Back to all articles

How We Outperformed Rapido Bike-Taxi App’s CPA KPI by 45% While Scaling Up the Campaign x13

Download the PDF version of the case study

Campaign Goals worked with Rapido, a bike-taxi app, to acquire engaged users for their Android app. Being India’s first and largest bike-taxi service, Rapido operates in 100+ cities and has 10M+ customers. The campaign goal was to acquire new paying customers (users who complete at least one ride on a bike or a taxi) in the targeted cities where Rapido operates.

Results at a Glance:

lower CPA

than the KPI

Campaign Scaling

within 4 months

Average number

of fare estimates per user


When we started running the campaign for Rapido, our initial approach was to generate lookalike audiences based on the loyal users that make frequent rides. In the first week alone, we managed to acquire users who would complete the first ride, thus outperforming the CPA (First Ride Complete) KPI. 

ML Model Ideation

Processing over 1.5M QPS of in-app inventory, our in-house DSP quickly accumulated enough data for the ML algorithm to establish a robust install-event model focusing on a younger audience with a moderate number of apps installed on their devices.

Our age prediction capabilities allowed us to identify the best-performing age groups and adjust the bidding strategy to focus on users whose predicted age was within the group boundaries.

Due to the fact that the Rapido app is very well-known
in the market, users tend to install the app on their new device in case they ever need it. However, users with a higher device age who install the app have a higher intent of using the app.

ML Model Ideation

After accumulating more positive signals (installs and target actions), the ML algorithm improved the model to optimize the campaign towards users who fit the defined targeting criteria and were more likely to produce the target action (complete the first ride). As a result, was able to bring in highly converting users, which led to the second campaign scaling x13 within 4 months. Programmatic UA Process Machine learning

Campaign Results

CPA (First Ride) Cost and Campaign Scaling

The initial user segmentation, based on a lookalike audience built according to the behavioral patterns of the app users, has produced CPA (First Ride Complete) below the KPI from week 1.

Towards the end of the exploration period, once enough data had been accumulated for the ML algorithm to build the install-target model, we started to gradually scale the campaign week-over-week. Within 4 weeks, we managed to increase the install volume by nearly 300% while generating a CPA (First Ride) cost 45% lower than required. 

Within 13 weeks, the scale increased x11 while still producing the CPA within the KPI.
The continuous ML model improvement ensured maintenance of the CPA within the KPI, which allowed for a massive campaign scaling x13 vs. the initial phase.

Graph 1. CPA and Campaign Scaling

programmatic campaign scaling

Bid Optimization

Considering that taxi use intent is time-sensitive, the ML algorithm altered the bidding strategy to bid slightly higher—just enough to win the auction—during the hours when users are most likely to use a taxi app.

This approach allowed us to bring the users with a higher intent of completing the target action (First Ride).

Graph 2. Correlation Between eCPM and CVR%

The graph displays a positive correlation between eCPM price and the conversion rate per timeslot: the higher the eCPM is, the higher the conversion rate is. While the eCPM is higher on those timeslots that are predicted to bring better-converting users, the lowest CPA (First Ride Complete) was achieved on the timeslot where the eCPM was the highest.

Another level of bid optimization was achieved by using our age prediction capabilities. The ML algorithm adjusted the targeting model to bid higher on younger users between 18 to 25 years of age, as this user group showed the best conversion rate while bringing the lowest CPA.

Finally, the ML algorithm was able to identify that users of high-end devices are more likely to use the Rapido app and optimize the bid accordingly.

“The team at was able to understand and deliver as per our business goals and KPIs right from the beginning with commendable transparency. With data driven experimentation and the platform’s high scalability, rapidly became an important contributor towards our acquisition goals.”

Arsh Bhatt, Performance Marketing Manager @ Rapido

About Rapido

Rapido is a bike taxi and auto service spread widely across all of India from Tier I to Tier III cities. The app allows you to book bike taxis and autos with the minimum wait time, maximum safety and is super easy on your pockets. Rapido’s journey since its inception has been nothing short of remarkable, with an exhaustive user base across bike taxis and autos of over 20 million, (pegged to touch over 40 million in 2022), 9 lakh daily order across bike and auto taxis, over 400,000 Captains, and over 25 million app downloads, and operations in over 100+ cities. Today, Rapido is playing a pivotal role in transforming commuting for Indians and has become a disruptor in the mobility landscape.

“The Rapido team has great conviction on the Programmatic medium; they were open to testing and clearly understood how data can be leveraged for the greater success of a performance-based digital marketing campaign. This trust helped the team to collaborate internally and scale the campaign with quality.”

Rajesh Fatnani, India Country Manager @

About is a mobile-first DSP operating worldwide. Using our proprietary bidder and machine-learning algorithms, we offer transparent, performance-driven, highly-targeted UA and retargeting solutions at scale with access to over 1.5 Million ad auctions per second. We are trusted by CoinDCX, Games24x7, Ubisoft, Tilting point and many others. strives to be more than just a vendor for its partners, but a partner that helps generate actual value, growth, and broad marketing insights that can be used across channels.

Thank you! Your subscription has been confirmed. You'll hear from us soon.