Keyword research for B2B SaaS should start where money changes hands, not where search volume is highest. Pull the exact phrases buyers use from sales calls, support tickets and lost deals, expand them into alternative, comparison, use-case and integration terms, and score each by business value and your odds of ranking. Then group them into clusters so every cluster gets one page built to convert.
Why search volume is the wrong starting point
The usual approach is to type the category into a tool and sort by volume. For B2B SaaS, that produces a list dominated by definitions and “what is” queries, searched by students, job seekers and people years away from buying.
The terms that produce demos look different. They’re specific, often name a competitor or a tool the buyer already uses, and tools frequently show them as zero volume. That tells you less than it seems: volume figures are modeled estimates, and they tend to be least reliable for exactly the niche phrases B2B buyers type.
A hypothetical example with made-up numbers: “what is revenue operations” might show thousands of monthly searches, while “[competitor] alternative for agencies” shows 20. If the second turns a couple of visitors a month into qualified demos on five-figure contracts, it’s worth more than the first for as long as it ranks.
So flip the order. Start from what buyers say, and use volume as a tiebreaker rather than a filter.
Mining sales calls, support tickets and lost deals
Your best keyword sources are the words prospects and customers already use, in your own systems first and on public review sites second.
- Sales calls. Listen for the problem in the buyer’s words, the trigger that made them look now, and the tools they’re comparing. “We’re outgrowing spreadsheets” is a keyword seed.
- CRM closed-lost reasons. Where reps log them, lost deals name the competitors that beat you and the objections that stopped the deal. Each is a page you don’t have yet.
- Closed-won “previous tool” data. This shows what customers switch from: the raw list for alternatives pages.
- Support tickets. These surface integration needs and “how do I” phrasing that prospects also search before buying.
- Google Search Console. Queries already earning impressions at an average position of roughly 11 to 20 are often the quickest wins, and they reveal phrasing you’d never guess.
- Review sites. Competitor reviews on G2 or Capterra list what people dislike about the incumbent.
Copy phrases verbatim into a sheet with the source next to each. Don’t polish them; the raw phrasing is what people type.
When I run this with a sales team, the most useful output is rarely a surprise keyword. It’s clarity on which three or four competitors actually show up in deals, often a different list from the one on the board slides.
Bottom-of-funnel keyword types
Sort the raw phrases into the patterns that tend to carry buying intent:
| Type | Pattern | Who’s searching |
|---|---|---|
| Alternative | “[competitor] alternatives” | Unhappy users of a competitor |
| Comparison | “[product] vs [competitor]”, “[A] vs [B]” | Buyers down to a shortlist |
| Category + qualifier | “[category] software for [industry]” | Buyers building a shortlist |
| Use case | “[category] for [job to be done]” | Buyers with a specific problem |
| Integration | “[category] with [tool] integration” | Buyers checking stack fit |
| Pricing | “[competitor] pricing”, “[category] cost” | Buyers building a budget |
| Problem-aware | “how to [fix the problem you solve]” | People feeling the pain, not yet shopping |
Work top to bottom. Alternative and comparison terms sit closest to a purchase; problem-aware terms bring earlier readers who need nurturing first. A newer site should build the first five rows before writing many articles.
Competitor, alternative and comparison terms
Build the competitor list from your deals, not from market maps. Include the non-obvious alternatives buyers weigh: spreadsheets, an in-house build, an agency, or doing nothing. “Build vs buy [category]” is worth checking, because vendors often skip it and the searcher is actively deciding.
For each real competitor, you typically end up with three targets:
- “[competitor] alternatives”: realistic options, you included, with who each one suits
- “[product] vs [competitor]”: a direct comparison for buyers with both on the shortlist
- “[competitor] pricing”: usually best served by a section on the comparison page
Also check “[competitor A] vs [competitor B]” for pairs that appear in your deals. You’re not in the query, but the reader is mid-evaluation, and a fair comparison that adds you as a third option can earn the click.
These pages only work if they’re honest. Say where the competitor is the better fit, date your pricing and feature claims, and cover switching: migration effort, data import, contract timing. A rigged table won’t convert or earn links, and invites trouble when you’re naming another company’s product.
Use-case and integration keywords
Use-case keywords come from how customers describe the job, not from your feature list. Group them three ways:
- By role: “[category] for finance teams”
- By industry: “[category] for agencies”
- By job: “[category] for month-end close”
Only build a use-case page when you have something specific to say: a workflow, a template, the objections that segment raises. Swapping the industry name into the same template is how thin pages happen.
