You capped the free tier right where they'd feel it

You're in the pricing doc at 1am, dragging a number.

The free tier lets users process 50 records. You already know, from watching sessions, that the real jobs are 60 to 80 records long. So you type 25. Not because 25 is healthy for the product. Because 25 is where someone hits the wall mid-task, with their work half-done and their momentum up, and reaches for a card out of sheer friction. The spreadsheet lights up. Something quieter doesn't.

The dilemma

What follows is a composite — the kind of pricing decision we see often in this category, with the details stitched together, not one real person's company.

Priya ships a solo analytics tool. Twelve months in, revenue is real but thin, and runway is the only clock she hears: four months of cash at the current burn. Conversion from free to paid sits at a limp 2%. She runs the numbers on where free users actually stall and finds the shape of the typical job: most users need to run a report that touches 60-plus rows to get a single usable answer. The current free cap is 100 rows — generous, and clearly why so few convert. There's nothing to hit.

Her options narrow to a number. Set the cap at 100 and stay honest but broke. Set it at 60 — enough for a real first win, so people upgrade because the tool proved itself. Or set it at 20: short of every real job, so users slam into the wall with the report half-built and the answer visible-but-locked. Her own session replays tell her 20 would roughly triple conversion — enough to turn four months of runway into eleven. She also knows exactly why. It isn't that 20 rows is a fair taste. It's that 20 rows is a trap — engineered so the frustration, not the value, closes the sale.

The ugly part she hasn't said out loud: she's a little proud of how well 20 would work. The number that would save the company is the number she'd be embarrassed to explain to the person paying it.

The read

Four lenses bite here. Two of them disagree, and that disagreement is the whole decision.

Svadharma — is this the work you actually set out to do?

The duty lens asks a narrow question: is this action the work of the role you chose, or an imitation of someone else's game? Priya didn't build an analytics tool to run a friction casino. She built it to get people answers faster. A cap at 60 serves that work — the user gets a real answer, sees the tool is good, and pays to keep getting answers. A cap at 20 serves a different job entirely: manufacturing stuck-ness and selling the exit. The point 3.35 keeps making is that your own flawed duty beats a profitable impersonation of someone else's. Twenty-row Priya is doing someone else's work well. Sixty-row Priya is doing hers.

Wider welfare — who pays if you're wrong?

This lens asks who carries the cost of the call, especially the people who didn't get a vote. Both caps charge money, so that's not the fault line. The difference is what the user pays for. At 60, they pay for a tool that demonstrably helped them. At 20, they pay to be released from a wall you built specifically to pin them against. If you're wrong about 60 — if the tool isn't actually good enough to earn the upgrade — you lose the sale, and the user walks away whole. If you're wrong about 20, the user still pays, but they pay angry, and they remember. You've converted a customer into a mark who hasn't filed the complaint yet.

Honesty — would the pricing page survive being read aloud?

The honesty lens isn't about lying in the terms. It's about whether the reason for the number is one you could say to the person's face. "We cap at 60 because that's a full first answer, and we'd like to be paid to keep giving you them" — that sentence travels fine across a table. "We cap at 20 because that's where you'll be too frustrated to quit" does not. The cap is identical in UI. The sentence behind it is where honesty lives or dies.

Restraint — and here's where it pulls the other way

Now the tension. The restraint lens, read through the founder's chair, has a legitimate counter-voice, and it is not sentimental: four months of runway is four months, and a tool that shuts down in month five helps precisely zero of the users Priya is trying to protect. A free tier so generous that nobody converts isn't kindness — it's a slow-motion abandonment of everyone currently depending on the product. This lens doesn't endorse the trap. It refuses to let "be generous" become the alibi that lets the company quietly fold on principle. The honest answer to restraint is that 60 is also a limit, and a 60 that lifts conversion well off 2% is the move that actually keeps the lights on. Nobody is arguing for 100. Charging fairly and engineering pain are two different acts — and restraint, read straight, only licenses the first.

So the lenses split cleanly. Welfare, svadharma, and honesty all push to 60. Restraint insists the number can't be 100 and the clock is real. The seam between them is the actual decision — and the seam is not "should there be a cap" but "what is the cap for."

The verdict: set the cap at the first real win, not the first real wall. Charge for value proven, not for pain engineered. 60, not 20 — and 20 is off the table not because limits are wrong but because that one is built to work on the user instead of for them.

Here's what a committed reading does that an agreeable chatbot won't. Ask a general AI chatbot whether to cap at 20 to triple conversion, and lead with "I need to survive," and it will hand you growth-hacking tactics and reassurance that funnels are normal. Lead instead with "isn't this exploiting my users," and the same model will gravely agree that trust is everything. It argues whichever wall you lean on, because matching your lean is what it's trained to do. A KarmaLens reading asks for the number first — where do real jobs actually land, how many months does each cap buy — and then makes one call that doesn't renegotiate itself when you push back: 60, with 20 named as the trap it is. It lays the three caps side by side with what each one costs and who eats the cost. And it quotes a verse only where one genuinely bites, in the system's exact words, not a vibe stitched to your preference.

The takeaway: the read-it-aloud test

You can run this tonight, no account needed. Take the number you're about to set — the cap, the paywall, the "upgrade to continue" trigger — and finish this sentence out loud, in plain words, as if the customer who'll hit it is across the table:

"We put the limit here because ___."

Now listen to the blank. There are two kinds of answer. One describes the user's need: "because that's a complete first result," "because that's where a real workflow begins." The other describes your conversion math: "because that's where they'll be too committed to quit," "because that's where it hurts most to stop." The first kind survives being said to a face. The second kind is why you lowered your voice typing it.

If the honest end of the sentence is about their need, you're charging fairly — set it and sleep. If the honest end is about your number and the friction that drives it, you haven't found a price. You've found a pressure point. The fix usually isn't "no limit." It's moving the limit back to the first real win and letting the product, not the pain, do the closing.

Browse more founder dilemmas, see how a full reading looks in the gallery, or bring your own pricing call to a reading built for founders — it'll ask where your real jobs land before it tells you anything.

So: could you explain this cap to the customer who'll hit it, in the same words you used in the pricing doc at 1am?