To find keywords your indie SaaS can actually rank for, skip the big category terms and look for the specific phrases your buyers already type when they have the problem you solve: the words they use in support emails and signup answers, Google autocomplete and "People also ask" questions, and, once you have a few impressions, the queries in Search Console. Then open the real results page for each phrase and only keep the ones where forums, thin pages or small sites are ranking. A new site with no authority wins narrow, high-intent searches first, and it wins them one page per intent. Below: why head terms are a trap, where real keyword ideas come from, how to match intent to a page type, how to read a results page instead of trusting a difficulty score, a small keyword sheet, how to group terms so your pages do not compete, how to pick what to do first, and a 14-day plan.
Why do big keywords fail a new indie SaaS site?
Most founders start the same way: type the category into a tool, see a big number next to something like "CRM for small business", and plan a page for it. Months later the page sits on page six.
Usually the page is fine. The term is the wrong fight for a site this young:
The results are already full. Broad category searches tend to show big brands, large review sites and long lists written by teams with years of links behind them. A new domain has none of that yet.
The intent is vague. Someone typing a two-word category could be a student, a journalist, a buyer at a large company or someone who wants a free spreadsheet. Even if you ranked, most visitors would not be your buyer.
It takes too long to learn anything. When the only targets are terms you cannot reach for a year, you get no feedback. You cannot tell if the writing is the problem, the page type is wrong, or the keyword was never winnable.
Small sites do better the other way round. Pick narrow phrases where the searcher has a clear job, the current results are weak, and your product is a real answer. Each brings fewer visitors, but closer to buying, and a handful of pages that rank beats one page that ranks for nothing.
One note before going further: ranking for your own product name is a separate job with its own fixes. If people searching your exact name cannot find you, start with getting your startup to rank for its own name before anything here.
A running example (made up)
The rest of the post uses a fictional product. Quaverlog is a made-up lesson scheduling and practice-tracking tool for independent music teachers running a small studio. It has a new domain, a homepage, a few docs pages and no blog. Everything about Quaverlog and the keywords and scores below is made up to show the method. None of it is real search data.
Where do real keyword ideas come from?
A keyword tool is a decent place to check ideas and a poor place to find them, because it shows what many people search, and many people are not your market. Better ideas come from places where your buyers already wrote the problem down in their own words.
Your customers' own words
Start with the text you already have:
Support emails and chat. Look at how people describe a problem before they know your feature name. A Quaverlog customer would not write "recurring event rules". They would write "how do I move one lesson without breaking the whole term".
Signup answers. If your signup asks "what made you look for a tool like this?", the answers are a keyword list. If it does not ask yet, add the question this week.
Sales and setup calls. Write down the exact sentence someone uses to explain why they are on the call. The first sentence is usually the search they did.
Copy phrases word for word. "Missed lesson policy" and "makeup lesson policy" may mean the same thing to you, and very different things to the person searching.
Google autocomplete and "People also ask"
Type the start of a customer phrase into Google and read what it suggests. Google's help page on how autocomplete predictions work says (as of Oct 2026) that predictions "reflect real searches that have been done on Google", and that the system also considers the language of the query, the location it comes from, trending interest and your own past searches. So treat suggestions as proof that a phrasing exists, not as a ranking of what is most popular. The same page says autocomplete does not simply show the most common queries and should not be compared against Google Trends. Try it in a private window too, so your own history does not shape what you see.
Then scroll the results page. The "People also ask" box and the related searches at the bottom show the follow-up questions, which are often better targets because they are more specific.
Search Console, once you have impressions
As soon as your site gets any search traffic at all, Search Console becomes the best keyword source you have, because it shows searches where Google already thinks you are relevant. In the Search Console Performance report, the Queries tab groups data by the search people typed and shows clicks, impressions, click-through rate and average position. Per Google's help page (as of Oct 2026), the default view covers the past three months, and average position is the position of the topmost result from your site.
The useful rows are queries with impressions but almost no clicks, especially ones you never wrote a page for. A page that answers that exact question directly is often the fastest win you will find. Google also notes that you might not see your own site when you search the query yourself, because results depend on time, place, device and search history, so trust the report over a manual check.
Competitor help centres and docs
Read the article titles in the help centres of the bigger tools in your space. They are a map of what customers struggle with. You are not copying anything, just noticing which jobs come up again and again and checking whether the results for them are weak. If you already publish docs, turning product docs into organic signups covers how to shape those pages.
