When Beautiful Design Kills Conversion
Kate Syuma joined Miro in 2017—back when it was still called Realtime Board—and there was no onboarding to speak of. Just a signup form and a blank canvas. As the first product designer on the growth team, she built one of the most visually striking onboarding flows in SaaS: users selected their use case from a menu, and animated GIFs of Kanban boards and roadmaps danced on the right side of the screen. Twitter loved it. Designers screenshot it. The metrics collapsed.
"After launching that, we saw some drops and then we analyzed why it's happening with user interviews and uncovered that users are distracted by these visuals," Syuma recalls. "They start interacting with them, especially if they look like the product itself." The team had confused delight with utility. The animations weren't clarifying value—they were competing for attention with the actual signup questions. Users thought the GIFs were clickable prototypes and abandoned the flow trying to interact with fake UI.
It doesn't mean that we should just remove these visuals completely and make it clean and white. It's more important to iterate and to test multiple versions. These visuals turn back actually, but in a different way that presents the value but doesn't make this confusion that it looks like a real product.
— Kate Syuma
The insight stuck. Clean, centered signup flows outperformed clever ones. But Syuma didn't swing to brutalism. She learned to treat visuals like seasoning: necessary, but only if they clarify the core value without pretending to be the product. The current Miro onboarding brings visuals back—just static, illustrative, non-interactive. Enough to prime intent, not enough to derail it.
The COVID Inflection: When Your ICP Rewrites Itself Overnight
Then 2020 hit. Miro's user base exploded from early-adopter design teams to millions of non-tech-savvy remote workers who had never used a digital whiteboard. The entire activation funnel had to be rewritten for people who didn't understand what "visual collaboration" meant. Syuma's team faced a choice: scale self-service or inject human touch.
They tried both. One experiment introduced a "collaborative onboarding" flow where a virtual guide—part chatbot, part human representative—walked new users through their first board. The team prototyped it using Synthesia, an AI video tool, years before ChatGPT made synthetic media ubiquitous. User feedback was glowing. The scalability was not. "Even today, looking at the quality of the videos, sometimes it's not the level of quality that you really want to introduce to your end customers, especially for such a user-centric and quality-oriented product like Mira," Syuma explains. The experiment validated demand for human touch but died on the altar of production costs.
What survived was ruthless personalization. Syuma's team doubled down on profiling questions at signup—role, use case, team size—even though conventional wisdom warned it would crater conversion. It didn't. Referencing insights from growth advisor Elena Verna, Syuma notes that adding profiling questions can drop signup conversion by roughly 10%, but activation rates often rise because users become more intentional. "These people will not have a negative effect on the activation funnel. And sometimes by introducing more questions, the activation can actually increase because you are personalizing the experience or the users are becoming more intentional while they are filling in this information."
The profiling information is not static. You shouldn't be afraid of adding extra questions to your signup flow. It will not drop extremely your conversion to the signup. And even if it drops, there is no negative effect on the activation funnel.
— Kate Syuma
At 50 Million Users, Big Bets Become Liabilities
By the time Miro crossed 50 million users, Syuma had learned the hardest lesson in growth: at scale, heroic redesigns are a trap. The cost of shipping a full onboarding overhaul was prohibitive. The learning curve was glacial. The risk of regression was existential. So the team stopped swinging for the fences and started hunting for leverage in the margins.
"You start thinking less radically in terms of big bets and just revamping everything, but you start being very intentional about what actually can move the needle," Syuma says. "You start searching for that very, very tactical sometimes thing that can actually move this needle." One of those tactical things: a "say hi" prompt for users joining a shared board. When someone landed on a collaborative board for the first time, Miro nudged them to drop a reaction—a wave, a thumbs-up. It introduced the reactions feature, created a micro-moment of social presence, and moved the aha moment metric for joiners without requiring a single line of new infrastructure.
Don't underestimate quick wins, don't underestimate these small improvements. They can actually have an impact. And sometimes big bets are overestimated and they don't have such an impact.
— Kate Syuma
Syuma began decomposing the activation metric by segment—joiners versus creators, use case by use case—and running tight, low-cost experiments against each cohort. The shift from "let's redesign onboarding" to "let's test one micro-interaction per week" was cultural as much as tactical. It required the team to accept that at scale, compounding small wins beats the lottery of the big redesign.
The Onboarding Report and the Quarterly Review Doctrine
After leaving Miro, Syuma surveyed more than 80 companies to understand how they approach onboarding. The result was a public research report published at onboard.report. The top finding: most teams don't revisit their onboarding flows regularly. They ship it once, move on, and let it rot. Syuma's prescription is simple and surgical: review end-to-end onboarding at least once a quarter with your product manager, designer, and growth lead. Surface bugs, UX friction, and backlog experiments. Treat it like a living system, not a launched feature.
She also found that activated users convert to paid plans at 3 to 10 times the rate of unactivated ones—a multiplier that justifies relentless iteration. But iteration requires a North Star. For Syuma, that's the activation metric itself, which she says can take six months to define properly. "It's a very difficult metric to define. It sometimes takes half of the year just to understand that it's more or less valid and it makes sense to have such a metric," she admits. "But you will refine it, you will change it over time. Starting thinking in that direction is already a good way."
If you just continue to iterate on the onboarding flow without having the activation definition in place, it will be difficult to understand what is working, what is not working, what is the core value of the product and what we are optimizing for.
— Kate Syuma
Syuma's framework is ruthlessly pragmatic: define activation early, revisit it annually, and never optimize for delight at the expense of clarity. The profiling questions you ask should change as your ICP evolves. The features you introduce in session one should be the smallest set that unlocks value, not a tour of your roadmap. And the experiments you run should be cheap, frequent, and segmented—because at scale, the safe bet is the one you can afford to get wrong.
The Anti-Overwhelm Doctrine
One pattern emerged across Syuma's research and tenure at Miro: users are overwhelmed not by complexity but by noise. Bombarding new users with dashboard tours, feature callouts, and lifecycle emails doesn't activate them—it fractures their attention. The antidote is editorial restraint. Show one or two features in the first session. Introduce reactions before you introduce comments. Let the second session expand the surface area. The onboarding flow isn't a product demo; it's a deliberate sequence of micro-commitments that compound into habit.
Syuma's advice to teams is to audit what they're asking users to do in those first moments and cut everything that doesn't directly ladder to the aha moment. That might mean fewer tooltips, fewer emails, fewer template options. It definitely means tracking which features correlate with activation and suppressing the rest. The goal isn't to show the product—it's to help users experience the core value before they churn. And that requires saying no to almost everything else.