What is Middleware?
Problem Statement
New users were dropping out before completing onboarding. Almost 90% never integrated their tools or saw their own data, which meant they never reached activation or their “aha” moment. Our onboarding was scattered, confusing, and failed to guide them through the critical path to value.
Problem Discovery
The issue surfaced in conversations with the sales and marketing teams. They were generating demand, booking demos, even convincing teams to sign up. But inside the product, enthusiasm collapsed.
PostHog data and LogRocket sessions showed users wandering aimlessly after login. Instead of following the onboarding checklist, they got distracted by an empty product and quit before setup.
We weren’t failing to acquire. We were failing to activate.
Research
Behavioral Analytics: Session recordings highlighted abandonment after the first login screen.
Data: Only ~10% of users completed onboarding.
User Expectations: Interviews confirmed a simple truth: people expect onboarding to guide them, not leave them stranded.
Cognitive Principles:
Cognitive Load: too many choices, too soon.
Broken Mental Models: onboarding should be linear, not scattered.
Perceived Progress: no visible milestones, no sense of advancement.
One pattern stood out: users who connected their tools were far more likely to stick around. Integrations weren’t just a feature, they were the hook.
My Hypothesis
If we structured onboarding as a flow instead of a checklist, users would feel guided, reduce decision fatigue, and stay motivated through a sense of progress.
And if we helped them integrate their tools before entering the product, they’d land in an environment that already felt like theirs. This leverages the endowment effect, people value what they’ve already invested effort into.
The Redesigned Flow
I restructured onboarding into a step-by-step journey:
Login → immediate entry, no distraction.
Select your goal → framing intent clarifies why they’re here.
Choose integrations → anchoring the product to their daily tools.
Select repos/projects → deeper personalization, sense of ownership.
Product tour (while data loads) → perceived progress during the 10–15 minute wait.
Land on the app with insights already there → payoff at the moment motivation is highest.
Psychology at play
Linear Flow: Reduced cognitive load by narrowing choices and guiding users step by step.
Progress Indicators: A progress bar created a sense of forward motion (perceived progress). Even a small visual nudge increased commitment.
Endowment Effect: By the time users landed on the app, they’d already “set it up” with their integrations. The app felt like their workspace.
Wait-time Utilization: Instead of staring at a spinner for 10 minutes, users were taken on a guided product tour. For some it was delightful, for others skippable—either way, we converted dead time into learning time.
Twists and Turns Along the Way
It wasn’t a straight path to get everyone aligned on this flow.
“What if they just want to skip around?”
“We don’t have precise data to show how long integration will take. It could be 2 mins or 15.”
“Won’t users drop off before they even see the product?”
That gave me the chance to explain the empty state problem.
“Showing the product without data is like handing someone a restaurant menu with no food on it. It doesn’t excite, it creates doubt. Let’s integrate first, then reveal value with their own data.”
Once framed this way, the team understood that excitement doesn’t come from seeing the product, it comes from seeing the product, with your data.
Before
A 2 step process. Scattered checklist which users mostly abandoned before connecting data.
After
Linear, guided onboarding that built momentum and delivered value faster.
Product Tour After Data Collection
Linear, guided onboarding that built momentum and delivered value faster.
Success Criteria
We know the onboarding redesign is successful if:
Onboarding Completion Rate: % of new users completing the entire onboarding flow increases from ~10% to at least 50–60%.
Activation Rate: % of new users reaching their first insights with their own data increases significantly.
Time to Value: Users reach the “aha” moment. i.e seeing dashboards populated with their own data within the same session, instead of abandoning during setup.
Drop-off Reduction: Sharp decrease in abandonment after login and during the integration wait-time window.
Qualitative Feedback: Fewer complaints from sales/marketing about stalled trials; positive signals from users in product tours and demo calls.
The Impact
Onboarding Completion Rate
Completion jumped from ~10% to over 60% within three months — a 6× improvement. More users were making it through the full flow without abandoning.
Activation Rate
The percentage of users reaching their first insights with their own data increased significantly. Instead of seeing an empty product, most new users landed directly on dashboards populated with their repos and integrations.
Time to Value
Users now hit their “aha” moment within the same session. What previously took multiple visits (or never happened at all) now happened seamlessly during onboarding.
Drop-off Reduction
The sharpest gains were at two critical points:
Post-login: fewer users abandoned after first landing.
Data loading window: the product tour kept users engaged instead of leaving during the 10–15 minute processing delay.
Qualitative Feedback
Sales reported far fewer stalled trials. Marketing flagged fewer complaints about “confusing onboarding.” In user calls, feedback shifted from “I didn’t know what to do” to “I liked how it guided me step by step.”
Reflection
Users don’t care about seeing your product.
They care about seeing their product [read: value]
With their data.
Moving them forward.
MORE work