Community threads
Forums, subreddits and niche groups where your buyers help each other are full of raw search phrases. Search them for the problem, not your category. The same question across several threads with no good answer linked is a strong sign.
A word on Google Trends
Google Trends is handy for comparing two phrasings or spotting seasonality. Just know what it shows. Google's FAQ about Google Trends data explains (as of Oct 2026) that the data is a sample, normalized and scaled from 0 to 100, and that searches made by very few people appear as "0". For the narrow phrases a new site should target, a flat zero in Trends does not mean nobody searches. It usually means the term is too small for Trends to show.
What kind of search is it, and what page should answer it?
Every phrase carries an intent: what the person wants to do next. The page type matters as much as the words. A long guide will struggle where everyone wants a template, and a product page will not rank for a "how to" question. These intent types matter most for a small SaaS:
Intent type | Made-up Quaverlog example | Page type that usually fits |
|---|---|---|
Problem-aware "how to" | how to reschedule one music lesson without changing the series | Answer-first guide or docs page that solves it, with the product as one option |
"X for Y" (tool for a specific audience) | lesson scheduling app for piano teachers | A landing page for that audience, with their words and their use case |
Alternative to a known tool | alternative to a spreadsheet for music students | An honest alternatives or comparison page |
Template or example | music studio makeup lesson policy template | A free, usable template on the page itself, no signup wall |
Integration or "with" terms | sync music lessons with Google Calendar | An integration page or docs page that shows exactly how it works |
Alternative searches deserve their own careful page, and the advice in writing a SaaS comparison page that ranks covers that. For the rest, the simple rule is: before you write anything, decide which row of this table the phrase belongs to.
How can you tell if you can actually rank for a keyword?
Tools give each keyword a difficulty score, which is an estimate from the tool's own data, and tools often disagree on the same term. A better test takes two minutes: search the phrase in a private window and read page one like a competitor would.
What you see on page one | What it likely means for a new site |
|---|---|
Forum threads, Q and A sites or community posts in the top five | An opening. People are asking, and no strong page answers it yet |
Thin pages, old posts, or pages that answer a slightly different question | An opening if you answer the exact question better and more directly |
Small sites and indie tools similar to yours | Winnable with a clearly better page |
A mix of one or two big brands and several smaller pages | Possible, but slower. Worth it only if the business value is high |
A wall of big brands, large review sites and huge directories | Skip for now. Come back when the site has more trust |
Results that serve a different intent than you planned (for example all templates, all videos) | Change the page type or drop the phrase |
While you are there, note which format wins and what the top results miss. If you cannot write a page that is clearly more useful than what ranks today, the phrase is not yours yet, whatever the difficulty number says.
This matches what Google itself says it rewards. Its guidance on creating helpful, people-first content asks (as of Oct 2026) whether content provides "substantial value when compared to other pages in search results", and lists producing "lots of content on many different topics in hopes that some of it might perform well" as a warning sign. Reading the results page is how you check the first question honestly.
What goes in a small keyword sheet?
You do not need a big tool to keep track of this. A spreadsheet with one row per phrase is enough, and it forces you to make a decision on each one. Here is what a few rows might look like for Quaverlog. All values are made up, and "value" and "winnable" are simple 1 to 3 judgements, not measurements.
Phrase (made up) | Where it came from | Intent | Page one today (made up) | Value 1-3 | Winnable 1-3 | Score | Decision |
|---|---|---|---|---|---|---|---|
music teacher software | Keyword tool | Vague category | Big brands and review sites | 2 | 1 | 2 | Skip for now |
how to reschedule one music lesson without changing the series | Support emails (several) | How to | Forum threads, one old post | 3 | 3 | 9 | Write first |
makeup lesson policy template for music teachers | Signup answers, autocomplete | Template | Small blogs, a PDF | 2 | 3 | 6 | Write second |
lesson scheduling app for piano teachers | Setup calls | X for Y | Mixed, two big tools | 3 | 2 | 6 | Landing page, later |
sync music lessons with Google Calendar | Support emails | Integration | Generic calendar help pages | 2 | 2 | 4 | Docs page |
Keep the "where it came from" column honest. Phrases from customers' own words usually beat phrases that only came from a tool. Add a "target URL" column once you assign a page.
How do you group keywords so your pages do not compete?
With 20 or 30 phrases, many will be the same question in different words. "Reschedule one lesson", "move a single lesson" and "change one lesson in a recurring schedule" are one intent. Three pages for them means three weak pages splitting attention. This is keyword cannibalization: your own pages competing for the same search.
