Founders ask what design costs. The more useful question is what it should be spent on this quarter, because the same budget buys a very different amount of progress depending on where the product actually is.
A team still looking for product-market fit and a team opening a second market need opposite things from a designer. Spend growth-stage money at the pre-fit stage and you get a beautiful interface for a product nobody has validated. Spend pre-fit money at the scaling stage and you get research that confirms what the team already knew.
This is how I split the work at each stage. The percentages are how I allocate, not a law of the category - but the ordering behind them holds across every product I have shipped.
Before product-market fit: buy certainty, not pixels
The job at this stage is to be wrong cheaply. Everything spent here is buying information about whether the thing should exist in the shape it is being built.
About 45% goes to research. Interviews, customer development, prototypes put in front of real situations. One good conversation costs less than an hour of interface work. It also sets the direction for months.
About 35% goes to structure: flows, states, the skeleton of the product. At this stage the logic is the product. The visual layer is a courtesy.
The last 20% buys enough interface to be used. That is the point. Someone has to be able to try it and tell you the truth about it.
The saving does not show up in the design budget. It shows up in the engineering budget that follows. A wrong assumption caught in a prototype costs a conversation. The same assumption caught after the build costs a refactor. That is the whole argument for spending here first.
What replaces the research at this stage is usually a persona document, which is the expensive version of guessing - I wrote about why in nobody hires a persona.
- 45% research: interviews, customer development, prototype testing
- 35% structure: flows, states, the working skeleton
- 20% interface: enough to be used and judged
Early growth: the money moves to the entry point
Once the product has a market and paid traffic is running, the constraint changes. The question is no longer whether people want this, it is what each new customer costs, and design now works directly on that number.
About 40% goes to conversion work. The pages traffic lands on, onboarding, the moments where someone activates or leaves. This is design's shortest path to the business. A better entry point buys more customers from a budget you are already spending.
About 30% goes to the design system. From here on, the cost of every new feature depends on how much of it has to be invented from scratch.
The rest splits two ways. Around 20% for instrumentation and iteration. Around 10% stays with research, so the team does not lose contact with the market it just found.
The system is the line founders defer longest and regret most. It is not a tidiness project. It decides how fast the next twenty features ship. I have made the case for building it around decisions rather than components in a design system is a decision system.
And the conversion half only works if the entry point is treated as part of the purchase - the case for that is in a weak landing page is a tax on your ad spend.
- 40% conversion: landing pages, onboarding, activation
- 30% design system: the cost floor for everything after this
- 20% analytics and iteration
- 10% continuing research
At this stage design is not competing with marketing for budget. It is deciding what the marketing budget is worth.
Scaling: retention, recognition, and the cost of a large team
The funnel works, the product is stable, and the problems become long-term ones: keeping people, being remembered, and shipping fast with more hands on the product.
About 35% goes to product experience. New scenarios, and the reasons someone stays past the first month. A customer who gets value every week builds their own work around the product. That is what lifetime value actually is.
About 30% goes to brand. At this stage it stops being decoration and starts being a conversion asset. Recognition lowers the cost of every interaction that follows it.
The remainder is infrastructure and reconnaissance. Around 20% scales the design system so a bigger team keeps its release speed. Around 15% researches the next segment or country before entering it, rather than after.
Founders find this line the hardest to accept. Brand was correctly deprioritised for years, so promoting it feels like a mistake. It is not.
Both identities I have built were designed for exactly this moment. by.studio and boostlab both had to hold on surfaces nobody art directs. An invoice. A status page. A repository avatar. Places where the mark is the only thing carrying recognition.
- 35% product experience and retention
- 30% brand as a conversion asset
- 20% scaling the design system
- 15% researching the next segment
The three rules underneath the numbers
The percentages will shift with the company. What holds is the reasoning that produced them.
Research protects capital. The cheapest place to be wrong is a conversation. The most expensive is production code.
A design system is infrastructure, not housekeeping. It sets the price of every feature that comes after it.
Numbers create objectivity. Design earns its budget when it aims at a metric someone can check, not at a preference someone can defend.
The failure I see most often is not underspending. It is spending the right amount at the wrong stage - polish before validation, or research after the answer is already in the analytics.
How I work out which stage a product is in
The first conversation is not about the interface. It is about which of these three situations the company is actually in, because founders routinely believe they are one stage further along than the evidence supports.
The tell is usually in the funnel, and it is easy to read once you look.
People arrive and do not understand what the product is? That is a pre-fit problem in a growth-stage costume. They understand it, sign up, and never come back? The money belongs in activation, not acquisition. Everything works and nobody remembers the name? That is a brand problem, and no amount of interface work will touch it.
Published prices for each kind of engagement are on the main page, which is unusual in this category and deliberate: a founder deciding where the next few thousand dollars goes should not have to book a call to find out the order of magnitude.
Design pays when it is pointed at the constraint the business currently has, and the constraint moves. Validation, then acquisition cost, then retention and recognition. Getting the order right matters more than getting the amount right, because the wrong stage of work does not simply underperform - it produces confident answers to questions nobody was asking yet.