Optimizing Thimble's Checkout Experience

A conversational checkout experience designed to maximize conversion based on the users industry.




Thimble sells flexible, short-term business insurance for the gig economy, with policies available by the hour, day, or week, up to a full year for growing businesses reaching the next level. Policies can be purchased in seconds via the web or app, with the ability to modify, pause, or cancel instantly whether work slows down or hiring picks up. This innovative product-offering in a truly archaic industry earned Thimble the 1st place spot on Fast Company’s Most Innovative Companies, in the "Small and Mighty" category.

Project Background

In 2021 Thimble launched three new insurance products which added additional steps to the funnel, more sophisticated underwriting, and increased pricing. These new products improved our unit economics, but at the cost of our conversion. After dropping by 22% in the first month, I was assigned the task of optimizing conversion.

I dedicate one week to data analysis and research, creating Amplitude dashboards with high-level conversion and a detailed funnel analysis, watching FullStory sessions, and calculating the tradeoff between higher conversion and higher order value. There were 2 primary points of drop-off:

  1. The Duration Picker: Thimble's initial product was our short-term, flexible policies by the hour, day, or month, but these short-term policies are only available for certain coverages. If a coverage like Business Equipment Protection is included, you have to buy a Year-Long policy, in the form of either Monthly Payments or a 1-time Annual Payment.
  2. The Coverage Selection Page: The page that advertises our newly released insurance coverages, and allows users to configure which coverages and limits they want. In an effort to increase cart size (Average Order Value), we had launched this page with most of the coverages defaulted to on.

After performing more detailed analysis on these screens, it was clear there were patterns based on the users industry. I created an experiment roadmap, top heavy with quick experiments aimed at offering certain gig-based industries (like DJ's and Photographers) shorter policies with less coverage, while maximizing the coverage offering to more established industries (like Contractors and Plumbers).

[.icon_star][.icon_star] Highlight

Over 3 months, the experiments I ran led to a 27% increase in conversion, while generally maintaining Average Order Value.

The Duration Picker Screen

Our traditional General Liability policies are available by the hour, day, or month, which is our strongest differentiator from competitors. Unfortunately due to regulatory restrictions, these new coverage were only available in the form of long term policies. In an effort to increase Average Order Value, we launched these new products by asking users to select their desired coverages first, and then displayed the available durations. If any of the newly launched coverages were selected, you were ineligible for short-term coverage.

Initial Duration Picker Screens, Based on Selected Coverages

The Challenge

The positive: we had a lot of users opting to proceed with additional coverages (but more on that later). But after proceeding, they'd discover that short-term policies were not available and grow frustrated. The screen did not provide a lot of clarity as to why short-term wasn't an option.

It was pretty clear that by explaining to users how to get a short-term policy, which was only buying General Liability and Professional Liability insurance, we could drastically increase conversion. But this was a classic case of whack-a-mole, Thimble had a number of targets, including very aggressive and important Average Order Values targets. Updating the experience to drive every user to short-term would have a heavy impact on our attachment rates (since these new products can't be purchased short-term), impacting Average Order Value.

The Solution

The experiment was to give users in select industries the option to select their duration up front, while highlighting the value of long-term policies that offer additional coverages. This would allow us to increase conversion of users looking for 1 day policies, where we didn't stand much of a chance converting them to annual policies. Data showed that even prior to launching these new coverages, there were clear segments of industries that were more inclined to buy long-term coverage, because they worked....long term! Whereas certain industries were heavily lopsided to short-term coverage, including some industries that were 99% short-term.

This experiment allowed certain industries to choose their duration first, highlighting the values of our longer-term policies and enhanced coverage offering, while still allowing short-term.

Overall conversion from start to purchase over time: When we launched 3 new insurance products which included additional questions and higher prices, conversion tanked. A month after, I put together a roadmap of 8 optimization experiments to run in 2 week intervals. 7/8 experiments increased conversion, ending in a 27% increase.

[.icon_star][.icon_star] Highlight

This experiment successfully increased conversion by 21% with only a 1% drop in Average Order Value, ultimately increasing revenue by 18%.

The Coverage Selection Screen

In rolling out these new coverages, we continued our approach to increasing Average Order Value by defaulting a lot of coverages to on. This resulted in some fairly jarring pricing being displayed, when users had just shopped around for General Liability in a few other places.

The coverage selection screen initially defaulted coverages to on to increase Attachment Rate and Average Order Value. We also only displayed 1 total price.

The Challenge

Again, Thimble had to prove these new products were a success, and that we were capable of launching and selling additional coverages to our customers. We have the option to make them fully optional, but that could have had a huge effect on our bottom line.

The Solution

Using data, I segmented industries based on their likelihood to buy these individual coverages. Now that we were allowing duration-selection first, we were already dealing with users that were committed to long term coverage for long term work. We could leave certain coverages optional and rely on marketing to upsell them, given the actual importance of these coverages to full-time businesses.

Each industry was assigned a value of defaulting to on or off for each coverage based on their prior behavior. I also updated the screen to included individual pricing for each coverage to entice users to add-on coverages for a small cost.

[.icon_star][.icon_star] Highlight

Over 3 months, this experiment increased conversion by an additional 16%. Attachment rate for two of the new products stayed above 70%, with Business Equipment Protection dropping from 75% to 41%.

The overall increase in conversion resulted in a net positive on Revenue, and we decided to roll this out permanently.

The updated flow defaulted different coverages to "on" based on the users industry, and the relative attachment rates across similar customers.