The fix is simple and a bit boring:
Group by intent, not by words. Put phrases in the same group if the same page would fully satisfy all of them. A quick check: search two of the phrases. If the results pages look mostly the same, they belong together.
One group, one page. Pick the clearest phrase as the main one for the title and first heading, and use the others naturally in subheadings and the answer.
Split only when the intent changes. "Reschedule one lesson" (a how-to) and "makeup lesson policy template" (a template) are different jobs, so they get different pages even though they are related.
Check before you write. Search your own site for the topic first. If a page already covers the intent, improve that page instead of adding a new one.
In your sheet, add a "group" column next to "target URL". Every group should end up with exactly one URL. If one group points at two URLs, merge them, and link related pages (the how-to and the template) to each other.
Which keywords should you go after first?
With groups in place, ordering is quick. Score each group on two things, using the same 1 to 3 scale as the sheet:
Business value. If someone searching this found your page, how likely is it that they are your buyer and your product helps right now? A teacher asking how to reschedule one lesson is in the middle of the exact pain Quaverlog solves. A student looking for a practice log is not the buyer.
Winnability. From your results-page read: are there openings on page one, and can you write something clearly better?
Multiply the two. Anything scoring 6 or more is a candidate for the next few weeks. Anything at 1 or 2 waits. Between two groups with the same score, pick the one where your page can show the product doing the job, not only describe it, because that is the page most likely to turn a visitor into a signup.
Then limit yourself: three to five finished pages over the next month beats twelve half-done ones. Google's helpful content guidance is direct on one related point: it asks whether you are writing to a particular word count because you heard Google prefers one, and answers "No, we don't." Write each page as long as the job needs and no longer.
What are the common traps?
Trusting the volume column. Tool volumes for small phrases are estimates, and many tools show zero for searches that clearly happen. If customers keep asking it, it is real.
One page per phrase. Writing a page for every variation splits your own effort and confuses which page should rank. Group first.
Researching forever. A sheet with 200 rows and no published pages teaches you nothing. Ship the top few, watch Search Console, then adjust.
Writing for the keyword, not the person. Repeating the phrase does not make a page useful. Answer the question in the first paragraph, then show how.
A 14-day plan
The day numbers are a planning guess for one founder doing this alongside everything else, not a measurement.
Days | What to do | Done when |
|---|---|---|
1-2 | Collect phrases from support emails, signup answers and call notes. Add a "what made you look for a tool like this?" question to signup if it is missing | 20 or more customer phrases copied word for word into a sheet |
3 | Run each phrase through autocomplete and "People also ask" in a private window; add the useful follow-up questions | Sheet has source and intent columns filled in |
4 | If you have impressions, export Search Console queries and add the ones with impressions but few clicks | Search Console rows added and marked as such |
5-6 | Read page one for every phrase; fill "page one today" and a 1 to 3 winnable score | Every row has a winnability judgement |
7 | Group phrases by intent, assign one target URL per group, score business value and sort by value times winnability | A ranked list of groups with one URL each |
8-12 | Write or improve the top two pages: answer-first, the right page type, the product shown doing the job, links between related pages | Two pages live and linked from somewhere relevant on your site |
13-14 | Check both pages are indexed with URL Inspection in Search Console, and note the target queries to check in a month | Pages indexed or requested, plus a dated note |
FAQ
Do I need a paid keyword tool to do keyword research?
Not to start. Customer words, autocomplete, "People also ask", community threads and a careful read of the results page get a new site most of the way. A tool helps later to find related phrases, but it should not be where your ideas come from.
Should I target keywords that show zero search volume?
Often, yes, if customers keep asking the question and the results page has openings. Tool estimates are rough for small phrases, and Google Trends shows low-volume terms as "0" by design. Check that the page would help a real buyer, then test it.
How many keywords should one page target?
One search intent per page. That intent can cover many phrasings, so a single page may show up for dozens of related queries. If two phrases need different answers or different page types, they need different pages.
How long does it take a new site to rank for a keyword?
There is no fixed timeline, and anyone quoting an exact one is guessing. Narrow phrases with weak results tend to move first. Watch impressions and average position for your target queries in Search Console and give each page a few weeks before judging it.
Where should you start this week?
Open your last 20 support emails and signup answers and copy out every sentence where someone describes the problem in their own words. Search the three most common ones in a private window. If page one is forums and thin posts, you have found your first page. And when that page is live and you want more people to see the product behind it, a listing on IndieHunt puts it in front of founders and early adopters who are already looking for new tools.
