A leading on-demand service provider in South-East Asia, who specialised in ride-hailing and logistics, wanted to remodel the existing driver incentives structure. The service provider wanted to reduce daily driver incentives, which accounted for nearly 40% of the drivers’ everyday income, and instead boost the organic income received on trip completions.
TheMathCompany identified gaps in the current incentives scheme and noticed that the payouts were disproportionate across tiers and did not effectively motivate/challenge drivers to accept more trips. A new personalized incentives scheme was outlined to improve driver trip acceptance rate and reduce disbursed incentives. The ideal distribution of drivers was identified as a pyramidal structure with the highest number of drivers in Tier 0 (non-bonus tier) followed by Tier 1, 2 and 3, which would be effectuated through the revised incentives scheme.
- The drivers were divided into control and experiment groups in the ratio 30:70 to test out the new incentives scheme for a 3-week duration. The experiment group had to be carefully picked out as drivers were highly sensitive to any sudden change in the incentives structure and were a part of a close-knit peer network. Select drivers, who were consistently in the non-bonus tier (Tier 0), were chosen to be a part of the experiment group that would be targeted under the new incentives scheme. The original incentives structure would apply to the control group
- The new incentives structure had an intermediate tier below Tier 1, focused on motivating drivers in Tier 0 to move up the incentives tier system. The incentive amount was calculated such that the bonus incentive for the intermediate tier was lesser than the bonus incentive for Tier 1. Personalized shuffle cards and push-notifications were sent to drivers to inform them about the change in incentives structure
- Presuming that the experiment group would be more motivated to move up the tiers under the new scheme, the number of trips available for drivers to reach higher tiers was automatically expected to reduce, and thereby help in achieving the ideal driver distribution
The outcome of the experiment was measured in terms of trip acceptance rates and incentives paid out per trip across experiment and control groups.
- The experiment proved that personalized incentives schemes can work for different segments of drivers
- It recorded a ~70% reduction in total incentives disbursed to the experiment group with a 13 % reduction in average number of trips completed by top-tier drivers which helped in optimizing driver distribution across tiers