Custom software building is evolving into a product company

Seeing an increasing number of companies that build ‘custom’ software for large enterprises. Lots are growing incredibly well, have recurring revenue and are getting good valuations both in the VC and in the PE world.

Why does this matter? The line between project consulting and product software business is becoming a blurred one. Custom software creation is usually not top business model — it’s a labor+margin per hour business with project-driven, one-time revenue that body shops like Accenture would do. Pure software products are built once and resold as-is to all customers for a recurring fee, justified by cloud hosting and continuous improvement.

In reality, most vertical software companies do customization and data structuring work for clients on onboarding (maybe 5-10% is custom?). But this is not necessarily bad — more customization can mean more customer retention. Software is usually focused around a function or industry so that the deliverable is mostly reusable blocks but also custom enough for the client’s need. Customized code creation is (or was) expensive, and would raise CAC and maintenance cost too much for how much a single client paid — so you have to build for the average client’s workflows and have the client adapt.

Palantir evolved this — they concluded that for VERY large clients, customization made sense. ~30-70% of their deliverable is custom every time. They have reusable pre-built blocks (which were part of previously custom builds), and the rest is done custom for the client. This helped them charge a subscription. And because the software is highly custom, it has very good retention.

Now LLMs have significantly reduced the cost of code. What would have taken a year in 2024 now takes 1 month (quote from a developer doing this model). So Palantir-style plays can travel downmarket. We are seeing startups building custom software (also called AI) for all industries - they start fully custom and over time amass a set of reusable blocks. At the core, they are building software that builds custom software. A lot are being able to charge recurring revenue for it. And clients are highly demanding it.

My view, keep on doing this! Being everything for everyone opens up the client funnel wider. On the other side, focusing allows a company to be ‘known’ and better able to consult the client on what they need. That is why, at McKinsey, senior people focus on a single industry or function, even if the advice is custom every time.

It’s ok to start broad to find what users want. But even if the cost of software drops to zero, I would not underestimate the benefits of focusing around a client-type.