Redesigning Ride-Hailing as a Conversation

Case study in the series behind How I Whiteboard Product Interviews Live. Method: The Live Whiteboard Method.

These were two Instacart whiteboards in the same round, and they look unrelated until you notice they are the same move twice: find the place a problem actually lives, and design straight at it. The first board was an accessibility prompt. The second was a retention prompt. Here is how each one went.

Board one: design an accessible ride-hailing experience

The prompt was to design a more inclusive ride-hailing experience for visually impaired riders. The fast, wrong version of this is to add a screen-reader pass to the existing app and call it inclusive. I did not start there.

I grounded "visually impaired" as a spectrum

The first thing on the board was a refusal to treat "visually impaired" as one user. I drew a spectrum: color-blindness at one end, low vision in the middle, full blindness at the other, and asked the same question at each point. What breaks here? A color-blind rider and a fully blind rider hit completely different walls, and a design that helps one can be useless to the other. Accessibility is not a checkbox, it is a range, and naming the range is what keeps you from designing for an average user who does not exist.

I redesigned the visual UI as a conversation

The reframe that unlocked the board: stop treating this as a visual interface to be patched, and redesign it as a conversation. I drew a user input / system output model and wrote out the full verbal flow the way a rider would actually experience it.

The system asks "what are you wearing?" so the driver can find the rider. It says "your car is thirty feet ahead of you" instead of dropping a pin on a map the rider cannot see. It prompts "make sure to ask the driver their name" before getting in. Spoken out as dialogue, the experience designs itself, and gaps a visual mockup would hide become obvious.

Writing it as a conversation also surfaced a whole safety cluster that a sighted designer would simply never reach: confirming the driver's identity by voice, knowing where the car is in physical space rather than on a map, having a verbal check before entering the vehicle. Those are not nice-to-haves for this rider. They are the difference between a usable product and an unsafe one.

I closed the loop with metrics and a validation plan

I do not leave a solution un-instrumented, so the board ended with how we would know it worked and what we would test before committing. Success metrics: the percentage of visually impaired riders who complete a ride, and a decrease in time to book. A validation plan: clickable prototypes, usability testing with actual visually impaired riders, and a verbal script you can run by reading the flow aloud and seeing if a person could complete a ride from the words alone.

Board two: grow membership retention

The second prompt was different on the surface: a newly formed retention team needs to reduce churn and grow membership. Where do they even start?

I found the crux on the funnel

I drew an activation funnel: first order, then second order, then third order. Then I circled the second order in yellow, and the entire strategy condensed into one insight. The jump from the first order to the second is the leverage point. People who place a second order behave like members. People who stop at one were never really activated. So the retention team should not spread itself across the whole funnel. It should obsess over one transition: getting a first-time orderer to come back once.

That is the crux move. Most of a strategy is figuring out which single step decides the outcome, and then refusing to be distracted by the others.

I reasoned with real numbers, even illustrative ones

To pressure-test it, I reasoned out loud with concrete numbers on churn timing and membership conversion, even where the exact figures were illustrative. The point of putting numbers on the board is not false precision. It is to show you are thinking in magnitudes, that you know roughly how big the second-order drop-off is and what moving it would be worth. A strategy that cannot estimate its own upside is a wish.

What both boards were really showing

Two different prompts, the same way of thinking:

• Design for the edge as a spectrum, not as a single imagined user.
• When a visual interface excludes someone, change the medium, do not just patch the surface.
• Never leave a solution un-instrumented: name the metric and the way you would validate it.
• Find the one step that actually decides the outcome, and aim the whole strategy at it.

The accessibility board found the move by changing the medium. The retention board found it by circling one transition. Same discipline: locate where the problem really lives, then design straight at it.

If you want to see this live, bring me a real product problem and watch me whiteboard it end to end: that is the Watch Me Think offer. To practice the method yourself, the Whiteboard Prompt Translator scaffolds the six frames from any prompt.