SaaS Keyword Research: A One-Person Process That Works

November 8, 202613 min read

On this page

TL;DR: SaaS keyword research works when you start from the buyer's own words and rank keywords by buying intent before volume. In a 2023 Grow and Convert case study for Geekbot, bottom-of-funnel articles converted at 4.78% against 0.19% for top-of-funnel ones, so treat the volume number as one input, not the verdict (growandconvert.com).

| Step | What you do | What you should see | |---|---|---| | 1. Define the buyer | Write two sentences: who they are, why your product fits | Both sentences written without vagueness | | 2. Collect their phrasing | Pull verbatim phrases from calls, tickets, reviews, threads | A phrase bank in the buyer's words | | 3. Brainstorm three types | Category, competitors and alternatives, jobs to be done | A seed sheet with three buckets | | 4. Verify intent | Read the SERP for each candidate | Pages that sell or compare, not only explain | | 5. Check volume and KD | Tune both together in a tool | Keywords that are reachable and still worth it | | 6. Map and re-run | Match each keyword to a page, repeat every three to six months | One target page per keyword | | Verdict | Geekbot: BOTF 4.78% vs TOTF 0.19% conversion | Rank by intent first |

Who is the keyword for? Write the buyer definition first

Prerequisites: a live product (or a clear draft of one), a blank spreadsheet, and an afternoon. Free tools are enough; one source states the process can be followed with free alternatives to Ahrefs (madx.digital).

Write two sentences. The first covers segment, company size, role and the situation the buyer is in right now. The second says why your product is the obvious fit for that person. You should see: two sentences a stranger could repeat back. If you cannot write both, stop here, because keyword research will only make the vagueness look organised (platypusseo.com).
Optional: split by role. If you already have customers, segment them by company size, industry, job title and common pain points (linkedin.com). Then map keywords to the people on the buying committee (C-suite, IT, department heads or end-users) and their concerns. Gartner reports that 64% of B2B purchase decisions involve four or more people, so one persona rarely covers the whole sale. You should see: a short list of roles, each with one concern.

If you are a solo founder with a handful of customers, skip step 2 and revisit it once you have more data.

Where do you find the words buyers actually use?

Build a phrase bank. Keep every phrase verbatim; do not tidy it into marketing language.

[ ] Sales calls and demo recordings
[ ] Support tickets
[ ] G2 and Capterra reviews, for your product and competitors (platypusseo.com)
[ ] Community threads
[ ] Your site's internal search queries, which show what visitors want once they arrive and flag what your content is missing (questiondb.io)
[ ] Prompts buyers type into ChatGPT, alongside what they type into Google (platypusseo.com)

The last item matters more each quarter. Peec's analysis of 20 million ChatGPT fan-out queries found average length roughly doubled from about six words to about twelve between October 2025 and January 2026 (platypusseo.com). A G2 survey found 22% of buyers' first chatbot prompts are requirements-based: the situation first, the category or competitor second. So save full sentences that describe a situation, not just two-word terms.

Then get marketing, sales and support in one room. They talk to customers directly and will surface ideas a tool never will (questiondb.io). If you ask ChatGPT for seed ideas, label them unvalidated, since none of it is backed by real search data yet.

You should see: a sheet of raw phrases, none yet checked for volume.

Which three keyword types should you brainstorm?

Grow and Convert's framework names three types of buying-intent keywords: category keywords, competitors and alternatives, and jobs to be done.

| Bucket | Example shape | Pitfall | |---|---|---| | Category | "CRM software for manufacturing" instead of "CRM software" (linkedin.com) | The super obvious terms are often highly competitive (growandconvert.com) | | Competitors and alternatives | "[competitor] alternatives", "[you] vs [competitor]" | A competitor targeting a term does not make it relevant to you (growandconvert.com) | | Jobs to be done | Question-style phrases from People Also Ask or AnswerThePublic | Easy to drift into purely informational topics |

Category. If you can go beyond the first half dozen terms that describe your product, you can probably find valuable keywords with lower competition (growandconvert.com). Industry modifiers help: Ahrefs states that keywords with industry modifiers have lower competition but higher relevance (linkedin.com). Find modifiers in Google Keyword Planner or Ahrefs' Keyword Explorer, and check which ones competitors' product pages use successfully.

