RILLA · ONBOARDING · CASE STUDY · 2025

Self-Serve Onboarding

I replaced Rilla’s 1:1 manager onboarding with a self-serve system that teaches the product through practice.

It moved day-one return from 5% to 40% and gave each CSM back ~3 hours a week.

Role
Product Designer
Team
1 designer, 2 engineers
Timeline
Sept 2025 to March 2026
Shipped
All managers, March 2026

Coaching was too complicated to learn alone

Managers coach reps through ride-alongs: reviewing recorded conversations, finding teachable moments, and leaving feedback.

The workflow was complex enough that every new manager needed a 1-hour onboarding call with Customer Success. Many never scheduled one.

The Rilla app, dimmed and blurred, behind the modal new managers saw before self-serve onboarding. Under the eyebrow "Ridealongs made easy" it reads "Welcome to the End of Ridealongs", then: "Rilla has built something new to make your ridealong experience 10x better. You'll review faster, coach better, and never go back to the old ways again. Let us show you how! This will not cost you anything extra." Below an onboarding video, a yellow button reads "Book your Personalized Onboarding", with "Already booked? Dismiss" underneath.
Before: getting started meant scheduling a call with Customer Success.

Our goal was simple: teach managers to coach without needing a CSM.

The Rilla app: a ride-along list with one conversation open.
For this onboarding, use your keyboard only! Pressenterto continue.
You’ll be learning how to use your keyboard only so you can coach in super speed.
Hit enter to start your ridealong!
enter
End of prototype

Dummy ridealongs start here.

After: managers can start learning the product immediately, on their own.

We built onboarding one friction point at a time

Every day, I watched session replays of every manager who onboarded, paired them with Mixpanel and Grafana data, and shipped another iteration.

HOW, EVERY DAY watch every replay isolate friction ship it that day measure REPEAT

Watch every replay → isolate friction → ship → measure → repeat

Not one big redesign, but a series of small corrections backed by real behavior.

Completion

First, managers had to actually finish onboarding

Our first version asked managers to coach a real rep immediately. For someone still learning the product, the stakes were too high.

So I built dummy ride-alongs where managers could practice without sending feedback to a real rep.

A dummy ride-along in Rilla, titled "Dunder Mifflin // Practice Ridealong #1". Two utterances from Michael Scott and a customer sit above a media player. Everything is dimmed except the comment box in the right gutter, which holds a comment already written for the manager: "Michael, try to build rapport with the customer before you start bragging!". A tooltip above it reads "Hit enter to send your first comment!".
A practice call with nothing at stake. The comment is pre-written so managers can focus on learning the interaction.

Teach one core loop

I stripped the experience down to three actions:

Find a moment → Leave a comment → End the ride-along

Rilla was keyboard-first, but many managers reached for the mouse. So each action got an obvious button with the shortcut built in, teaching the keyboard instead of assuming it.

The same three actions, whether you click or use the keyboard.

One problem solved

Those iterations moved onboarding completion from 50% to ~100%.

ONBOARDING COMPLETION before · sept 2025 50% after · march 2026 ~100%

But solving completion exposed the next problem.

Retention

Then we had to get them to come back

Nearly everyone could finish onboarding. But only 5% came back the next day to coach.

Watching their sessions revealed what happened: managers left the guided experience, returned to the full product’s wall of tools, and froze.

Finishing onboarding wasn’t enough. We had to make the transition into the real product work.

WHERE THEY FELL OFF finished onboarding ~100% came back the next day 5%

Unlock complexity as managers become ready for it

Borrowing from how games introduce mechanics gradually, I built an unlock system that introduces tools as managers complete real ride-alongs.

Core coaching → Rick AI after 3 → Script trackers after 6 → more over time

UNLOCKED BY COACHING, NOT BY TIME FIRST REAL RIDE-ALONG Jump Jump straight to the key moments in a call. AFTER 3 RIDE-ALONGS Rick AI Answers any question about the conversation. AFTER 6 Scripts Scores reps on their own sales process. LATER AI Summary Clips Voice comments Integrations Analytics Leaderboards 3 6 … completed ride-alongs

Managers keep practicing the core coaching loop while learning one new capability at a time.

Locked, but never hidden

Future tools stay visible even before managers can use them.

That preserves the shape of the product while giving managers something to work toward. Coaching their reps is the only thing standing between them and the next tool.

An annotated diagram of the Rick AI panel while locked. The tab bar and heading are unchanged; a grayed wireframe stands in for the conversation, labeled "sneak peek of locked feature". Over it sits a card reading "3 more ridealongs to unlock Rick", labeled "how to unlock feature", above "Rick answers any question about the conversation", labeled "what the feature does".
Locked, but never hidden.

Make learning feel like a reward

Unlocking a new tool shouldn’t feel like being handed more software to learn. So I used animation and sound to make each unlock feel earned.

Rick unlocking after enough ride-alongs. Best with sound on.
Outcome

From manual onboarding to managers coaching on their own

50% → ~100%
Finish onboarding
5% → 40%
Come back on day one to coach
77%
New orgs active after a week
~3 hrs/week
Returned to each CSM
Across a 15-person CS team, that removed roughly 9 hours of onboarding calls each day.