A useful question to ask of any software firm is: what would happen to their clients, in practical terms, if the software quietly began to lie? The answer tells you who the firm actually builds for. At Semperr, the answer is uncomfortable, and we like it that way.
A misreported revenue figure in a Trace dashboard, on the wrong Friday evening, can cost a client a deal. A missed citation in a Clad brief can cost a client a ruling. An unreliable point-of-sale in the back of a Dev-built bakery can cost the owner the only afternoon in the week when she gets to sit down. These are our clients. They did not buy our software to feel modern. They bought it because the software they had was costing them.
Three kinds of serious
Our clients fall into three broad categories. They look, on paper, like very different businesses. In practice, they have more in common with one another than with most software customers.
The analyst. She prices a deal on a Friday evening. The figures she ships upstream to the partner are going to be cited, by memory, in a meeting the following Tuesday. She does not have time to check our numbers against a second source; if she has to, we have failed. Trace exists for her.
The partner. He files a brief on Monday. The judge will read it. The opposing counsel will read it. His own name will be on it. A broken citation — or a case summary that is subtly wrong — is not an embarrassment; it is a professional event. Clad exists for him.
The proprietor. She opens the shop at six. She has six hours before the first lunch rush to do everything that does not involve a customer. The one hour a week she spends on "the system" is one hour not spent on bread. If the system costs her more than an hour, she is paying us to subtract from her week. Dev exists for her.
The phone, again
Every account at Semperr is assigned to a real human being, with a name, an email, and the authority to fix whatever is broken. We answer the phone. We do not hide behind ticket queues. We do not ask the client to produce a video of the bug.
This is not generosity. It is, in the strictest sense, the product. A software firm that cannot be reached when its software breaks is selling a gamble, not a tool. We are not in the business of selling gambles to people whose week depends on the outcome.
Who we do not serve
There is a kind of prospect we politely decline. It is the prospect who wants a clever dashboard, a good-looking slide for a board meeting, or a badge of technical sophistication. We are delighted that these products exist. We do not make them.
We also decline work where the cost of being wrong is low. If the figures do not matter much — if the dashboard is a vanity, the brief is a draft, the shop is a side project — there are cheaper, faster, looser tools. We would rather our clients use those tools and come back when the stakes change.
A shorter list than you'd think
The practical consequence of all of this is that our client list is shorter than it could be. We have walked away from work we could have done. We have referred prospects to other firms. We have, on more than one occasion, lost a competitive process because the winning firm promised more and charged less.
We are comfortable with this. A short list of serious clients, served well for a long time, is the only kind of business we know how to run. We intend to keep running it.
— Henry