Integration keywords are underrated. Someone searching “[category] that integrates with [CRM]” has told you their stack and their requirement. Pull the list from your integrations directory and support tickets, then give each page substance: what syncs, in which direction, how often, and what it doesn’t do. With dozens of integrations that share a structure, templated pages can be legitimate; the programmatic SEO guide covers when that works.
Scoring keywords by business value
Two scores matter more than volume.
Business value (0-3)
- 3: your product is the direct answer and the searcher is choosing a tool
- 2: your product solves the problem, but the searcher is still defining it
- 1: related to your space, but mentioning the product is a stretch
- 0: no realistic path to a customer. Drop it, whatever the math says
Win likelihood (1-3). Search the keyword and read page one.
- 3: results are weak, dated or from sites no stronger than yours, and match the page type you’d build
- 2: a mix of strong and beatable results
- 1: dominated by major publishers, review platforms or much bigger competitors
Then treat volume as a tier (1-3), relative to your niche rather than to the web (1 = low, 3 = high), so a 40-search term with value 3 isn’t buried under a 4,000-search term with value 1.
Priority = Value × Win × Volume tier, from 1 to 27 for every keyword you keep. An example with made-up keywords and scores:
| Keyword | Type | Value | Win | Volume tier | Priority |
|---|---|---|---|---|---|
| [competitor] alternatives | Alternative | 3 | 2 | 2 | 12 |
| [category] hubspot integration | Integration | 3 | 3 | 1 | 9 |
| [product] vs [competitor] | Comparison | 3 | 3 | 1 | 9 |
| how to reduce [problem] | Problem-aware | 2 | 2 | 2 | 8 |
| [category] for agencies | Use case | 3 | 2 | 1 | 6 |
| what is [category] | Educational | 1 | 1 | 3 | 3 |
The “what is” term has the most searches and lands at the bottom. That’s the point.
Win likelihood also sets expectations. A newer domain chasing win-1 terms is signing up for a long project; the SEO timeline guide covers realistic ranges.
Grouping keywords into clusters and page types
A scored list still isn’t a plan. Several keywords usually belong on one page, and separate pages for near-duplicates split your authority and leave Google guessing which URL to rank.
Let the results page decide. If two keywords return mostly the same top results, they’re one cluster and one page. If one shows comparison pages and the other shows how-to articles, they need separate pages.
Then map each cluster to a page type:
| Cluster example | Page type | Primary job |
|---|---|---|
| [competitor] alternatives, apps like [competitor] | Alternatives page | Get switchers onto your shortlist |
| [product] vs [competitor], [competitor] pricing | Comparison page | Win head-to-head evaluations |
| [category] for [industry] and variants | Use-case page | Prove fit for a specific buyer |
| [category] [tool] integration, sync [tool] | Integration page | Remove a stack objection |
| how to [solve problem], [problem] checklist | Article or template | Capture earlier demand and route it down |
Educational articles should link down to the commercial pages in their cluster, and commercial pages should link to each other where buyers compare. That linking is how top-of-funnel content feeds pipeline instead of just traffic.
Every cluster gets one primary keyword and one URL. If two existing pages target the same cluster, merge them and redirect the weaker one. This mapping is also what sets the build order in my technical SEO and content systems work.
Keyword research template
Keep everything in one sheet with these columns: keyword (exact phrase), source (call, lost deal, ticket, Search Console, tool), type, cluster, page type, target URL, value, win, volume tier, priority, status and owner.
Then run the process as a checklist:
- Collect verbatim phrases from at least ten recent sales calls, two quarters of closed-lost reasons and a sample of support tickets
- List the competitors and non-obvious alternatives that actually appear in deals
- Expand phrases into the bottom-of-funnel types above, then check volume
- Score business value and win likelihood for every keyword
- Cluster by results-page overlap and assign one page type and one URL per cluster
- Build the top five to ten clusters first, starting with alternatives, comparisons and integrations
- Track demo requests and pipeline by first landing page in your CRM, not just rankings
- Repeat the sales and support mining quarterly; new competitors appear in calls before they appear in tools
Get it built
If you want a keyword plan built from your own sales and CRM data rather than a tool export, get in touch and tell me where pipeline is coming from today. You can see how engagements are structured on the pricing page.