How We Do B2B Keyword Research (Including the Keyword We Skipped)

Most keyword research content describes a framework: check volume, check difficulty, pick your targets. It rarely shows the actual decision-making, including the calls that do not go the way you would expect. 

This is that walkthrough: the real process we used to research keywords for this article, a live example of a keyword we evaluated and chose not to chase, a real competitor analysis, and why the internal linking work that follows connects to indexing and authority, not just tidiness.

The most important part isn't any single step. It's the why, threaded through the article.

Keyword Research: Gathering, Comparing, and Choosing What to Target

Where We Gather Keywords, and Why

A keyword tool is a starting point, not a strategy. 

For those running Google Ads or who have access to it, we start with Keyword Planner. The reason is simple: there's no better source of demand data than the people already paying for traffic on the platform that sells it. This step isn't mandatory, more of a nice-to-check before we dig in. If you don't have a paid platform to check against, skip straight to SE Ranking or your company's platform for keyword research.

Before that raw list goes any further, it gets a quick double-check, once from us, once from Claude, mainly to catch anything obviously off before spending time verifying it properly.

Next, take the list in its raw form and run it through SE Ranking to verify volume, difficulty, and, most importantly, pull a historical SERP composition. Across a given keyword set, this splits results into keywords a competitor ranks for that we do not, keywords we both rank for, and keywords we rank for that they do not, along with a read on whether the SERP is genuinely made up of competitors, or of non-competitive sites like tool vendors or major publishers that happen to rank but aren't part of your actual market.

The reason to gather from a competitor rather than a generic seed list is simple. 

A competitor's actual ranking keywords reflect demand that has already proven it can support a page worth ranking. A keyword tool's suggestion list has not been tested against anything.

The Real Evaluation

For this article, we ran the keyword research process on itself. Here are a few of the keywords that we searched:

• keyword research: Difficulty 100, roughly 5,400 searches a month
• SEO keyword research: Difficulty 91, roughly 480 searches a month
• keyword gap analysis: Difficulty 67, roughly 10 searches a month
• B2B keyword research: Difficulty 19, roughly 10 searches a month

Volume and difficulty figures pulled from SE Ranking's Keyword Research tool (July 15, 2026)

Difficulty scores are useful, but they are not a direct measurement. 

According to Ahrefs, KD is calculated from the number of referring domains the current top 10 pages have, and it explicitly does not account for on-page factors. It tells you how many links you'd need. It doesn't tell you who already holds them or whether they can be outcompeted, which is exactly what turned up here. 

It was dominated end to end by Semrush, Ahrefs, WordStream, Google's own Keyword Planner, Wordtracker, KeywordTool.io, SpyFu, Moz, HubSpot, and Neil Patel: tool vendors competing on tool functionality, not educational content.

Simply put, these are companies with deep pockets, dedicated in-house SEO and content teams, and the resources to defend the top of that SERP indefinitely. Not a fight worth picking for a site our size.

We took a temporary step back. What about AI search results instead? We decided to aim for getting cited there rather than ranking there. We also needed to check whether an AI answer engine like ChatGPT or Perplexity had already broken into that SERP, or the smaller "B2B keyword research" SERP. Neither had.

None of that guarantees a citation, and we know it. 

But here our own position is no different from the in-house teams and demand generation marketers reading this: you do not win every keyword, and sometimes you still need to finish the piece anyway, because it connects the rest of the cluster. At that point, the job stops being about ranking one article and starts being about maximizing what a finished piece of work can do.

Outside of organic search, we can pull sections of this article for social media, mainly through team members' own LinkedIn profiles right now, for additional traffic and, potentially, a client who never would have searched for this keyword at all.

Would we bill a client for an article we didn't expect to rank? Not on its own. Content is a sum of its parts: some pieces carry more weight than others, and that's fine if the series holds together enough to keep someone reading the next one. That's what moves a buyer, not one article ranking, but a body of work worth finishing. We'd only write this for a client if their pillar needed it, the same way a history section earns its place only when the topic requires it.

