Keyword research for SaaS: start from what you sell

Keyword research for SaaS starts from the product, not from a seed list. Write down what you sell in the words a buyer would use (the category, each use case, each rival you replace, each problem you fix), size every search, then check where your own Search Console already shows you. A SaaS keyword plan built that way is short, specific, and full of searches from people who could pay.

Most SaaS keyword research goes the other way: type a broad term into a tool, sort by volume, and write posts for the top of the list. That produces traffic from people who will never sign up, because the biggest searches near any software category are the ones furthest from a purchase.

TL;DR:

• Start from the product: your pricing page, feature pages and docs already name most of the searches worth winning.
• A SaaS plan needs four groups: category, use case, comparison and alternative, and problem searches.
• Size each search with real demand, and treat "not priced" as unknown, never zero.
• Check your own Search Console first. A search you already rank 11 to 20 for is the cheapest win on the list.
• Drop searches whose searcher has no budget or no use for what you sell, however big they are.

Why is keyword research different for SaaS?

Because a SaaS buyer searches for a job to be done, then a category, then a shortlist, and the product only appears at the end. "How to find pages not indexed" is the job. "Indexing checker tool" is the category. "[Rival] alternative" is the shortlist. Each step has less volume and more intent than the one before it.

Most keyword lists stop at the first step because that is where the volume is. For a SaaS site that already earns from search, the money sits in the last two, and those searches are long, specific and often too small for a keyword tool to price.

What four kinds of keywords does a SaaS keyword plan need?

Four groups, in roughly this order of value:

1. Category searches. What the product is, in the buyer's words: "Search Console MCP server", "SEO tool for developers", "SaaS SEO tool". Few, competitive, and the clearest signal of a buyer.
2. Comparison and alternative searches. "[Rival] alternative", "[rival] vs [you]", "best [category] for [audience]". Small, and written by people with a shortlist open in another tab.
3. Use case searches. One per job the product does: "find low CTR pages", "track SEO changes". Each one maps to a feature page you probably already have.
4. Problem searches. The question someone asks before they know a tool exists: "why did my impressions drop". These carry the most volume and the least intent. Their job is to link the reader to a page in one of the first three groups.

A plan with only the fourth group gets traffic and no signups. A plan with only the first three has nothing pointing at its money pages. Most SaaS sites need both, weighted towards the first three once the site has traffic.

How do you find SaaS keywords from your own product?

Read your own site the way a buyer would, then write down every noun phrase they might search:

• Pricing page: the category name, the plan names, who each plan is for.
• Feature pages: one use case search per feature, in plain words, not your internal feature name.
• Docs and integrations: "[your category] for [tool you integrate with]" for every integration you support.
• Sales calls and support email: every rival a prospect named, and every problem they described in their own words.

That list will be shorter than a keyword tool export, and almost every line on it will be about something you actually sell. That is the point.

How do you size and prioritise SaaS keywords?

Size each search with real demand, and read the gaps honestly. A search the tool won't price is often a long-tail buyer search below its floor, not a search nobody makes. Judge those on what the product says they are worth.

Then check your own Search Console before choosing. It shows which of those searches Google already shows you for and where. A search you rank 11 to 20 for needs one improved page. A search you have never appeared for needs a new one. Do the first kind first. Your own data under-counts here, because Google hides rare searches from the query report, so a long-tail search missing from it may still be sending you impressions.

Last, cut. Remove any search whose searcher has no repo, no budget, or no use for what you sell, however big it is. A plan of thirty searches you can win beats three hundred you cannot.

What tools do SaaS teams use for keyword research?

Most use a keyword database (Ahrefs, Semrush) for volume and rival rankings, and Search Console for where they stand. The database is good at size and bad at fit, because it starts from your seed. Search Console is your real position on your real searches, and it forgets after 16 months.

QueryScope does the research from the other end. Your agent reads the repo for what the site sells and proposes the searches; QueryScope sizes them, reads your own Search Console for where you stand, and names the rival searches you never appear for. The plan it keeps is one file in your repo that your team commits. The roundup of SEO tools for SaaS teams sorts the wider field by the job each tool does.

What can't keyword research tell a SaaS team?

Whether a search will turn into a customer. Volume says how many people search, intent labels say roughly what they want, and Search Console says where you rank. None of them sees past the click. A page can win the search it was built for and still not convert, and that is a product page problem, not a keyword one.

So treat the plan as the searches worth winning, chosen from what you sell, and measure each one on whether you win it. What the visitors do next is your analytics' job.

Sources

• Google Search Central, A deep dive into Search Console performance data filtering and limits (rare queries are left out of the query table while their clicks stay in the totals).
• Google Search Console Help, Performance report (data is kept for 16 months).