Skipline Tech
UI/UX Design
Data Engg & Analytics
Mobile App
A Geo-fence based Ad-Tech platform reached out to synctactic. They provide restaurants and retail locations in cities a platform to directly market to customers based on the customer’s location.
Industry
AD-Tech
Company size
<100+ Employee
What we did
- Data Engg & Analytics
- Website
- Mobile App Development
Objective
A Geo-fence based Ad-Tech platform reached out to synctactic. They provide restaurants and retail locations in cities a platform to directly market to customers based on the customer’s location. The app collects a lot of data on the customer like geofences entered, tables booked in restaurants, shopping coupons redeemed etc. The company wants to use the data to enable smart features like
Provide a live score on a restaurant based on how many people are at the restaurant at a given point of time.
Multiple Demand Forecasting models would need to be trained and evaluated
Create segments based on customer behaviour
Challenges
The team worked with a 3rdm party app developer to build the app. The app is configured to collect a lot of data and is stored on MySQL. The live data sources such as geo-location pings are sent to a webhook that the platform subscribes to. The team required to build machine learning models and deploy them on the apps. This came with huge cost and time implications as it was difficult to hire data scientists and the app developers lacked expertise in ML/AI. The team wanted a solution that can be easily used by them without having to build an entire data science team and data product from scratch.
High Cost to Build
Solution
Live scoring model
The team created a project and added the MySQL DB, webhooks and social media streams as sources of data. The pipe was configured to combine the MySQL data with the live location data of the users. This processed data was then fed to a math function which calculated a score based on how many ppl were at the restaurant, events at the restaurant, traffic data near the restaurant and social media mentions. This score was then pushed to a REST API client on the platform which exposes an API for the live score. The app used this API to display the live score to its user.
Customer Segmentation
The customer behaviour data from the MySQL DB was used to build the features for the clustering algorithm. The features were calculated using the code engine to add the derived columns to the dataset. This data is fed to the clustering algorithm which is configured to generate customer segments based on their behaviour. Segments like Big Spenders, Window Shoppers, Budget eaters, Party Goers etc. were created using the model. When new users installed the app their data was also fed into the model which bucketed them into specific segments with high accuracy.
Recommendation Engine
Users who used the app when walking in a shopping district or near restaurants were recommended with offers at nearby restaurants or shops. The data required to build recommendations were the live location feed from the webhook, the segmentation data from the model in another project and also the customer behaviour data. A recommendation algorithm was integrated using the code engine. The output of the model was fed to the REST client which exposed the APIs for the recommendations. The engine was able to target the right segments with the right offers hence improving the redemption rate and also revenue to the business.
Outcome
With Synctactic AI, the company was able to bring in a data science based approach to look after their operations and hence improve customer satisfaction. They could optimise routes and stick to SLAs and Reduce operational costs by equipping drivers with feedback systems. Synctactic AI provided them a way to leverage a scalable and flexible data infrastructure for all their data science initiatives.
Improved team productivity
Recent Work
Launch your product with Arokee