In a world filled with hospitality options, Sonder stands out for its unique combination of beautiful spaces and mobile-first service. Combined with the shift toward remote work and longer destination-focused stays, Sonder’s explosive growth in the travel industry seemed inevitable.
Persona has been a long-time partner of Sonder since 2020, whose operational model of contactless stays necessitates an automated check-in method powered by digital identity verification. But as Sonder expanded globally, its Trust and Safety team, which is responsible for identifying problematic guests and fraudulent reservations, wanted to scale and streamline their processes without adding friction for guests.
Problem: Sonder's T&S team faced a backlog of manual reviews and sophisticated fraud during its hypergrowth
With Sonder’s booking volume growing 45% year over year, its Trust and Safety team was spending an increasing amount of time manually reviewing tens of thousands of upcoming bookings at Sonder properties. This influx of bookings also brought bad actors who continuously evolved their fraud tactics to make it past manual review, leaving the team feeling like they were playing “whack-a-mole” with fraudsters and people looking to break house rules, such as throwing parties.
As such, Sonder began searching for a way to scale its bookings approval process without throwing people at the problem or sacrificing the ability to catch these risky guests.
Solution: Sonder streamlines manual review and fraud detection via Persona-powered machine learning
Sonder knew if it wanted to streamline its manual review process, it needed to be able to consolidate external signals with data from its own systems. As it already uses Persona to verify users during check-in (and during other parts of the guest experience), it decided to partner with Persona’s product and data science team to build a custom machine learning-powered monitoring system that either automatically approves bookings or sends risky bookings to the team’s manual review queue for further investigation.
Alex Sheaf, the head of the department, explains that Sonder intentionally chose to identify potentially fraudulent bookings via a machine learning approach rather than relying on a legacy rules-based approach to keep up with rapidly changing fraud patterns.
Since implementing bookings monitoring via Persona, Sonder's Trust and Safety team has seen significant benefits:
Improved guest experience due to a 75% decrease in manual review time
Before Sonder’s automated bookings monitoring system powered by Persona was implemented, guests who booked a stay with Sonder often had to wait for an extended period to hear if their reservation was approved, as Sonder’s trust & safety team worked through the manual review queue. This was frustrating for both guests and Sonder’s Trust and Safety team, who want to provide the best possible experience.
With Persona’s platform now automating most decisions, Sonder has reduced time spent manually reviewing bookings by 75%, reducing the cost associated with a booking by an equivalent amount. This gives Sonder’s team more time to focus on the truly risky bookings, as they know these have been flagged by the model.
Thanks to this new process, Sonder has also seen its NPS increase, as guests hear back on the status of their booking in hours rather than weeks.
Nuanced risk detection powered by machine learning
Rather than having to manually factor hospitality-specific concepts like length of stay into its decisions, Sonder can easily model these signals inside Persona without its own dedicated data science or engineering resources.
Persona’s machine learning model incorporates over 50 distinct signals — from guest data collected by Persona to booking information like length of stay and how far a guest books in advance — which are ingested via Persona’s Workflows and Transactions. This flexibility makes it easy for Sonder to bring in key business data at the start and layer in additional signals over time to improve future iterations of the system.
“There are many Sonder-specific signals that contribute to whether a guest’s booking is risky, and it was simple to incorporate them all into the model. We can easily add more over time and keep evolving our risk model without having to retrofit a signal to Persona.”
By assessing signals from across the guest experience, rather than just identity or payments data like with traditional transactions monitoring solutions, Sonder can automatically approve around 85% of bookings while identifying fraud in a more nuanced way than a manual reviewer could accomplish on their own. The machine learning approach makes it possible for the team to adapt quickly to new patterns of bad actors and fraudsters rather than having to endlessly orchestrate dozens of possible rules.
Comprehensive fraud protection with integrated identity verification and transaction monitoring
Since 2020, Persona has seamlessly grown from Sonder’s identity verification provider to a full-on onboarding, fraud prevention, and monitoring partner, according to Alex. The chief reason? Persona makes it easy to leverage and compare data from one interaction a guest has with Sonder to any other interaction, minimizing friction for both guests and the Trust and Safety team.
For instance, Sonder can catch instances of non-compliance with its terms of service by comparing the ID used during check-in with the personal information on the booking. Persona even helps Sonder catch stolen credit card use by comparing IDs used during bookings against information from the credit card transaction.
These verification touchpoints give Sonder multiple ways to defend against fraud, freeing up its team to focus on the bookings that require the most attention.
As the booking monitoring system incorporates more customers and signals, Alex and team hope Persona will continue to get even better at detecting fraud. As he puts it, “We have all this data we want to bring to Persona to create better fraud signals. And eventually, we want to be able to find risky guests at not just the point of booking, but at any other point of the guest lifecycle.”
Result: Sonder manages risk across the guest experience by balancing automation and manual review
While Sonder previously thought that its business model required a labor-intensive trust and safety process, it now knows that a balance of automation and manual review inside of Persona is the key to keeping up with its hypergrowth.
The most recent use case for Persona may be booking monitoring, but Alex believes it won’t end there. As he puts it, “The Persona team feels like an extension of ours. For new solutions, we choose to go with Persona since we can customize to our needs and requirements while leveraging what we already have versus having to find more vendors.”
With Persona serving as their command center for all things identity and fraud prevention, Sonder is now free to focus on what it does best — giving guests a better place to stay via efficient, tech-driven operations.
