Pricing a tiny tool: three numbers I tried, and what happened

Free, $5, $9. Three prices for the same tool, one week each. The differences were larger and less flattering than I expected.

Because I could not decide what to charge for the CSV-to-summary tool, I did what any confused indie hacker with too much time does. I ran a completely uncontrolled, unblinded, non-statistically-valid experiment on myself and my traffic for three weeks. Here is what happened.

Week 1: free

I put the tool up with a "free during beta" banner and a button that said "Get free access." Fourteen signups in the first three days. Two of them uploaded a CSV. Zero of the fourteen replied to any of my three onboarding emails. Two of them wrote to me with support requests inside the first week — one polite, one demanding an integration with software I had never heard of.

Signal: at zero dollars, I could not tell whether anyone actually valued the thing. Everyone signed up; almost nobody used it. Support emails were the same volume as when I later had paying customers, but with no revenue to justify the time.

Week 2: $5 one-time

I changed the landing page to "$5 one time, use it forever." Eight signups over the week. Three converted to paying, which was a much better ratio than free, though of course a lower total. Support emails from paying customers were noticeably calmer. Nobody asked for the software-I-had-never-heard-of integration.

Signal: paying customers are self-selecting. Even a small amount of money changes who shows up. And a one-time price felt to me, and clearly to the market, like a lower bar than a recurring one, because it was.

The point of a small price is not the money. It is the sorting.

Week 3: $9/month

Same tool, same landing page, price changed to "$9/month, cancel any time." Six signups. Four converted. Two of those four were the same people who had bought the $5 one-time version the previous week — I sent them a private link at their old price, and they upgraded voluntarily "because it's easier to expense a subscription than a one-off." I did not know that was a real reason people mentioned. Now I do.

Signal: recurring pricing selected for people who used the tool at least monthly — which is exactly who I want as a customer. The absolute number of signups went down; the quality went up. Support emails dropped further and shifted from "how do I" to "what if you added."

What I did not test properly

Every honest self-report deserves a list of things it did not actually measure. Here is mine:

• Three weeks is not statistical. The traffic was small, and the messaging changed with the price, which contaminates the comparison. Take the specific numbers with a fistful of salt.
• I did not try a higher price. My hunch is that the same tool at $19/month would have done better than $9 for this segment (self-employed and small-firm accountants), but I did not run it because I wanted to lock in a recurring number and keep shipping.
• I did not try annual pricing. That is on the list.

What surprised me

Three things surprised me, in ascending order of importance.

The smallest surprise: free users produced more support work per dollar of revenue than any other cohort, by a very wide margin. If you have ever wondered why "your free tier is your worst customer" is a cliché — it is because it happens to be true, at least for me, at least for this shape of tool.

The middle surprise: the $5 one-time price felt, in my chest, like it was worth less than the $9 subscription, even though $5 was more money than the first month of the subscription. People upgraded from one-time to monthly at my suggestion without complaint. Monthly framing was, for this audience, easier to say yes to.

The biggest surprise: charging money made me a nicer person to my users. When I had free users, I resented the support requests. When I had paying users, I welcomed them, because each one was information about how to keep them. Nothing about the tool changed; only who was on the other side of the email did.

Where the price is now

$9/month, cancel any time. Annual coming this quarter. The banner that used to say "free during beta" now says nothing — the tool is just the tool, priced like a normal thing, and it turns out normal things sell fine.