Blog / Build with AI /
Turning a client problem into a product
Client work and product work look like opposites. One is paid for a specific person. The other is built for strangers. But the best small products often start as someone else’s request.
My argument: a client problem is the cheapest market research you’ll ever get, and the discipline is in telling a one-off from a pattern. Most requests are one-offs. A few are the same problem wearing different clothes. Those are the ones worth turning into a product.
For context, my first Framer plugin work was for a client in early 2025, and my own plugin collection came in January 2026. What follows is the general shape of the idea, without naming anyone, since client details stay private.
Why client problems make good product ideas
- The problem is real. Someone was willing to pay to have it solved.
- You know the context. You’ve seen how the person works, what they tried first and where they got stuck.
- You have a first user. Even one happy client is proof the solution can work.
- You’ve already built something. A custom solution is a head start on the product version.
Compare that to inventing an idea at a desk. You’d spend months wondering if anyone wants it.
The test: is it a pattern?
Before you build anything, ask five questions.
- Have I heard this problem more than once? Three different clients asking for similar things is a signal. One is an anecdote.
- Does it come up in different kinds of businesses? If only one niche has the problem, the market may be small, though a small market isn’t always bad.
- Do people currently solve it badly? Spreadsheets, copy-paste, five different tools. Workarounds are a good sign.
- Can the solution be self-serve? If it needs a conversation each time, it’s a service, not a product. My own direction is toward self-serve products, such as plugins, templates and tools, over anything that needs a lot of support.
- Would people pay, and for what? Not “would they like it”, but “what are they paying with now?” Time, money, or tolerance of a bad experience?
If you can answer yes to at least three, it’s worth a small experiment.
Separate the client’s version from the general version
A client’s solution is wrapped in their branding, their data and their quirks. A product has to survive without that. Do this exercise:
- Write what the client asked for.
- Underline the parts that are about their business specifically.
- Rewrite what’s left as “a person who [does X] needs to [achieve Y] without [pain Z]”.
That sentence is your product statement. If it’s still sensible with the client’s name removed, you’re on to something.
Respect the client
Two rules that I think are non-negotiable:
- Check who owns what. If you were paid to build it, read your agreement. Don’t reuse their code, content, data or brand without clear permission. Often you can reuse your own methods and ideas while keeping their specific work separate.
- Never use client data. Screenshots, examples and testimonials need explicit consent. Build fresh demo content.
This isn’t only ethics. A client who feels used will not recommend you again, and referrals are usually the best source of work.
Build the smallest version
Productising doesn’t mean rebuilding everything. It means:
- Take the core function and remove the client-specific parts.
- Add the boring product parts: settings, empty states, error messages and instructions.
- Write the listing or landing page as if the reader knew nothing.
- Release it somewhere that lets you learn quickly.
For Framer, a plugin marketplace is one such place. For WordPress, a free plugin directory. For others, a template store or a simple page with a payment link. I compared the options in selling digital products: platforms compared.
The idea of narrow beats broad appears again and again. Building one feature well vs ten features badly is about exactly this temptation: once you’ve got the first version, you’ll want to add everything every client ever asked.
The strongest counter-argument
“Don’t productise client work: it distracts you from clients and most products fail.”
Both are true. Most products don’t sell much, and time spent on a product is time not billed. My answer is to make the experiment small and cheap. A weekend or two to turn existing work into a simple version. If it gets traction, invest. If it doesn’t, you’ve lost little and learned something about the market.
There’s also a risk on the other side: staying in client work forever, because it’s safe, means you never own anything. A product gives you an asset that keeps working when you’re not.
Mistakes to avoid
- Building for one client and calling it a product. If only they use it, it’s custom work.
- Over-generalising, trying to cover every possible request until the product is vague.
- Pricing without thinking. Free to test is fine; free forever with no plan is not. Pricing deserves its own post, and it’s coming.
- Ignoring support load. A product that generates constant questions takes over your time. Support emails for a small product: what people ask sets expectations.
A quick worked example
Say a client keeps asking you to check that every page on their site has a title and description. You do it by hand each month. Three other clients ask for the same thing in their own words. The shared problem is “a site owner wants to see missing page basics without opening every page”. That sentence survives with the client’s name removed, it’s repeated, it’s currently solved with copy-paste, and it can be self-serve. That’s a candidate. A request like “make the homepage banner match our new brand” fails the test, because it’s about one business.
Key takeaways
- Client requests are free market research.
- Look for patterns: the same problem, heard more than once.
- Strip out everything specific to the client, and keep the general need.
- Respect ownership and confidentiality.
- Build the smallest self-serve version, then learn from real users.
Related reading
Think of your last three client requests. Is there a thread running through them?