It was about a year ago when we were trying to get customers to use our churn prediction system. We had developed a tool that could tell a business which of their customers were likely to leave their platform. We received a good response from the market for our product and we had some paying customers. But we soon realised something was missing. The churn prediction system only elaborated on the problem we were tackling i.e churn. But it wasn’t exactly helping businesses solve this problem.
Now, is a problem that very few businesses pay attention to. This isn’t because it’s irrelevant but because of lack of awareness.
We hear marketers in large businesses throw around words like ‘ growth’ and ‘increasing acquisition’ but very few focus on or realise the importance of retaining existing customers. One of the major lessons we learnt from this realisation was that our product was aiming to help solve a problem that most people didn’t know they had in the first place. It’s like how a marketing effort for a home blood sugar testing kit must have been before people realised what diabetes was and that it was a problem. So in order to make our product useful we had to create awareness about churn and how it negatively impacts businesses.
Coming back to our churn prediction system. Once we started explaining churn to our target customers, our interactions with them brought to us another major realisation. Our customers not only wanted to know which customers were going to churn out but also what they could do to stop it.
This realisation was the reason for the birth of MARS- The Marax Action Recommendation System. After months of research, brainstorming, trial, error, and experimentation, we developed a tool that used deep reinforcement learning to learn from past user data and recommend an action from a pool of marketing actions that is most suited for a given user to improve a selected business goal.
MARS at the time dealt with all kinds of mobile marketing actions — engagement notifications, value proposition for new users, discounts, cashback offers, product specific offers to name a few. We worked exclusively with mobile app based businesses and so we thought we’d found our specific niche. But we soon realised even that this domain was too broad for us be known as the ‘experts’ in this field.
While running campaigns for our customers and learning about their current business problems, we realised that most businesses today use offers (discount offers to be specific) to acquire and retain customers. However the problem they were facing wasn’t in selecting the right marketing action to send to each user. The major problem to be solved was targeting the right users and spending the right amount of their marketing budget on these users. At a broad level, customer segmentation makes this strategy possible. Sending different amounts of discounts to different segments allows for smart research backed spending, which on average reduces cost and increases the ROI of marketing campaigns with minimum effort.
Three months ago, we decided to move from the broad marketing action recommendation system we were developing and get deep into a niche. Months of experimentation and marketing campaigns had led to our system developing an expertise in a very specific marketing domain — discount and promotional marketing. We then started building a new version of MARS — a marketing tool that learns from past user behaviour and sends out tailor made offers and discounts to each customer using AI. Every offer that is sent out to a customer is worth just the right amount to get the customer to complete a transaction and this is done, within an allocated marketing budget. So you get maximum conversions at minimum cost.
Once we had built our MVP, we were lucky to have India's largest bike taxi business — Rapido sign up for a series of pilot campaigns. We went live for Rapido, while also using the observed results to refine and polish our product.
After months of hard work, failures, successes and a focused but open minded approach to solving a specific problem, we’re now heading towards an official launch for MARS on Product Hunt. We’ve built our latest, most polished version of MARS to include a complete end to end campaign creation feature, integration capability with several mobile marketing platforms, campaigns with four types of offer constructs and an analytics dashboard.