SaaS keyword research

A keyword tool can only show you demand that already has a name. For a SaaS product, a large share of the terms worth owning have either no measurable volume, a volume figure rounded so hard it is meaningless, or a name your buyers use that nobody has indexed yet. Sorting an Ahrefs export by monthly searches and keeping the top 50 will reliably produce a plan that ranks for things your buyers do not search and misses the things they do.

So build the seed list somewhere else, then use the tool to size and check it.

Where SaaS seed keywords actually come from

Six sources, in the order we work through them. None of them is a keyword tool, and together they usually produce 200 to 400 raw candidates before any filtering.

The G2 and Capterra category tree. Find your category page, then read the sidebar. The related categories, the sub categories and the filter facets are a taxonomy built from what buyers actually shortlist, maintained by people with a commercial incentive to get it right. Field service management software sitting next to work order software and dispatch software tells you those are three separate buyer vocabularies, not synonyms.

Competitor navigation and footers. Crawl the header, footer and sitemap of your four closest rivals. Their /solutions/, /use-cases/ and /integrations/ URL slugs are keyword research someone else paid for. The footer is often more useful than the blog, because footer links reflect deliberate architecture rather than whatever a writer felt like publishing.

Sales call transcripts. Pull 20 recordings from Gong, Chorus or whatever records your calls, and search for the moment the prospect describes their current process. The phrase they use is almost never your category name. One payments company we know found buyers saying chase failed card payments while the entire marketing site said involuntary churn recovery.

In app search logs. If your product has a search box, it is a free panel of exactly your users typing exactly their problems. Export a quarter of queries. The top 40 are usually feature names, but the long tail is job phrasing you can build content around.

Support tickets and the help centre. Group the last 500 tickets by topic. Anything with a volume spike has a public search equivalent. This source also tells you where the help centre may already be competing with the marketing site for the same query, which matters later.

Reddit, Slack communities and private forums. Search the subreddits your buyers use for recommendations and alternatives to. The threads contain both the vocabulary and the objections, and they double as distribution once the page exists. The same reading habit powers a lot of what works on X for B2B SaaS, where the useful signal is the phrasing rather than the reach.

Sort everything into four intent tiers

Every candidate belongs in exactly one tier. The tier decides the page type, the priority and what you are allowed to expect from it in reporting.

Tier 1 is where a company under 5 million dollars of ARR should spend its first two quarters. The patterns and modifiers are catalogued in bottom of funnel keyword patterns, and the case for building those pages before the blog is made properly in bottom of funnel SEO for SaaS.

Tier 4 is the one people argue about. It earns links and it builds topical association, both of which matter. It also produces almost no pipeline directly, so it should be funded from a link building budget rather than a demand generation one, and judged on referring domains earned rather than trials. Treating it as a pipeline channel is how content programmes get cancelled in month nine.

Score every keyword on pipeline value, not volume

One formula, applied to every row before anything gets written.

Modelled monthly value = monthly searches x expected CTR at your realistic position x visitor to trial rate x trial to paid rate x ACV x gross margin

Then divide by effort. We use a simple effort proxy: months to reach the realistic position, multiplied by the referring domains the current top three hold. It is crude and it is still better than arguing about which term feels important.

Three rules that keep the model honest. Cap expected CTR at 20 percent even at position one, because AI Overviews and SERP features take a variable share. Use your realistic position, which for most sub DR 30 sites means position 8 to 15 on anything competitive rather than position 1. And use gross margin adjusted ACV, because an 85 percent margin and a 45 percent margin produce very different answers on the same traffic.

A worked example at a 12,000 dollar ACV

Three real keyword shapes for a project management product selling at 12,000 dollars a year with an 80 percent gross margin. One has 672 times the search volume of another. It finishes last.

The head term has the volume, the brand halo and the slide that impresses a board. It produces the least money, takes the longest, and needs a link budget most Series A companies do not have. The vertical term with 90 searches produces more than twice the head term’s value on two referring domains and four months of patience. That pattern repeats across almost every B2B SaaS category we have modelled, which is why vertical SaaS SEO is usually the fastest route for a company that sells into two or three identifiable industries.

Model your own version rather than trusting ours. The organic traffic forecast calculator takes the same inputs and produces a month by month curve, which is more useful than a steady state number when you need to tell a board when the line moves.

