Thumbtack: Customer Booking Experience

Biggest lever – Scheduling Step

The scheduling step hadn’t been touched in a few years. Our goal was to reduce drop-offs at this step, by running an experiment.

Project Goal

Get more customers to a booking by improving how we capture their scheduling preferences. We do this to provide a simpler, quicker and more relevant booking experience with pros.

Quantitative goal

The baseline conversion (direct leads flow, instead of booking flow) for this same step was 10% higher. We wanted to close this gap and increase booking project share by 10%.

Diagramming, as my thought tool of choice

Unhappy with the ops-driven solution, I brought this diagram to the team for a discussion. My engineering and analytics partner added their input that changed the course of this experiment.

Customer hypothesis

Capturing customer availability first and showing them time-slots that overlap with pro availability will reduce drop-off and increase booking share.

Benefit to the customer?

• We’d be doing the heavy lifting for customers and presenting bookable dates and times that work for them and their chosen pro.
• We’d improve the UX to present the options in a scannable and easy to select way.
• We’re capturing their preferences first >> giving them their desired flexibility (based on the fact that 15% customers choose to suggest availability today)

Setting up the experiment

I worked with my PM and analytics partner to define the control group and variant based on the diagrams above.

Designs

Below, are sample screens for the variant group that I designed. The solution had to balance several business, UX, interaction and engineering considerations.

I collaborated with a content designer for the copy on the screens. The engineering team was consistently engaged to ensure the designs were feasible. This included one special case, where technical challenges had to be manoeuvred together to push the experiment out.

Design iteration

The initial solution presented one time-slot that matched their preferences and the pro’s. However, given how the pro’s availability was set up in the database, this timing might not have been suitable for the customer. For example, the time-slot, most times would’ve been between 6AM and 9AM, which is clearly too early for any job.

Our goal was to ensure that customers could move past this step as quickly as possible, with a time-slot that worked for them. The engineering lead and I brainstormed an iteration and the logic:

Logic

We would show the first available and overlapping time-slot from each group of timings, so that we can scatter the hours in the day providing varied options to the customer.

Experiment duration

4-5 weeks

Results

We hit our target!
Booking share was 10% higher in the variant group.
The scheduling step experiment formed the foundation for the ‘Booking Upsell’ experiment.

New data from the analytics team

Independent of the above experiment, the analytics org shared that the ‘hire rate’ for a direct lead project was a lot higher than previously believed. This meant that we no longer could afford the conversion loss that was coming by pushing customers into a booking flow.

Note: True percentage data is hidden due to confidentiality.

Problem

Customer

A lot of low intent customers are entering the booking flow and dropping off as a result.

Business

There is currently no fallback experience to capture these customers who drop-off, hampering our conversion goals.

We’re losing customers who could have led to a hire if they created a direct lead project instead.
‍

Trade-off

We don’t want to hamper the experience for high-intent customers who would’ve made a booking anyway from doing so.
‍

Solution

By re-arranging the steps of the direct leads flow, we will be able to create a DL project for customers right after the scheduling step.

Right after this step, we will recommend that the customer continues to book a specific time-slot by bringing in a strong value proposition. If they choose to do it, they upgrade to a  booking. Else, we’ve captured them as a direct lead project.

This came from the lead engineer

The proposal of this flow came from one of the engineers, who was closest to the technical feasibility of the above solution. My fear with this solution was that it created a project for a customer without much clarity, which could lead to accidental projects and thus downstream cancellations.
‍

Diagramming, as a tool for discussion

To facilitate a discussion with my cross-functional partners, I used diagrams to illustrate the proposed solution as well as some alternatives.

Design principles to the rescue

However, it became clear that the suggestion from the team had higher chances of helping us hit our targets and balance the trade-off.

My role was then to ensure that the customer was well-informed and ‘hand-held’ through the process, so that they don’t make accidental bookings.

1. Over-communicate

• Use information to provide clarity about the impact of their actions at every step
• Reinforce actions that have longer term, multi-side implications (project was created)
• Surface the relevant information at the right moments to soften the commitment a customer needs to make. I.e. Provide information to reinforce their freedom at every step

2. Flexibility over direction

• Let the customer decide what kind of booking they’d like to make

Pre-launch business implication

When a PM from an adjacent team reached out and expressed concerns about certain revenue implications of this solution, I worked with her to dive head-first into the revenue and engineering systems to solve for those implications. I detailed out user flows that ensured that we balanced revenue with the right user experience for this experiment.
‍

Next steps

Given that this experiment was directionally positive, it is now shipped nationwide. The team’s next responsibility is to increase booking share with this experience.