This year I cancelled Harvest, ClickUp and Framer.
The Pony team now runs on PonyHQ, an app I built myself. Iām a designer. The first version took about two days.
It saves us $11,108 a year.
The money is the least interesting part.
A price tells you what the seller thinks of you
Iād used Harvest for ten years, since my freelance days. Two months ago our price went up 13x.
Nothing about the product had changed. So what was I being charged for?
We were charged a percentage of the amount we invoice. Huh? Not the number of invoices, but the amount it says on the invoice. That was their new pricing model. How clever.
A percentage of what I invoice is a tax on my studioās revenue. Harvest did none of the work behind those invoices.
So why would a company think it could ask for that?
My switching cost.
Harvest was bought by Bending Spoons last year. The same owner has raised prices at Evernote and WeTransfer too. The bet is always the same: leaving will hurt more than paying.
That bet was safe for twenty years. Moving off a tool meant months of pain, so a vendor could charge you for the pain.
My switching cost turned out to be two days.
When your vendor gets acquired, read the announcement as an early price notice.
Software with one customer
PonyHQ tracks our time, runs our projects, creates invoices, pushes them to Xero and shows the profit on every project.
That list could describe a hundred products. The difference is who it was built for.
Harvest has to work for a law firm in Ohio and a dev shop in Berlin. PonyHQ only has to work for one branding studio in London.
Every feature in Harvest that I never touched was there because someone else needed it. I was paying for their needs. Every feature in PonyHQ is there because we use it.
Each person on the team sees what they need to see. We have a level of transparency we never had before, because I decided what transparency means for us. No product manager in another country decided it for me.
Why SaaS never did this
SaaS was built to scale. One product, sold to as many customers as possible.
A product built for ten thousand companies is designed for the average of them. No company is the average.
So for twenty years we bent our companies around other peopleās software. We changed how we worked to fit the tool, then called it best practice.
Building for each customer individually was the thing that didnāt scale. Do that and you were told you had a service business.
It was a fair trade while code was expensive. Code is cheap now.
Iām far from alone. Retool and McKinsey both found about a third of companies have already built something in place of buying it.
Jason Lemkin at SaaStr calls these N=1 apps: software with exactly one customer. His warning to vendors is the part I keep thinking about. The customer who builds their own never complains and never moves to a competitor. They cancel, and the vendor never finds out why.
Whatās scarce now
When building gets cheap, knowing what to build becomes the valuable part.
Ten years as a Harvest user was my spec. I knew which parts we relied on and which weād never opened. No vendor will ever know my studio that well.
The person closest to the problem can now build the solution. That used to need a product team, a roadmap and a year.
So the advantage moves to people who understand their own business in detail. Knowing exactly how your company works used to be an operations skill. Now itās how you get your software.
What Iād tell a founder
Rent what makes you the same as everyone else. Own what makes you different.
Accounting, payments, security: rent. We kept Xero. I have no interest in maintaining accounting software.
How your team works, what you measure, who sees what: own.
The easiest wins are tools where you pay per seat for a workflow that is really yours. Time tracking, project management, internal dashboards.
Start with the tool you know best. You already have the spec.
And be honest about the cost. What you build, you maintain. I donāt know yet what PonyHQ will cost me over a year. Ask me in twelve months.
Where I think this goes
Three guesses. I could be wrong on all of them.
Lock-in stops working as a pricing strategy. A 13x increase only works if leaving is painful. Once customers know they can rebuild, the price of any simple tool has a ceiling: what it costs to replace it.
How a company works becomes part of its brand again. When every studio runs on the same five tools, every studio ends up with the same process, the same reports and the same blind spots. Shared software makes companies comparable. I spend my working life helping founders become impossible to compare, and Iād missed that our own tools were pulling us the other way.
The old insult flips. āThatās a service businessā used to mean you couldnāt scale. If AI does the work of fitting software to each customer, building for one customer at a time scales fine. Some of the best software companies of the next decade will look a lot like service businesses.
Finally, a service
For twenty years, software as a service meant renting the same product as everyone else.
A service is made around the person receiving it. A tailor gives you a service. A clothes shop gives you a product.
I think the name is about to become true. Software made around you, changed when you change, priced for what it does for you.
A 13x price rise pushed me into all of this. Iām almost grateful.
Love,
Stef

