Keyword research is the process of finding the words and phrases people type into search engines and AI assistants, then deciding which of them deserve a page on your site. Good research does three things: it collects candidate searches, it works out what the person behind each search wants, and it ranks them by what they are worth to the business. For a SaaS company, the last step matters most, because a small search made by buyers ready to compare products can be worth more than a large one made by students.
Where the candidate list comes from
Most teams start in a keyword tool, but the better starting point is your own customers. Sales calls, support tickets, lost-deal notes and onboarding questions show the exact language buyers use, including terms a tool would never suggest because the volume is too small to measure.
After that, a keyword tool is useful for widening the list, grouping variants and checking rough demand. Google Search Console shows the searches you already appear for, which is the most reliable data you have because it comes from Google rather than a third-party estimate.
- Customer conversations: sales calls, demos, support tickets, churn reasons
- Search Console queries you already get impressions for
- Competitor pages that rank for searches you care about
- Keyword tools for expansion, related terms and rough volume
- People Also Ask boxes and autocomplete for question phrasing
Judging each search
For every candidate, answer four questions. What does the searcher want? Could your product be part of the answer? How hard would it be to rank? How much is a visit worth if it converts?
Search volume and keyword difficulty help with the third question, but both are estimates from third-party tools, and different tools give different numbers for the same keyword. Treat them as a rough sort, not a forecast.
Prioritizing by business value
A common failure is sorting the spreadsheet by volume and writing from the top down. That produces traffic that rarely turns into pipeline. A more useful order puts searches close to a buying decision first: comparisons, alternatives, pricing, integrations and specific use cases. Broad educational topics come later, once the pages that convert are in place.
Some teams add a simple score, such as a 1 to 3 rating for business value multiplied by a 1 to 3 rating for how winnable the search looks. The exact formula matters less than making the value judgment explicit.
Keyword research and AI search
People increasingly ask AI assistants longer, more conversational questions, and Google's AI features can expand one query into several related searches behind the scenes. Keyword research still applies, but the unit shifts toward topics and questions rather than single exact phrases. Covering a topic thoroughly, with clear answers to the follow-up questions a buyer would ask, tends to serve both classic results and AI answers.
Example
A hypothetical scheduling tool
Imagine a made-up SaaS company, Slotwise, that sells scheduling software to clinics. A tool says “online scheduling” has far more volume than “clinic scheduling software with insurance verification”.
The sales team reports that nearly every closed deal mentioned insurance verification. Slotwise writes the narrow page first, then pages for “Slotwise vs [competitor]” and the integrations prospects ask about. The broad topic waits until those pages exist, because the narrow searches are the ones buyers make the week they choose.
Common mistakes
Sorting by volume alone
High volume usually means broad intent and strong competition. Without a value judgment, the plan fills up with articles that bring visits and few customers.
Skipping the customer research
Keyword tools under-report small, specific B2B searches. Buyer language from sales and support fills that gap and often points to the highest-value pages.
Treating tool numbers as facts
Volume and difficulty scores are modeled estimates. Use them to compare keywords within one tool, not as promises about traffic.
Questions
Review it whenever the product, market or competitor set changes, and check Search Console regularly for new queries you are starting to appear for. A full rework is rarely needed more than once or twice a year.
Not to start. Search Console, autocomplete, People Also Ask and your own customer conversations go a long way. Paid tools help with scale, competitor data and volume estimates.
Yes, though the focus moves toward topics and the questions buyers ask. Knowing what people search for, and what they want when they do, is still the basis for deciding what to publish.
