SaaS Competitive Analysis: Read What Rivals' Customers Say

Most competitive analyses are a feature matrix. Ten competitors down the side, twenty features across the top, ticks and crosses in the middle. It looks thorough, and it answers the wrong question: it tells you what rivals built, not why their customers are unhappy.

The people who will pay you are already paying someone else, and they are unhappy about something specific. That something is written down in public, in their own words.

Why the Feature Matrix Misses It

A feature matrix compares what products do. Customers do not switch because a feature is missing from their current tool — they switch when a feature that exists, or a policy they accepted once, starts costing them something.

You can see this in the language. Nobody posts “the competitor lacks webhooks”. They post “I have to export a CSV and re-import it every week” or “they raised prices again and support never answers”. The first sentence is a specification for a feature. The second is a reason to leave.

So build the analysis out of the second kind of sentence.

Where the Complaints Live

You do not need privileged data for this. You need the places where paying customers talk to each other, or to the vendor.

• Product communities. The subreddit or forum dedicated to the tool you are comparing yourself with. People there are users, they are technical enough to explain the problem, and they are not being polite.
• Support forums and issue trackers. For developer tools, GitHub issues with many participants and no resolution are a map of what the vendor has decided not to fix.
• Review sites. Useful, but read them as a corpus rather than as ratings: look for phrases that repeat across different reviewers. Repeat phrasing is signal; a single angry review is noise.
• Switching threads. Search the tool’s name together with “alternative to”, “switched from” and “canceled my subscription”. These threads are the closest thing to a public list of churn reasons.

Keep the raw material verbatim: the sentence, who wrote it, and the URL. A paraphrase is your interpretation; the quote is evidence. If you cannot find the sentence again, you cannot show it to anyone later.

The Method

1. Pick two to four alternatives people actually pay for. Not twelve. You are reading deeply, not surveying widely.
2. Collect complaints, not features. For each alternative, gather a few dozen posts where a user describes friction. Ignore the ones that are just praise with a small criticism attached.
3. Group by cause, not by wording. “Too expensive”, “the per-seat pricing is brutal” and “I pay for five seats and two of them are idle” are one problem, not three.
4. Separate price from value. “Expensive” means the price model does not match how the customer uses the product. “Waste of money” means the product does not deliver what it promised. They need different answers.
5. Note the workaround. What do people do instead? A spreadsheet, a hired freelancer, a script they maintain themselves? A workaround is a rough specification of what to build.
6. Count independent authors, not posts. Ten posts from the same angry person is one voice. Five posts from five people in different communities is a pattern.

Reading a Complaint Correctly

Complaints are short, and the obvious reading is usually wrong.

• “It is too expensive.” The number is rarely the issue — the pricing model is. Per-seat pricing punishes exactly the growing teams who would otherwise expand.
• “It is too complicated.” This is almost always about the first hour, not the whole product. Nobody rejects a tool after mastering it.
• “They never ship what we ask for.” The vendor has a different priority list. That is your opening: a small team can take one ignored request and own it.
• “Support takes days.” Not a feature at all, which makes it the cheapest differentiator there is.
• “It used to be good.” Something changed: a re-design, an acquisition, a new pricing page. Find what changed and you find the migration.

And keep the counter-evidence in view. The people who are happy do not post. If a complaint appears constantly in a community of a few hundred users, that is a real pattern; if it appears twice in a product with a hundred thousand users, it is a rounding error. A competitive analysis that only collects complaints will always conclude the market is wide open.

Turning Findings into a Decision

At the end you want one paragraph, not a document. Something like:

Customers of the two mid-market tools repeatedly describe the same onboarding problem. Both vendors have raised prices in the last year, and the complaints about per-seat pricing appear across communities. Support response time is mentioned as a reason to leave. A small tool that does the core job, charges per workspace, and answers support within a day would not need to beat them on features.

Check that paragraph against four questions:

• How many independent people said it?
• Is the fix a product change, a pricing change, or a service promise?
• Can one person ship the first version?
• Is there evidence people are content with the current state, which would contradict you?

If two of the four questions have no answer, you have reading material, not an analysis.

Mistakes That Make the Analysis Useless

• Treating review sites as a survey. Reviews are selected, and the selection favours strong opinions in both directions.
• Comparing feature lists. It produces a roadmap for a worse copy of your competitor.
• Only analysing the market leader. The complaints about the leader are generic; the complaints about the number three are specific, and the number three is the one losing customers right now.
• Cleaning up the quotes. Once you smooth a complaint into your own sentence, it stops being evidence and starts being your assumption.
• Stopping at the first pattern. The first complaint you notice is usually the one you already believed.

FAQ

How many competitors belong in a SaaS competitive analysis?

Two to four for the reading, plus a glance at the rest. Depth beats coverage: a dozen posts each from four tools tells you more than one post each from twenty.

Is a review site a valid source?

As a corpus of phrasing, yes: repeated complaints across independent reviewers are worth counting. As a measure of how many people are unhappy, no — review sites over-represent people with strong feelings, and some are curated by the vendor.

What if every complaint is about price?

Then the gap is pricing, not features, and that is still a gap. Look at how the complaint is phrased: “expensive for what it does” and “expensive because of the seat model” point to different products.

Can I do this without talking to customers?

Yes, and you should start there — public complaints are unprompted, which is their advantage. Talking to people confirms what you found; it rarely finds it for you.

Reading hundreds of threads by hand is the reliable part, and the slow part. The same job in a few minutes is what a research report is for: it collects the complaints around your topic, keeps the quotes checkable against the original thread, and separates recurring problems from one-off noise. If you are earlier in the process, start with finding pain points you can build on or the manual version in competitor research for indie hackers. And before you commit to the gap you found, read willingness-to-pay signals — a complaint is not a budget.