Pricing Your Claude Skill: A Seller's Field Guide

A seller messaged me last month, furious. Her skill had sold eleven copies in three weeks at $7. Eleven copies. She'd spent two weekends building a skill that turns a messy CSV of expenses into a clean, categorized report ready for an accountant, and she was about to delete the listing and call the whole thing a failure.

I asked her one question: how much time does it save the buyer? About forty minutes per use, she figured, maybe twice a month. So we did the math out loud. Forty minutes, twice a month, for someone whose time is worth even a modest $45 an hour — that's roughly $60 of value every month, forever. She was charging $7 once. No wonder it felt disposable. It looked disposable. We relisted at $29. It outsold the $7 version in the first week.

That's the whole problem in one story. Pricing is the question every seller agonizes over and almost nobody approaches systematically. Here's the framework I've watched work.

Price from value, not from effort

How long it took you to build is irrelevant to the buyer. They never see your weekend. They see what the thing does for them on Tuesday at 4pm when they're drowning. The only anchor that matters is this: how much time or money does one purchase save the buyer in a month?

If the answer is "two hours," and their time is worth $50 an hour, you have a lot of room. A skill that saves $400 a month in labor is cheap at $40 — you're charging ten percent of one month's value, once, and they own it forever. That ratio should feel almost too generous to you. It won't feel that way to the buyer. To them it's a rounding error against what they get back.

The trap is that effort feels like it should count for something. It doesn't, not directly. A skill that took you forty hours but saves the buyer ten minutes a month is worth less than one you wrote in an afternoon that saves them a full day. The market prices outcomes. Build accordingly, and price accordingly.

The trust floor is real

Below a certain price, cheapness stops being attractive and becomes a liability. On a buy-once marketplace, buyers can't try it for a month and cancel. They commit, they download, they own it. That changes how they read the price tag.

Under roughly $10, a lot of buyers read low price as low confidence. The unconscious logic goes: if the person who made this won't ask for real money, why should I trust it with my actual work? A $4 skill doesn't look like a bargain. It looks like a gamble that wasn't worth pricing seriously. You're not winning the frugal buyer at $4 — you're scaring off the serious one.

This is the opposite of the intuition most first-time sellers carry in from consumer apps, where racing to the bottom is the whole game. Here, the floor protects you. Respect it.

Where things tend to land

• $9 to $19 — single-task utilities, nicely polished. One bounded job, done well. Impulse-buy range; the buyer barely thinks about it.
• $25 to $49 — skills that encode real expertise or untangle a genuinely annoying workflow. The buyer has felt this pain personally.
• $60 to $129 — agents and multi-step systems that run a process autonomously and replace a tool or a recurring chunk of someone's week.

These aren't laws. They're the bands where buyers' eyebrows stay down. A skill does one bounded task, so it lives in the lower ranges. An agent runs a multi-step process on its own, carries more weight, and earns a bigger number. If you're charging agent prices for a single-task skill, the description had better justify it hard.

A worked example, start to finish

Let's price something real. Say you've built an agent that takes a folder of raw podcast audio and returns show notes, timestamped chapters, three pull-quote candidates, and a drafted social thread. End to end, autonomous. A podcaster runs it on every episode.

Start with value. Doing all that by hand is easily ninety minutes per episode. A podcaster publishing weekly runs it four-plus times a month. At a freelance editor's rate of $40 an hour, that's about $240 of monthly value — and that's before you count the episodes they'd have skipped because the busywork wasn't worth it.

Price at five to ten percent of monthly value, once. That math points at the low-to-mid twenties per month of value — but this is a one-time purchase against value that recurs every single month the buyer keeps podcasting. So you anchor higher: this is an autonomous agent replacing a real chunk of weekly work, squarely in the $60 to $129 band.

Now the commission. Skillmint takes a marketplace cut per sale, roughly 20 to 30 percent. On a $89 sale at, say, 25 percent, you keep about $67. If that take-home feels thin for the value you're delivering, the answer is not to price at $94.71 to "cover the fee." The answer is $99 — clean, confident, and still a trivial expense against $240 a month saved. Forty sales of that agent is roughly $2,970 in your pocket, for a thing you built once.

That's the shape of every good pricing decision: value first, band second, round number last, commission folded into the round number rather than bolted onto it.

The round number ending in 9

Price with your take-home in mind, but never try to pass the commission to the buyer with an ugly number. $39 outsells $44.71 every time. The round number ending in 9 reads as a decision someone made on purpose; the precise number reads as a spreadsheet leaking onto the buyer. $19, $29, $49, $89, $99 — these look like confidence. $43.12 looks like you're nervous.

This is the cheapest lever you have. It costs nothing and it changes how the price feels. Use it.

A practical way to find your number

1. Write down the monthly value to the buyer, in dollars. Be honest, not flattering.
2. Anchor to the band your thing belongs in — utility, expertise, or autonomous agent.
3. Round to a clean number ending in 9.
4. Launch slightly higher than feels comfortable.
5. Watch for two weeks before touching it. Discounting later is easy and even fun. Raising the price on a thing people already bought is not.

The sellers who regret their pricing almost always regret going too low. I have never once met a seller haunted by the customers they lost at $49 who would have bought at $29. I've met dozens haunted by the margin and the credibility they torched at $7.

Here's the sharper version, because it's the thing nobody says out loud: underpricing is not humility. It's a failure of nerve dressed up as generosity. You are not doing the buyer a favor by charging $7 for something worth $60 a month to them — you're teaching them not to take it seriously, and teaching yourself that the work wasn't worth defending. Charge the number the value supports. Pick it, ship it, and let real sales tell you if you were wrong. They usually tell you that you were too low.