How to validate a SaaS idea before you build it

To validate a SaaS idea before building, find people who already ask for it in public, talk to ten of them about how they solve the problem today, and then ask for a commitment: a call, a pilot or a pre-order. An idea is valid when strangers give you time or money, not when friends say it sounds useful.

Most advice on this topic stops at "talk to users" and "make a landing page". This playbook gives you the order of the steps, what to write at each step, and the number that tells you to go on or to stop. It takes two to three weeks of evenings, and you write no product code until step 5.

Step 1: Write the problem as one sentence about one buyer

Validation fails when the idea is too wide to test. "A tool for small businesses" gives you nothing to search for and nobody specific to ask. Write one sentence in this shape: who has the problem, when it happens, and what it costs them today.

• Who: one job title in one kind of company, for example the office manager of a dental clinic with two to five chairs.
• When: the moment the problem shows up, for example when a patient cancels the same morning.
• Cost today: what they lose or what they do by hand, for example an empty chair and twenty minutes of calls to fill it.
• Current fix: the spreadsheet, the agency, the competitor or the habit they use now.

If you cannot fill the last line, that is your first finding. A problem with no current fix is often a problem nobody pays to solve.

Step 2: Find the people who already ask for it

Before you ask anyone a question, look for people who asked it themselves. Search Reddit, Hacker News, X and Threads for the words your buyer uses, not the words you use: "how do you fill last-minute cancellations" rather than "dental scheduling automation". Each thread where someone describes the problem in their own words is a data point, and each reply under it shows you the fixes people already try.

Keep a simple log with four columns: the link, the exact words of the problem, the fix they use now, and whether anyone mentioned paying for something. Twenty threads is a good first target. If you cannot find five after an evening of searching, the problem is rare, or your buyer calls it something else. Try their words before you drop the idea.

This is also where many founders get banned. Asking questions in a community is fine. Dropping a link to a signup page is not. Our guide on how to promote your SaaS on Reddit without getting banned covers the rules of each subreddit, and the same habit of reading before posting applies on every platform. LitePal does this search for you: paste your site and it lists the public threads where people ask for a product like yours, with a reply drafted for each that you post from your own account.

Step 3: Talk to ten people about their past, not your idea

Reply to the people from your log, or write to them, and ask for fifteen minutes. In the call, ask about what they did the last time the problem happened. People describe their past well and predict their future badly, so "would you use this" gives you polite noise.

• Tell me about the last time this happened. What did you do?
• What did that cost you in time or money?
• What have you tried to fix it? Why did you stop?
• Who else is involved when it happens?
• If you could fix one part of it, which part?

Do not show a demo and do not pitch. Write down their exact words, because those words become your headline later. After ten calls, count how many people described the same problem without your help. Seven out of ten is a strong signal. Three out of ten means your sentence from step 1 is wrong, and you go back and rewrite it.

If you write to strangers by email to ask for these calls, the rules for cold email still apply to you. Google tells senders to keep spam rates reported in Postmaster Tools below 0.3%. That is 3 complaints in 1,000 messages, so a short, personal note to a small list is safer than a big blast. Our post on how to keep cold emails out of spam explains the setup.

Step 4: Ask for a commitment you can count

Interest is cheap. To validate a SaaS idea you need people to give up something: their time, their data or their money. Pick the strongest ask the stage allows.

• A second call with their manager or partner present.
• A pilot where you solve the problem by hand for one week.
• A pre-order or a paid pilot at a price you name on the call.
• A letter of intent that names the problem and the date they want it fixed.

Doing the work by hand is not a trick. It shows you every step the product must do, and it shows you which steps the buyer does not care about. If nobody accepts any of these asks after ten good calls, you have learned that cheaply. That is a valid result too.

Step 5: Test the words with a landing page

Only now build a page. Use the buyer's words from step 3 as the headline, one sentence on the current fix and its cost, and one action: book a call or join the pilot. Then send traffic you can trace, for example the people from your log and your replies in threads.

Know what a normal result looks like before you judge it. Unbounce's 2024 benchmark found that the median conversion rate for SaaS landing pages is 3.8%, compared to 6.6% for all industries. That benchmark comes from 57 million conversions across over 41 thousand landing pages and counts every kind of conversion, so treat it as a floor for a page that asks only for an email.

A worked example. Say a scheduling tool for dentists sends 400 visitors to its page from threads and replies. At the 3.8% median, 400 × 0.038 = 15.2, so about 15 visitors would leave an email. If the page gets 30, the words land twice as well as a typical SaaS page. If it gets 4, the headline or the audience is wrong, and you change one of them before you send more traffic. With only 400 visitors, treat the result as a direction, not a final answer.

Step 6: Decide, then show something people can try

Put the numbers side by side: threads found, calls held, people who named the problem unprompted, commitments, and page conversion. Go on when strangers committed and the page beat the median. Rewrite the problem sentence when calls went well but nobody committed. Stop when you could not find people who ask for it.

When you go on, build the smallest version that solves the problem from step 1, and put it in front of people. Hacker News is a good place for a first test, but it has firm rules: Show HN is for something you've made that other people can play with, and the same page says not to post landing pages or fundraisers. The site guidelines also ask you not to use HN primarily for promotion. Our guide on how to launch on Hacker News walks through the post itself.

A checklist to validate a SaaS idea

• One sentence: who, when, cost today, current fix.
• Twenty public threads logged in the buyer's own words.
• Ten calls about the past, with no pitch and no demo.
• Seven of ten people named the problem without your help.
• At least one commitment: a pilot, a pre-order or a second call.
• A landing page with the buyer's words, measured against the 3.8% median.
• A decision written down: go on, rewrite or stop.

Start with step 2 today. Paste your site into LitePal to see the public threads where people already describe the problem you want to solve, each with a drafted reply you can post yourself.