The keyword we ended up targeting for this article, B2B keyword research, was a deliberate exception, chosen because nothing better cleared the bar. Most of the time the numbers work out better than that. Here's what the same process looked like on a real project.

Comparing Against a Competitor

We are looking to start our own internal SEO work, which requires content to be created. To find out what to write about, we look up the terms prospective clients actually search for. 

These are the same terms behind How to Choose an SEO Agency in Canada: SEO consultant, SEO companies, SEO expert, SEO consulting service, best SEO agency Canada, best Canadian SEO company, top SEO services in Canada, and SEO services Canada.

The goal is to find keywords you can rank for with volume that can support traffic growth. For our in-house team, the ceiling is 70 difficulty and volume over 100; that's our criteria, and it's most likely different from yours. What matters is setting your own limits and moving on to the next step: pulling competitor data.

A competitor's unique keyword gap is the list of terms they rank for that we do not, matched against whichever piece of content you are trying to win.

We filtered the export for a minimum viable search volume and a difficulty ceiling actually within reach, and excluded city-specific and out-of-market terms (SEO Toronto, SEO service Ottawa, website design in winnipeg) and out-of-market terms (ga4, google ads fees, cost for google ads, anything about paid media pricing rather than choosing an SEO provider). 

Filtering is not fully mechanical; it takes a judgment pass specific to who you are comparing against. Where you want to rank. What you want to rank for. A competitor's client case study pages and service lines outside your own market will show up in the raw export and AI can help exclude this to whittle down your keyword list.

The top ranking competitor for these words showed a starting point of 4,340 keywords across 227 pages from a Canadian marketing agency we don't rank for (data pulled July 16, 2026 from SEranking).

Next our AI agent applied our internal rules. 

The volume and difficulty ceiling alone cut that to under 800. Followed by excluding city-specific terms and obvious out-of-market noise took another pass, and even that wasn't clean on the first try, phrasing like cost of a google ad or advertising rates on google doesn't match a simple exclusion list the way GA4 or SEO Toronto does.

Which is exactly why this step needs judgment, not just a fixed rule set, and that judgment comes with experience.

What was left after getting it genuinely clean: a little over 500 keywords worth looking at, spread across 65 pages. On paper, that's still a much bigger footprint than ours.

This is more realistic territory for finding what we should actually be looking at, but 500 keywords is still just a list, not a plan. The next step is checking whether they group into one real topic instead of several separate ideas. Google doesn't rank a page by matching keywords one at a time; it uses language models to match a page's actual meaning to the intent behind a search, not just the literal words typed in.

We didn't have to guess at this. 

The proof was already in the data: one single page on this agency's site ranks for five of the eight target keywords at the same time, and a second page ranks for the other three. 

If these were genuinely different topics, it would be unusual for one page to rank for several of them simultaneously. That overlap is the real evidence they're one search intent, someone trying to evaluate and choose an SEO provider, not eight separate targets. This agency splits that intent across two pages instead of answering it in one place: a city-specific page pulling about 1,219 visits a month, and a national page pulling about 202.

More pages and more keywords don't mean better execution. 

Consolidating one search intent into a single, thorough page beats splitting it across several thin ones, which is exactly what How to Choose an SEO Agency in Canada does instead.

Why This Matters: E-E-A-T

None of this works if it isn't true. Every number in this article, the difficulty scores, the competitor's traffic, the keyword we chose not to chase, is real, checked against actual data rather than asserted. 

That's what Google's Search Quality Rater Guidelines call Experience, Expertise, Authoritativeness, and Trust, or E-E-A-T, a framework Google explains in its own documentation on creating helpful, reliable, people-first content. 

