Why We Went One-Time Purchase Instead of Subscriptions

Somewhere in a pitch meeting last year, an investor drew the classic chart on a whiteboard: the smooth, up-and-to-the-right line of monthly recurring revenue. Then he drew a jagged, spiky line next to it. "This is you on one-time purchases," he said, tapping the spiky one. "Don't be the spiky line."

We're the spiky line. On purpose. Nearly every experienced person who looked at Skillmint told us the same thing: make it a subscription, recurring revenue is the holy grail, predictability is everything. They weren't wrong about the math. They were wrong about us. Here's the full reasoning, including the parts that cost us.

The skill runs on your machine, not ours

This is the load-bearing fact, so we'll start here. When you buy a skill on Skillmint, you download a file. That file runs locally, inside your own Claude instance, on your own hardware, using your own tokens. We host nothing. We execute nothing. We don't sit in the middle of every request watching a meter tick.

That detail quietly decides the whole pricing model. A subscription is the right shape when you, the seller, have an ongoing cost: servers humming, bandwidth flowing, a model burning GPU cycles every time someone hits the button. Charging monthly makes sense because you're paying monthly. None of that is true for us. Once you've downloaded a skill, our marginal cost of you using it ten thousand more times is exactly zero.

So what would we be charging you for, month after month? Not compute, because there isn't any on our end. Not hosting, because the thing lives on your disk. A monthly fee with no underlying cost to justify it isn't a service. It's rent on something you already took home. We couldn't write the sentence that explained the recurring charge without it sounding like a shakedown, so we didn't build the charge.

What a subscription actually feels like when you stop paying

Let me make this concrete, because "recurring revenue" sounds clean on a whiteboard and feels like something else entirely when you're the one being billed.

Imagine you're a freelance contract reviewer. In January you buy access to a skill that reads a lease, flags the predatory clauses, and drafts plain-English summaries for your clients. It's great. It saves you two hours per contract. You pay your fourteen dollars a month and you don't think about it.

Then work slows down in the summer. You don't have a contract to review for six weeks. The skill hasn't changed. It's the same file it was in January, sitting right there. But because it's a subscription, the fourteen dollars keeps leaving your account for a tool you're not using, and the moment you cancel to save money, the thing stops working entirely. Not degraded. Off. A client emails you a lease in August and now you're either resubscribing under mild duress or doing it the slow way again.

Nothing about the software got worse. Your relationship to it did. That's the trick at the center of most subscriptions: the value was delivered once, but the billing pretends it's delivered continuously. We didn't want to run a business that depends on you forgetting to cancel, or on the small panic of re-upping right when you need the tool most. That's revenue extracted from friction, and friction is a miserable thing to optimize.

With one-time purchase, your slow summer costs you nothing. The skill you bought in January is still there in August, still working, asking for nothing. You bought it. It's yours. That is how owning a tool is supposed to feel.

We didn't want to sell hostages

Generalize the scenario above and you get our actual objection. Subscriptions create a quiet hostility baked into the product: the thing you depend on is also the thing held just out of reach, renewed only by your continued payment. The longer you rely on it, the more it can cost you to walk away, and the seller knows it.

We didn't want a catalog of hostages. We didn't want sellers whose best growth lever is making cancellation scary. A tool you've paid for should not get worse the instant you stop paying, because nothing about the tool changed. The only thing that changed is that someone flipped a switch on their end to remind you who's holding the keys. We'd rather not hold the keys at all.

Sellers get paid for the work, not for the lock-in

The pricing model you choose decides what your sellers optimize for, whether you like it or not. Subscriptions reward retention mechanics: dripping out value slowly so there's always a reason to stay one more month, building switching costs, making the off-ramp narrow. Some of that produces genuinely better software. A lot of it produces dark patterns and dashboards nobody asked for.

One-time purchase points the incentive somewhere healthier. A seller earns by building something good enough that you'll pay for it today, on its own merits, with no promise of future leverage. They take their cut and move on to the next thing. We take a commission on each sale, somewhere in the range of twenty to thirty percent, and that's the entire shape of the relationship. No clawbacks, no retention quotas, no "engagement" metrics dressed up as value. Make a thing worth buying, sell it, get paid.

The trade-offs we signed up for

We're not going to pretend this is a free lunch. The advisors had real points, and here are the bills we agreed to pay:

• Revenue is lumpier. There's no smooth recurring base to forecast against. A big month doesn't guarantee the next one, and our finance spreadsheet looks more like a heartbeat monitor than a staircase.
• Sellers have to keep shipping. You can't coast on a subscriber list that auto-renews while you're at the beach. Every dollar of new revenue comes from a new sale, which means sellers are pushed to keep making new and better things. That's good for buyers and genuinely harder for sellers.
• Discovery is the whole game, and it's hard. When growth comes from new purchases rather than retained subscriptions, we live and die on whether people can find the right skill at the right moment. We have to be relentless about search, recommendations, and trust signals, because we don't get a second monthly bite at the same customer for the same product.

We decided lumpier-but-honest beat smoother-but-extractive. That's a values call as much as a business one, and we'd rather be upfront that it's both.

None of these are hypothetical. We feel each of them every month. We took them anyway because the alternative was building a business whose health depended on customers not paying close attention, and that's a foundation we didn't trust to hold.

What you actually get

Strip away our reasoning and here's the deal from your side of the table. You buy a skill once. You download it. You own it forever. It keeps working whether or not Skillmint is having a good quarter, whether or not we get acquired, whether or not we're even around in three years, because the file lives on your machine and runs on your Claude.

No surprise price hikes. No "your plan has changed" emails written by someone whose job is to soften a number going up. No losing a tool you depend on because a card expired and you didn't catch the failed-payment notice in your spam folder. You paid for it. It's done. The transaction is complete in the old-fashioned sense where complete actually means complete.

We might be leaving money on the table. The smooth-line people are probably right that we'd make more with a subscription, at least for a while. We think we're buying something worth more than that delta: a marketplace people don't have to keep one wary eye on. A place where "I bought this" and "I own this" mean the same thing. We'll take the spiky line.