Automating the billing of Tesla Supercharging
Background
Roughly 90% of the electric vehicles (EVs) listed on Turo are Teslas.
When a guest is renting a Tesla on Turo, they will inevitably need to charge the car and most often will go to a Tesla Supercharging station. Unlike a traditional gas station, though, no credit card or payment method is required at the time of charging - the owner of the Tesla is automatically billed to the payment method on file for their account.
While this is extremely convenient for guests, it creates work for hosts: after the trip ends, a host must collect charging receipts from their account, add them up, and send the guest a bill through Turo.
Problems
Gathering evidence and invoicing is tedious + blocks hosts from scaling
Through talking with hosts, we quickly understood how time consuming it could be to invoice guests their Supercharging when hosting only one or two vehicles. However, hosts who wished to scale and create large fleets of Teslas cited that the billing process would become overwhelming, and require them to either:
- take a loss by not invoicing guests
- hire support employees to handle the monotonous task of collecting evidence and invoicing
In either case, hosts saw this as a significant blocker to their ability to profitably scale their businesses.
Manual billing results in support costs (manual agent review)
After hosts collect evidence and bill their guest, there is a chance that the guest will either:
- dispute the bill
- miss notifications for/ignore the bill
In these cases, a support team at Turo would manually review the contents of the invoice and determine if the guest owes the amount, or if the host is potentially abusing the invoice system.
Annually, these types of support reviews cost Turo between $3.75M and $6.25M.
Manual error can lead to denied invoices + host dissatisfaction
The invoicing process for Supercharging requires the host to not only collect up all charging receipts as evidence, but manually add them to produce a single total. This could result in manual errors and miscalculations.
Small arithmetic errors might not seem like a big deal, but Turo's policies at the time (to curb hosts abusing the invoicing system) was to deny an invoice if there were ANY inaccuracies (even $0.01 off).
Hosts could always resubmit an invoice in this case, but the damage would be done: hosts already felt they were putting in too much work only to be denied for potentially trivial errors.
Project goals
Increase host efficiency when invoicing for Supercharging
Through an integration with Tesla via a newly released Fleet API, we believed that we could automatically pull a list of Supercharging transactions that occurred during the guest's trip.
Since these transactions would be from a verified source of truth (Tesla), hosts could forego collecting/providing receipts and avoid potential math errors.
Reduce support costs associated with invoice review
Similar to the host benefit, any invoice line items which were automatically pulled from Tesla could be deemed as 'pre-verified' and thus would not need a full review, allowing agents to focus on reviewing other aspects of the invoice.
If an invoice were to contain ONLY pre-verified transactions, it could be automatically adjudicated by the system and not require manual agent review at all.
Key metrics
Primary
70% of Supercharging invoice items automatically generated
Secondary
Reduction in support 'contacts' (manual reviews)
Increase in host CSAT
Early stages
Field research: Hertz
I rented a Tesla through Hertz to better understand how Teslas work, as well as any Hertz-specific policies that might inform our work.
Field research: Turo
I also rented a Tesla through an employee-host at Turo to better understand the experience we were offering, as well as being able to observe the invoicing process alongside the host.
Speaking with hosts
Sat in on research calls with hosts to understand their experience hosting Teslas on Turo, primarily focused on their pain points with Supercharging and invoicing.
X-functional ideation workshop
I lead an early ideation workshop with product and engineering partners to share what context we had so far, and open up a space to folks to provide early input into the direction as well as any additional considerations from a project perspective based on experience with similar projects in the past.
Mid-stage
Laying out preliminary user flows
Based on preliminary requirements, we knew that we would need flows to handle:
- hosts linking their Tesla accounts to enable us to make API calls
- management functionality for their linked account
- interface to review and modify transactions that were pulled in via the API and send them to their guests.
End-to-end wireframes
For each of the core flows we identified as a team, I created rough wireframes to block out the screens and serve as a foundation to begin iterating and gathering feedback on the flow and interface.
Refined through multiple rounds of team-wide feedback
During this time, I would frequently bring design explorations and iterations to weekly "Show and Tell" meetings, where designers across the org could provide feedback and weigh in on the direction.
Usability + validation testing with hosts
With designs in a decent state, we tested prototypes with a panel of 10 hosts (mix of internal and external) to gather feedback and double check we weren't going off the rails with any of our proposed flows.
Insights through testing
Hosts were excited about the feature, and had no issues completing tasks using the prototype
Hosts within our testing cohort were excited to see this much-requested feature coming to life.
Ability to manually add additional receipts needed
One host brought up a very important edge case during test: often guests would be running late when returning due to trying to charge the car last-minute, not expecting it to take very long. As a result, guests were often doing that charging past the actual trip end time - meaning that it would not be automatically pulled in via API calls.
As such, we would need to ensure hosts had the ability to manually add additional Supercharging costs/receipts to account for those missed by the API.
Hosts desire lenience with 'hidden' idle fees
Most guests renting a Tesla have never driven one before, and are unaware of certain Tesla-specific policies.
One of these policies is the 'idle fee': a car left at the charging station after it has reached the maximum amount will begin to incur a by-the-minute fee based on how busy the station is.
Hosts mentioned during testing that if the guest doesn't seem like a bad actor, they would often eat the cost of the fee in order to preserve a positive guest relationship.
As such, we needed to ensure that these fees could be displayed separately (if they existed) so that hosts could make the choice to include or exclude them based on what they believed was best in the situation.
Late stage
Applying feedback from testing
Based on the insights from testing, we continued to refine and converge on final designs.
Formal design review
Every project at Turo must undergo a formal design review, attended by anyone interested and any Director+ team member.
This serves as an opportunity for anyone across domains to provide input on the final design or catch any opportunities to create parity between product areas.
Engineering kickoff + collaboration
Once the designs were fully reviewed and signed off on, we held a kickoff with the engineering team to review all the flows and ask question/provide feedback.
While the engineering team was building the flows, I would attend multiple weekly engineering syncs to field additional questions, work through new information/complications with the Tesla API, as well as make adjustments to the design specs based on feedback or uncovered constraints.
Finished product
Flow: onboarding and linking a Tesla account
Flow: reviewing automatically generated invoice items + adding manual items
Outcomes
Metrics
On average, 72% of Supercharging invoice line items were automatically generated.
Annualized support cost savings of $800k - $1.2M.
Reception from hosts
We were excited to hear almost exclusively positive reception from Host Success Managers (who worked closely with hosts) and hosts themselves.