Screen for cannibalisation before the map gets approved

Two pages chasing one query is the most common self inflicted wound in SaaS SEO, and a keyword map is where it gets designed in. Catch it on the sheet, not six months later in the rankings.

The help centre clash deserves attention because it is specific to software companies. Docs and support content frequently outrank the marketing page for the product’s own feature names, which is fine for a customer and bad for a buyer who lands on an API reference instead of a product page. Decide which surface owns which intent and enforce it in the map.

The columns a writer can actually work from

A keyword map is a briefing document, not a research artefact. If a writer has to interpret it, two writers will interpret it differently. These are the columns we use.

The search intent evidence column does more work than its width suggests. One line saying top 5 are all listicles from review sites, no vendor pages stops a writer producing a product page that was never going to rank. A downloadable version with the formulas already in it sits at the SaaS keyword map template, and build your SaaS keyword map walks through filling one in with a real product if you would rather follow along than start from a blank sheet.

When the tool returns zero

Category creating products get nothing back from a keyword tool, and the standard advice at that point is to give up on SEO. That is half right. You cannot rank for a category name nobody searches. You can rank for four other things.

• The problem, not the product. Buyers search their symptom long before they search your solution. Size demand for the symptom phrasing instead.
• The adjacent category they currently shop in. If your buyers are solving this with spreadsheets or with an older tool class, that older class has volume and their alternatives queries are open to you.
• The job title plus task queries. How to forecast headcount as a cfo has demand even where the tool category does not.
• A free tool that produces the calculation people already search for. This also earns the links that everything else on the map needs.

Sizing latent demand is genuinely hard and we are not going to pretend a formula solves it. The practical test: if the total modelled value of every viable term is under 2,000 dollars a month, SEO is the wrong first channel and should be deferred until the category name has traction. Spend the quarter on outbound or community instead, and revisit in two.

Map the whole buying committee, including the people who block deals

Every guide tells you to map the buying committee and then produces four pages for the practitioner. The people who actually block a deal search too, and their queries are cheap because no vendor targets them.

Two of those rows are usually empty on a B2B SaaS site. Security queries carrying a company name have low volume and enormous intent, since nobody searches for a vendor’s SOC 2 status casually. A trust page that ranks for yourbrand security saves the sales team a week per enterprise deal and costs an afternoon to build.

Add a role column to the map so this stays visible. If every row says practitioner, the map is incomplete regardless of how well scored it is.

Three keyword types we stopped building for

Positions, so you can disagree with them deliberately.

Broad definitional terms outside your category. What is project management sends traffic and produces nothing. It is also the exact shape AI answers absorb most completely. Build glossary entries for terms inside your category where you have a product angle, and skip the rest.

Salary, job description and certification queries. They have volume, they rank easily, and the searcher is a job seeker. HubSpot’s decline is the cautionary version of this at scale: broad audience content detached from the product stopped earning its place once Google grew stricter about relevance to the site’s core subject.

Competitor brand terms with no switching modifier. Ranking for a rival’s bare brand name brings their existing customers looking for a login page. Target rival alternatives, rival pricing and rival vs instead, where the modifier carries the intent.

How often to rebuild it

Quarterly refresh, annual rebuild. The quarterly pass takes two hours: add new terms from Search Console, retire rows where the SERP shape has changed, and re-score anything affected by a pricing or conversion rate change. The annual rebuild takes two days and is only worth it after a positioning change, a new product line or a category shift.

One trigger overrides the calendar. If your ACV moves by more than 30 percent in either direction, re-score immediately, because every priority on the sheet was calculated against the old number and the ordering will have changed.

What to do next

Spend the first hour on Search Console positions 11 to 30 and 20 sales call transcripts, not on a keyword tool. Build 60 rows, score them, and delete everything below the top 25. Then ship the top five before adding anything to the list.

If you want the tooling decisions settled first, SaaS SEO tools covers what is worth paying for at each stage. Once the map is approved, the constraint shifts to authority rather than planning, which is where SaaS link building picks up. The whole sequence, from map to published pages to links, runs as four sessions in the SaaS SEO Sprint, and the strategic context sits on the SaaS SEO pillar.