Summary
Data signals driven event optimization' Optimizing campaigns on events created to trigger on feasible locations from the backend made us achieve growth in our feasible contribution to the total lead volume at a reduced cost per activation. Providing the required ML for feasibility to our campaigns helped us maximize results in an efficient way
Challenge
In today's scenario, where India is going digital in its roots, it becomes very important for fiber service provider companies to reach out and extend their support in different cities & localities. Airtel Xstream Fiber is one such product that thrives on expanding base in cities/localities to provide fiber connections to households. With this, comes the challenge of feasibility to the business. In a particular locality, there might be certain blocks/buildings that would be fiberized and certain would not be. To grow a business in such circumstances digitally, where the business can only grow in those fiberized localities of each market becomes a challenge.
Objective
The objective here was to implement a localized strategy with heavy indexation on feasible locations and cutting down inefficient spends in the account, thus lowering the cost per activation of the account. To understand the challenge and tackle it in a cost-efficient way required us to take both tech level and media strategy level approach. We could not go ahead with targeting the entire universe and counting every lead as business as that would lead to heavy inefficiencies in the business
Strategy
Since We could not go ahead with targeting the entire universe and counting every lead as business as that would lead to heavy inefficiencies in the business, we needed to breakdown the addresses of the consumers into buckets and measure each bucket separately to create a holistic and workable digital strategy. Broadband businesses have heavy dependencies on street addresses/Pin codes. To target only those locations where the business can provide a connection required us to break each location and classify them into buckets based on our availability. To do this, we had to break the campaigns to a city level approach. To execute this, we had to leverage the channels that we can closely monitor and can provide us the required scale. We needed to go after our Business-as-usual channels embedded with the feasibility pixels designed specially to fire on feasible locations from the backed after each lead gets dropped
Data
Hence, we picked Google search & Facebook. We ran AB testing on search & Facebook, where one campaign(A)/cohort was optimized on overall lead pixel and the other campaign(B)/cohort was optimized on feasibility pixels and the results were measured. we bifurcated the entire country into 2 variables basis the confidence we had on converting a sale and the fiber penetration in that area. Understanding the feasibility challenge and tackling it in a cost-efficient way required us to smartly take up both tech level and media strategy level approaches. The need of the hour was to break down the addresses of the consumers into 4 buckets The 4 buckets that we classify our addressed to are: Feasible High Confidence (HCF): Feasible lead with high confidence of conversion Feasible low confidence (LCF): Feasible lead with low confidence of conversion Not Feasible High Confidence (HCNF): Not Feasible with high confidence of conversion Not Feasible low confidence (LCNF): Not Feasible with low confidence of conversion At the website backend, each lead that a user drops with their address gets classified automatically in one of these four buckets according to the fiber availability. Since the business differs in terms of home passes available, fiberization, competitors and audience for each city, it was of utmost importance that we look at each city with a different parameter. Each city shows different penetration for this product
Solution
We created four events based on the above bucketing and pushed them into the Google & Facebook interface. Our Idea was to give the account the required learning in terms of feasibility to cash on it later After setting up campaigns on city level and giving each one of them required differentiated push in terms of budgets and bids, we optimized the campaigns on a combination of each feasibility bucket Hypothesis was that, since the campaigns had enough learning for each of the four events, pushing them to feasibility buckets would increase our feasibility contribution to the entire lead volume and would reduce our inefficient spends in the account amounting to a total reduction in our CPA This testing took us months at a stretch to implement as it was tough to identify which combination would work the best in terms of the city responsiveness as we didn't want to lose on any business. We took it phase wise to not affect the entire account learning. Each city went through A/B testing and Pre-post testing to come to a conclusive result
Results
After rigorous testing across our media channels with the four events, we could achieve ~20% growth in our feasible contribution to the total lead volume at almost 30% reduced cost per activation. Our campaigns are equipped with feasibility understanding and can be pushed in the required direction to maximize results
