Don’t wait months to see it work
Every Wednesday you see something working, and you decide what we build next. You can look at the work any day of the week, so you never wait months to find out what you paid for.
01 · Where it started
Online, but not for sale
When Sara Paula Hoffman, a painter and sculptor, started working with us in December 2024, you could admire her paintings online, but you couldn’t buy one.
The usual next step is a plan: meetings, a proposal, and a website delivered some months later. We did something smaller. The first week answered one question: how should one painting be shown so a buyer can tell what it is, how big it is, and whether it’s for sale? At the end of that week she had listing designs to look at, not a document about them.
Every week after that was one more question, paid for as one piece of work. The second asked which places online her work belonged on. The third gathered her work into series, on a simple website built that same week. The fourth asked how her site should promote a gallery show and send buyers to the gallery. Then, in February, we spent a week counting, measuring and pricing every work she had. Nothing custom was built until the cheap version had shown what it couldn’t do.
Each week was a test drive, to learn one thing. Each result was a demo, in her hands, to decide what came next.
Any week could have been the last. That’s the point of working a week at a time: you never pay for more than you’ve seen working. The whole case lists every week as it was invoiced.
02 · The test
When her first show opened
The morning after her first gallery show opened, we saw the gallery listing three of her works at 40 to 53% under her own price.
Both prices had to be true at once. While the show ran, those three works were sold the gallery’s way, and anyone who asked about one was sent to the gallery. The rest of her work had to stay at her own price, and so did those three, the day the show ended.
Most websites would have written over each painting’s own price, and someone would have had to remember to change it back. We built something else that same morning: on her site, a gallery’s price sits on top of hers like a sale sticker on a price tag, with a start date and an end date. Peel the sticker off and the tag underneath is just as it was.
One sold. When the show closed, the other two came home, and putting them back took one change. No other work on her site moved by a dollar. Noticing the show had ended took longer: her site showed the gallery’s prices for twenty days before we changed them. So now the site checks the gallery’s listings and the show’s dates every morning, and sends her a note when something needs deciding. It never changes a price on its own.
Something happened that nobody had planned for, and the sticker came off clean.
03 · How we work
A test drive, then a demo, both with dates
The test drive
Built to find something out
A test drive is a week of work that answers one question about your business. If you know Scrum or Google’s design sprints, it works like a sprint: a fixed week, one goal, and something to show at the end. It works the way an exterminator treats one room and books the follow-up visit before the truck leaves: on that day, the ants are gone or they aren’t. Before we start, we write down what we expect to find, and what result would tell us we were wrong. Then we build the smallest thing that could give either answer. A test drive that could only ever say yes would tell you nothing, and cost you the same.
The demo
Built so you can decide
A demo is that follow-up visit. It’s the thing itself, working, and you open it and use it yourself: any day you want a look, since the work is open the whole week, and by the end of every week without fail. Beside it we tell you plainly what we learned, including what went against us. Then you choose: keep going, change direction, or stop. Stopping is a perfectly good answer, and at this size it’s a cheap one. That’s why we keep things this size.
04 · Why we work this way
A year of meetings, and nothing to use
Nearly every small business we sit down with tells us some version of this.
They hired someone, and plenty happened. There were meetings, research, a plan for the year and some genuinely good-looking slides, all made by capable people working hard. But at the end, the thing they’d paid for didn’t exist, and the money was gone.
Nobody lied. The trouble is that nothing in that kind of arrangement names a date on which something has to work. It’s a treatment with no follow-up visit booked: nobody is ever obliged to go back and look, and the work can carry on, in good faith, for a very long time.
A plan can go on forever. A test drive ends on a date.
05 · What we built for Paula
Four things we hold to, on her site
These four come from a short statement software people wrote in 2001, the Agile Manifesto. Here’s what each one meant for one artist’s website.
01 She keeps working the way she already works
Paula doesn’t fill in forms. When she agrees a show with a gallery, she writes the email she would have written anyway. Someone reads her inbox at the start of every working session and turns what she wrote into the change on her site. One day a program may read those emails for her. If it does, it will get the change ready and wait, and a person will still say yes.
02 The site describes itself
What search engines read about her work, and what AI assistants quote, is produced from the same list of works the site itself shows. Nobody has to keep those descriptions up to date by hand, so they can’t quietly fall out of date. Even the case study was written after the site went live, from the site.
03 A person makes the call
Every day the site compares what the gallery lists with what her own site shows. When the two disagree, it could pick one on its own. It doesn’t, because why two prices differ is a matter between Paula and her gallery, and no program knows which one is right. It points out the difference and leaves the decision to a person.
04 Ready for what nobody planned
Not every new piece belongs to a series, so the home page changed to lead with her recent work, and a piece no longer needs a series to go online. And when her first show closed, the two works that came home went back to her own price in one change, because from the second day of the show its prices had been kept apart from hers.
What it is not
A treadmill with a nicer name
Weekly test drives are meant to lower the risk, not to squeeze people. There are no scores, no meetings anyone dreads, and nothing carries on just because it’s on the calendar. If there’s nothing worth finding out this week, we’ll say so. The whole arrangement is there so that being wrong is quick and survivable, and so that you find out in a week rather than in a year.
Experiment-led · test drive to learn · demo to decide
Book a free first call.
Bring the thing that isn’t working. We’ll tell you what we’d build first, what it would teach us, and what would make us tell you to stop. If none of that is worth a week of your money, we’ll say that too.