Competitors and alternatives. Six competitor comparison articles for Grow and Convert's client Circuit brought in 149 organic signups over two years (growandconvert.com). Small page count, real signups.

Jobs to be done. Click into a few People Also Ask questions and more related questions keep surfacing (questiondb.io). AnswerThePublic is another source for long-tail questions about your product's features (linkedin.com).

You should see: a seed sheet with each phrase tagged by bucket, and at least a few modifier variants of your main category term.

Which keywords actually earn signups: intent or volume?

Here is the worked example. For the Geekbot client, Grow and Convert compared articles aimed at bottom-of-funnel (BOTF) keywords with those aimed at top-of-funnel (TOTF) ones (growandconvert.com).

| | BOTF articles | TOTF articles | |---|---|---| | Articles published | 22 | 42 | | Organic traffic | 28,190 pageviews | 204,303 visitors | | Conversion rate | 4.78% | 0.19% | | Conversions | 1,348 | 397 |

The TOTF set had nearly twice as many articles and nearly 10X the organic traffic, yet produced fewer conversions (growandconvert.com). The reason is simple: very few readers attracted by top-of-funnel content will ever be in the market for your product.

Decision rule: skip generic high-volume terms that pull in unqualified traffic. Long-tail keywords with lower volume but real conversion intent tend to win, because the searchers already know roughly what they want (questiondb.io).

When this is the wrong call: if your goal is brand awareness or building an email list rather than signups, top-of-funnel terms have a job. For a founder who needs paying users, they come second.

How do you verify buying intent before you write anything?

Checking the SERP is the second step in Grow and Convert's process. Do it for every candidate on your sheet.

| Step | What to do | What you should see | |---|---|---| | 1 | Search the term in a private window, set to your target country | Results a typical buyer there would get | | 2 | Read the first page | Product pages, comparison pages, "best X for Y" lists, alternatives pages | | 3 | Look for an AI Overview | Present or absent, noted in a column | | 4 | Label the row | buy, compare or learn |

If the first page is mostly definitions and how-tos, the term is informational. Label it learn and move it down.

Reading the AI Overview signal

On Google, Seer Interactive's study of 2.43 billion impressions saw an AI Overview for 95.4% of comparison-format pages (platypusseo.com). Transactional queries are much less exposed. Seer saw them on 4.9% of those queries, and an experiment by Agarwal and Sen saw them on 6% of transactional searches against 53% of informational ones. The cost of showing up is real: across 300,000 keywords, Ahrefs put the click loss at 58% of position-one clicks wherever an AI Overview appears.

So a learn term with an AI Overview is the weakest row on your sheet. A buy term without one is the strongest.

Spread the sheet across stages

Do not let every row sit at the same level of buyer certainty (platypusseo.com). Keep a mix of problem-aware terms, comparison terms and ready-to-buy terms, and fill the buy rows first.

Our position on comparison pages

Write them anyway. Comparison pages are where a buyer actually chooses, so judge them by signups rather than clicks. Semrush found that almost 90% of the time, the pages ChatGPT cites sit at organic position 21 or lower, and that AI-referred visitors convert at around four times the organic rate (platypusseo.com). Ranking first is not the only route to being seen.

You should see: every row on your sheet marked buy, compare or learn, with an AI Overview flag beside it.

How do you mine competitors without copying them?

Pick three to five competitors, the ones that actually show up in your SERP checks (linkedin.com).
Run a gap tool. In Semrush's Keyword Gap, enter your domain plus up to four competitor domains, confirm your target country, and click Compare (questiondb.io). The tool finds keywords they rank for that you don't (linkedin.com).
Select Missing in the keyword table to see the gaps (questiondb.io).
Filter for relevance. Just because a competitor targets a keyword does not make it relevant. If small businesses aren't your market, don't target "X software for small businesses" (growandconvert.com).
Do not assume rankings equal revenue. Lots of first-page rankings do not necessarily mean those articles bring in conversions (growandconvert.com).
Mix difficulty. Target the competitive keywords right away, since ranking against big names takes longer. For faster results, add a few less competitive keywords with high buying intent, but don't ignore the competitive ones (growandconvert.com).

