Product Design
Growth
Research
2025 ∙ Bark.com
United Kingdom

Bark is a UK-based marketplace connecting customers with local service professionals. When customers submit a request, relevant professionals receive the lead and can respond with quotes. Bark runs on a credit system: sellers purchase credits and spend them on leads, with the first purchase being a starter pack covering around 8 leads. Bark also has a mobile app, and the data was clear. Sellers who used it were more successful, largely because push notifications let them respond to leads faster than those relying on email.

App adoption was stuck at 7% between signup and second purchase, and 1st-to-2nd purchase conversion sat at 50%. Previous attempts to promote the app had all failed to move the needle: modals, emails, Chameleon banners. The problem wasn’t the message. It was the timing. Sellers were being asked to download the app at random moments in their journey, before they had any reason to care.

I led the end-to-end design for the solution, working closely with product and development teams. I had roughly two weeks to design and launch, running in parallel with other projects and aligned to sprint timelines.

Understanding the problem

The goal of every seller on Bark is simple: find leads, convert them to paying customers, make money. Speed of response is one of the biggest factors in winning a lead. The faster you reply, the better your chances. The app solves that directly. But before I could work out when to tell sellers that, I needed to understand where they actually were in their journey when it would mean something to them.

I mapped the full seller journey from signup to second purchase to find the right intervention point. What became clear quickly was how congested the early stages already were. From signup through first purchase, sellers were already being hit with onboarding tours, educational modals, and conversion prompts. There was no room for another message, and more importantly, sellers at this stage hadn’t yet experienced the platform enough to understand why speed mattered.

Immediately after first purchase, sellers were in a different kind of pressure: they’d just paid for leads and needed to respond fast. Any distraction at that moment risked them losing a lead to a faster competitor. Not the right moment either.

The sweet spot was after the 5th lead response. By that point, sellers had experienced wins and losses, understood that speed mattered, and had built the appetite to do something about it. There was also a practical dimension: the starter pack covers 8 to 10 leads depending on cost, so at lead 5, sellers still had enough credits left to actually try a new way of working. Downloading the app wasn’t just a good idea in theory — they had the runway to test it immediately. Previous attempts had failed because they interrupted sellers at the wrong moment. The 5th response was the first point where sellers had both the experience to understand their problem and enough credits remaining from their starter pack to try a new approach before needing to top up.

That was the breakthrough. Previous attempts had failed not because the message was wrong but because it was showing up at random points before sellers had any reason to care. This wasn’t going to be another random banner. It was going to be a strategic intervention at the moment of peak receptivity.

Design

With the timing clear, the message needed to match it. A seller who had just responded to their 5th lead already knew the problem. They didn’t need to be told that speed mattered — they’d felt it. What they needed was proof that the app would actually help, in a form specific enough to be credible. The copy I landed on was: “Respond to leads instantly on your Bark app to increase your chances of being hired by nearly 25%.” The 25% figure was a real product stat, not a marketing claim, and it gave sellers something concrete to weigh against the friction of downloading a new app.

Rather than a single touchpoint, I designed a progressive reminder system with decreasing intrusiveness. A modal appeared after the 5th lead response, creating a focused moment to make the case. If a seller dismissed it and came back in a later session without having downloaded the app, a less intrusive banner appeared in the lead list as a gentle nudge.

Navigation links to the app were also added as a permanent low-friction option throughout the product. Maximum focus at the peak motivation moment, then stepped-down reminders that respected the seller’s attention without vanishing entirely.

Before launch, I validated the modal comprehension with 10 sellers through an unmoderated test. But I knew the real validation had to be the A/B experiment itself, as no test could replicate the genuine frustration of losing leads over five real responses. Comprehension I could check; motivation I could only measure in the wild.

Results

We ran an A/B experiment until reaching 90% statistical significance, which took around five weeks. App adoption increase validated the core hypothesis: show the right message at the right moment, and sellers will act.

 The real question was whether getting sellers onto the app actually made them more competitive. It did. Median response time more than halved, dropping from 112 minutes to 45 minutes. Sellers responded to more leads and did so more consistently, likely because having the app removed the friction of sitting at a desktop waiting for email notifications.

The downstream effect was exactly what we’d hoped for. Sellers who responded faster got hired more often, and sellers who got hired came back for more credits. First-to-second pack conversion increased by 9%.

Reflection

The core insight here wasn’t a design decision. It was a sequencing decision. The app hadn’t changed, the message hadn’t changed, and the audience hadn’t changed. What changed was when we showed up. Timing the intervention to a moment of genuine motivation, backed by a stat sellers could connect to their own experience, was enough to move a metric that repeated attempts had failed to shift.

Two things I’d explore with more time. Testing different messaging variations beyond the 25% stat would have revealed whether different value propositions resonated with different seller types. And tracking adoption trends over the following months would have shown whether the lift sustained or whether follow-up interventions were needed as new seller cohorts came through.

London, UK