A client came to me with 18 months of traction, 2,000 active users, and an MVP that looked like it was styled with inline CSS and prayers.
Which it was.
The founder defended it immediately: “It’s scrappy. It works. Our users don’t care about design.”
I didn’t argue. I pulled up the dropout metrics instead.
47% abandoned signup at screen two. Not because anything was broken — the form worked, the buttons worked, the backend worked. Because nobody understood what they were supposed to do next. The UI gave them no signal, no guidance, no indication of what screen two was even for.
His MVP UX design had proved the idea worked. Two thousand active users in 18 months is real traction. But that same MVP, unchanged, was now costing him half his potential users. Every week. For 18 months.
That’s the moment every founder eventually faces: when scrappy stops being efficient and starts being expensive.
Why Founders Defend MVPs Past Their Expiry Date
The defence is always some version of: “Our users love it, we’re growing, why fix what isn’t broken?”
It’s a reasonable position for month three. By month eighteen it’s rationalisation.
The thing is — a good MVP should feel uncomfortable. If it doesn’t, you’ve spent too long smoothing out the wrong things. You don’t need a design system or a rebrand or a Figma file with 200 components. You need a working path from problem to value, built fast enough to prove someone wants it.
What this client had was exactly that. The MVP had done its job. The problem was he’d stopped treating it as a starting point and started treating it as a finished product.
The data told a different story:
Signup completion rate sat at 53%. Industry average for B2B SaaS is 75 to 85%. He was losing nearly half his signups before they’d seen the product.
Second session rate was 38%. Users who made it through signup weren’t coming back. Good products run at 60% or higher.
12% of users had discovered the key features. The rest were using a fraction of what they’d paid for, not because the features didn’t exist but because the MVP UX design had never surfaced them.
Support was running at 200 tickets a month, almost all variations of “I don’t know what to do next.”
Churn in the first 30 days: 31%.
None of these numbers were catastrophic in isolation. Together they described a product that had found demand and then quietly failed to convert it for a year and a half.
What We Actually Fixed
We didn’t rebuild anything. That’s the first thing I told him, and the first thing he needed to hear — because the fear of a six-month redesign is usually what keeps founders on their MVPs for too long. You don’t need to burn it down. You need to find the five things that are costing users and fix them.
Onboarding guidance. The signup flow dropped users into an empty dashboard with no indication of what to do. We added progressive disclosure — one prompt, one action, one step at a time. Not a feature tour, not a tooltip parade. Just a single line of copy at each decision point telling users what this screen was for and what to do on it.
Empty states. Every zero state in the product showed a blank screen with a generic message. We rewrote each one to answer the question users were actually asking: what is this feature, why is it empty, and what should I do right now. One empty state fix on the main dashboard cut support tickets about that feature by 40% in the first two weeks.
Navigation. Eight top-level navigation items for a product where users needed three of them regularly. The other five created decision paralysis on every page load. We reduced to four items, moved the rest into a secondary menu. Users stopped clicking around looking for things and started clicking toward them.
Feature visibility. The product’s most valuable features were three clicks deep with no signposting. Users who found them retained at significantly higher rates than users who didn’t — which meant the retention problem wasn’t the features, it was the UX design that had buried them. We surfaced two key features with contextual prompts that appeared after users completed specific actions.
Visual hierarchy. The primary CTAs on most screens looked identical to secondary actions. Same size, same weight, roughly the same colour. Users were making the wrong choice not because they wanted to but because the interface gave them no signal about which choice mattered. We fixed the hierarchy — made primary actions look primary, made secondary actions recede.
Five things. Six weeks of work. Same product, same core functionality, same codebase.
Results: signup completion went from 53% to 79%. Second session rate went from 38% to 64%. Feature discovery went from 12% to 41%. Support tickets dropped from 200 a month to 73. First-month churn went from 31% to 18%.
Not a redesign. A targeted fix of the MVP UX debt that had been compounding quietly for 18 months.
The Cost of Waiting
Here’s the math the founder didn’t want to do but needed to see.
47% signup abandonment across 18 months, at roughly 150 new signups per month: approximately 1,270 users who never made it past screen two. Not because the product wasn’t right for them — because the UI gave them no reason to continue.
At his average revenue per user, that abandonment represented somewhere between $180,000 and $240,000 in lost lifetime value. Rough math, conservative assumptions. But in the right order of magnitude.
The six weeks of product design work cost $14,000.
The 18 months of waiting cost roughly 15 times that, in users who left and never came back. Users who abandoned at screen two don’t return to see the improved version. They found something else, or gave up on the problem, or told someone the product was confusing.
This is what expensive inertia looks like. Not a dramatic failure. Just a quiet, consistent bleed across every cohort for a year and a half, invisible because the founder was looking at active users instead of abandoned ones.
When MVP UX Design Needs to Evolve
The signals are usually visible before founders are ready to act on them.
Support answering the same question more than 20 times a month is the clearest one. If users are repeatedly asking “how do I…” about the same feature, the UX design isn’t clear enough anymore. Not a training problem, not a documentation problem — a design problem.
Second-session rate dropping below 50% means the product isn’t successfully onboarding new users into habit. Early adopters forgive confusion because they’re invested in the idea. Users who find you in month twelve aren’t early adopters — they’re evaluating you against alternatives that have had time to mature.
New features that don’t fit the existing UI are a structural signal. When every addition requires a one-off solution — a new tab, a modal on top of a modal, a settings page that doesn’t match the rest of the settings — the MVP architecture has become a constraint rather than a foundation.
Feature adoption below 20% on core features almost always means the features exist but aren’t findable. The product UX is hiding value the product has already built.
None of these require a rebuild to fix. They require someone to look at the product the way a new user does — without the founder’s context, without knowledge of where things are, without patience for a UI that doesn’t explain itself.
Scrappy Has a Shelf Life
Your MVP UX design should be uncomfortable. Honest about what it is. Fast to build, fast to change, built to prove a hypothesis rather than impress anyone.
That’s exactly right for month one. It stops being right when you’ve found traction and you’re losing half your potential users to confusion that six weeks of focused work could fix.
The client with inline CSS and prayers had proved something real. Two thousand active users in 18 months is not luck. The product worked. The MVP UX design just never grew up alongside it.
The second version isn’t about polish. It’s about stopping the bleed — the quiet, consistent loss of users who arrived at a working product, couldn’t figure out what to do next, and left before they found out what they were missing.
Fix the five things that matter. Don’t rebuild everything. Don’t wait another 18 months.
The users you’re losing to confusion right now won’t come back to see the improved version.