That last point answers whether a small site can compete with big names: yes, on the terms they ignore, and over time on the contested ones, as long as you start the contested ones now.

You should see: a trimmed gap list where every row passes the "is this my buyer?" test.

How do you read volume and KD without trusting them blindly?

Treat volume and KD as a pair. Adjust both until the keywords are within reach and still worth the effort (questiondb.io). There is no universal cutoff, so set yours by what your own site can plausibly rank for.

| Signal | Keep or cut? | Why | |---|---|---| | High intent, high volume | Keep, write first | The best case | | High intent, low volume | Keep, write next | Grow and Convert prefers it over low intent with high volume once the best case runs out (growandconvert.com) | | Low intent, high volume | Cut or move down | Pulls in readers unlikely to buy | | Zero volume in the tool, but the phrase came from a customer call and the SERP shows buying intent | Keep | See below |

Why zero volume is not a veto

The most specific B2B phrases often show "zero volume" in Ahrefs or Semrush (platypusseo.com). Let the tool inform the call instead of making it. If a buyer used the exact wording on a call and the SERP looks like a buying page, keep the row.

Long-tail terms also match how buyers search. A Conductor study found that 48% of B2B buyers use long-tail keywords during research (linkedin.com).

A simple score

Give each term a search priority score (madx.digital). Make it a plain one: intent label first, reachable KD second, volume only as a tiebreaker. Paid suites are optional here. Our roundup of the best free SEO tools shows what the free options do well and where they leave gaps.

You should see: a ranked list where the top rows are high intent, reachable, and phrased the way buyers talk.

What do you do with the list, and when is this process the wrong choice?

Map and schedule

Match each target keyword to a page you already have, or flag it as a new page (madx.digital). Poor targets burn time and budget. Review the list every three to six months, or sooner if customers or the market visibly change (questiondb.io).

| Row on your sheet | Action | |---|---| | Matches an existing page | Update that page | | No matching page | Add to the new-page queue | | Labeled learn, no AI Overview | Schedule after the buy and compare rows |

When this is the wrong process

You cannot write the buyer definition. Fix positioning first, because keyword research on a vague audience only produces a tidy-looking list of the wrong terms.
Your buyers research mainly in chatbots. G2's March 2026 survey of 1,076 B2B software buyers reported that 71% use AI chatbots to research software (platypusseo.com). On Ahrefs' own site, AI search accounted for 0.5% of visits yet 12.1% of signups. If that looks like your audience, add prompt-level tracking to the keyword work instead of replacing it.

A bounded note on tooling

A recent DispatchSEO release made keyword research and rank checks run in your site's own country and language instead of always asking American Google (What's new · DispatchSEO). Details are in the changelog. The process above works without it.

Your next action: write the two-sentence buyer definition today, then open a blank sheet and start the phrase bank.

DispatchSEO: Your coding agent, running your SEO Keyword research, one SEO guide a day as a pull request, rankings and Search Console tracked daily.

Frequently asked questions

Can I do SaaS keyword research without paying for Ahrefs or Semrush?

Yes. One source states you can follow its process using free alternatives to Ahrefs [3]. Google Keyword Planner can surface industry modifiers [4], and People Also Ask is free to click through [5]. The free route costs you speed, not method, so lean harder on your own customer phrases.

How many competitors should I include in a keyword gap analysis?

Start with your top three to five competitors [4]. The Semrush Keyword Gap tool takes your domain plus up to four competitors [5], so a set of five fits in one run. Add more only if the first run gives you too few relevant gaps.

Do I still need keyword research if my buyers ask ChatGPT instead of Google?

Yes, because the work is now about phrasing as much as ranking. In the G2 survey, 69% of buyers said a chatbot led them to a different vendor than they had planned [2]. A 2026 Wynter survey of 101 mid-market SaaS CMOs found 84% now use AI tools for vendor discovery, up from 24% a year earlier [2]. Your phrase bank of buyer situations is what lets you write pages those tools can cite.