Internal Linking Setup

Why It Matters Beyond Good Practice

Internal linking is usually framed as tidiness. It's actually two concrete mechanisms: passing authority, and indexing. Google's own guidance is direct on both: every page you care about should have a link from at least one other page on your site, and Google uses links to find new pages to crawl in the first place. A new page with no internal links pointing to it inherits none of the site's existing authority, and it can sit undiscovered far longer than it should, sitemap or not.

The Actual Architecture

Here's this exact article's own place in that structure: it sits in a cluster feeding the digital marketing pillar page, which feeds Growth Marketing at the top level. It links sideways to How Much Does SEO Cost in Canada and How to Choose an SEO Agency in Canada, two sibling articles in the same cluster.

That's not a diagram. It's what the links in this piece are actually doing right now.

Execution Rules

• Link new pages from already-indexed, established pages specifically, not just from other new content.
  
  • Tip: go through your existing case studies or service pages to see if there is an opportunity to link.
• Use descriptive, keyword-relevant anchor text, not "click here" or a bare URL.
  
  • Tip: see how we tied SEO and GEO into the anchor text on that link, it tells search engines what the destination page is about instead of leaving them to guess.
• Link sideways between sibling articles in the same cluster where it's genuinely useful to the reader, not only upward to the pillar.
  
  • Tip: this is also for the reader, no one likes to play choose your own adventure and hope it is the correct article
• Treat every page you actually want found as needing at least one link from somewhere already in the index.
  
  • Tip: without this in mind, you can have an amazing piece of content that never sees the light of search results.

Confirming It Worked

Two tools close the loop once the links are in place. Search Console's Internal Links report, under Links in the left sidebar, shows which URLs receive the most internal links, a reasonable proxy for where authority is concentrating. 

The URL Inspection Tool checks a specific page's index status directly and lets you request indexing rather than waiting on the next crawl. Building the links and never checking either report is the same as skipping this step entirely.

Talk to our team if you'd rather walk through your own keyword map than build one from scratch.

Frequently Asked Questions

What's the actual process for doing B2B keyword research?

Start with your own seed keywords, then check the SERP for those terms to find out who's actually ranking. Pull that competitor's unique keyword gap, filter for realistic volume and difficulty, and cluster what's left by topic.

How do you decide which keywords are worth targeting?

Check the actual SERP, not just the difficulty score. If the pages ranking are non-competitors, a low difficulty number is meaningless, since there's no real prospect to win. Target keywords where the people searching are actual buyers and the field is companies you'd genuinely go up against.

What if the right keyword has too much competition or too little search volume?

If every realistic option is either too competitive or too small, target the smallest genuinely winnable option on purpose and get the rest of the value elsewhere: topical depth, internal linking, and earned trust signals, rather than skipping publication altogether.

What is E-E-A-T and how does it apply here?

E-E-A-T stands for Experience, Expertise, Authoritativeness, and Trust. It is not a ranking factor Google calculates directly; it is the quality framework its human raters use and its systems are built to reward. Content that documents a real process, shows real specificity, builds topical depth over time, and is honest about limitations reflects it.

How does internal linking affect indexing, not just rankings?

Google uses links to discover new pages to crawl. A page with no internal links pointing to it, sitting only in a sitemap, can go unindexed far longer than a comparable page linked from an established, already-indexed page on the same site.

References

Ahrefs. (n.d.). What does KD stand for in Keywords Explorer? Ahrefs Help Center. https://help.ahrefs.com/en/articles/72265-what-does-kd-stand-for-in-keywords-explorer

Google. (n.d.). SEO link best practices for Google. Google Search Central. https://developers.google.com/search/docs/crawling-indexing/links-crawlable 

Google. (n.d.). Creating helpful, reliable, people-first content. Google Search Central. https://developers.google.com/search/docs/fundamentals/creating-helpful-content

Google. (n.d.). How Google determines ranking results. How Search Works. https://www.google.com/search/howsearchworks/how-search-works/ranking-results/