spark