Multi-day Rental: Gig Car Share

Context:  Gig Car share is a mobility start up that focuses on providing an app-enabled car-sharing service similar to that of scooters.  The idea is that you open the app approach a car, unlock it, and do a short trip.  Cars can be parked anywhere streetside in the home zone and you are charged by the minute/hour.  Gas is free with the use of a gas card, and parking is free in the home zone.  With COVID utilization plummeting and to encourage usage cleanings were done much more frequently but no one felt as safe as before.  This experiment occurred in ~march of 2020.

Proposal: 

• Offer a long-term rental with the promise of cleaning before and after the ride to establish trust. 
  • Marketing pushed content showcasing the deep cleaning with disinfectant to build customer trust.
• Experiment with duration of days with a segment of people who have taken long trips in the past.  Prove out the MVP before investing resources in changing the product to showcase this offering.

Constraints: 

• There is no way for the customer to access this service directly during the experiment.  Messaging (email and push) all encouraged users to call the call center.
  • The call center would manually move the user to the correct user group that was associated with select cars for this offering.
    • They would be manually charged a lump sum, moved to the correct group, and a specific set of cars that were recently cleaned became accessible.
• The product is built to support automated pricing up to 24 hours, with distance caps and overages within a single day.  Everything over 24 hours requires manual resetting and tracking of “overage miles”.  This took significant bandwidth and was very prone to human error.
• There was no way to “restrict” the customer from moving from car to car once they were in this rental period.

Experiment:

• About 1000 people were chosen to participate and about 1000 people actively took the offer of 5 days for 300$

Experiment Results:

• The results of this experiment were remarkable positive and briefed as a win, showcasing a ~10% increase in revenue when compared two group
• This win was championed by the marketing team and presented quickly with tight siloed analysis between marketing and a single analytics employee.  The leadership team digested these highlights and pushed to implement them further.  After the decision was made the analysis was shared. 
• The entire company was hungry for a win and only two members of the product team unpacked it to understand the process
  • Red flags
    • Because of the short duration of the experiment, no maintenance costs were considered
      • Due to the manual nature of tracking distance and overage costs less than 20% of trips that exceeded 250 miles were held accountable.
      • Trips went dramatically further (including several states over) to more remote locations.
  • To expedite the analysis the marketing/analytics team agreed to forego A/B testing (the standard at the company) and simply compared the 1000 person group to itself, just 4 weeks earlier.
    • With lockdowns in the bay, there was a huge disruption to all things transportation.  As the lockdowns lessened the business as a whole “increase in revenue” over time.  There is no way to tell if it was due to normal progression or this experiment.

Impact:

• I started a “tiger team” to evaluate our experiment process to ensure appropriate stakeholders would be involved to avoid poor experiment hygiene.
  • Several members of the team were disappointed in the lack of transparency during the process.
• The product offering was implemented and was immediately “successful”. In the first month, there was a 30% increase in revenue, and 20% of total revenue was because of this offering.
• The product roadmap was shifted to prioritize the user interface for this offering.  Buttons and automation took about 12 weeks to fully implement.
• Cons:
  • Operational costs increased from 0.16$ a mile to 0.24$ in less than 1 quarter.
    • This is calculated by monthly operational total cost / total mileage
  • Parking tickets increase roughly 3x, with less than 60% recovery
  • Tows increased by 2x due to communications issues in remote areas. This results in 2 instances of media fallout.
  • Call hold time increase from an average of <15 seconds to ~20 minutes during business hours.
    • Because calls are not triaged, all calls waiting in the same queue.  This includes calls for simple fixes or questions to emergency “I’m stuck in the woods” types of calls.
  • Trip satisfaction average decreased from 4.8 to 4.5 across the entire market.

Business Rational: The Proposal

What Prompted This Project: COVID utilization

Descriptions of Details and Execution: Risk/Cost Analysis (Done too hastily and in a vacuum)

Descriptions of Details and Execution Part 2: Experiment

What insights were gathered from this experiment: Experiment Results

What was the results: Impacts

Lessons Learned and Recommendations: