How AWS helped Trapyz speed up customer segment creation time by 300 percent
Consumer journeys today span multiple digital and offline channels. This is why no effective marketing strategy is complete without an understanding of the consumer journey across channels for contextual and relevant personalisation.
As career marketers, Girish Vishwanath, Anil Kumar and Ranganathan Hulgundi believed in the power of digital marketing and saw it evolve as an effective channel in the last two decades. However, despite the enormous amount of data available for understanding audience behaviour, they saw a gap in marketers’ understanding of the consumer journey in the real-world.
“A large part of the data on these journeys like visits and transactions at stores are buried in CRM systems and tied to identifiers like phone numbers which cannot be verified or used to reach the consumer once they leave the channel, especially if the consumers have opted for Do Not Disturb (DND),” says Girish, Co-founder and CEO, Trapyz.
The trio decided to launch Trapyz in 2016 to focus on the real-world aspect of consumer journeys and help marketers understand their audiences based on their real-world visitation and intent.
Powering intent-based mobile personalisation
Based in Bengaluru, Trapyz is a privacy-first audience insights platform that uses Machine Learning to help marketers and brands get a better understanding of their audiences based on intent derived from real-world visitation.
“Trapyz only works with consented data to infer consumer preferences and behaviour and allows marketers to precisely segment and target audiences, improving their campaign effectiveness by more than 30 percent” says Anil, Co-founder and CBO.
The company handles huge volumes of streaming data and mines it for insights in near real-time.
“This ability to process billions of data scans with varying workloads required us to design an architecture that was not only scalable and robust, but also flexible,” says Ranga Co-founder and CTO.
The team has painstakingly designed their architecture to handle a throughput in excess of 24 billion logs in a month, which makes it one of the only audience platforms able to deliver on-demand querying and segmentation of very large processed datasets.
Working with AWS Cloud
In looking for a cloud provider who could partner with them on their scale plan and provide expert guidance on specific use cases, Amazon Web Services (AWS) turned out to fit the bill perfectly. Trapyz chose to build their solutions on AWS mainly because of its ease of use, setup, tools, completeness of stack, support and developer community, and the quality of handholding and incentives that AWS provides to young startups during their build phase.
“We considered AWS over competitors because of our engineering team’s familiarity with AWS and the credits we received from AWS which was crucial as we were testing and architecting our solution,” says Girish.
Their architecture deployment uses serverless components like AWS Lambda and Kinesis Firehose for ingesting data arriving from mobile devices. It is then organised and stored into S3 buckets from where it is picked up for processing into visitation insights by a number of automated scripts running on Elastic Compute Cloud (EC2) clusters (24 instances across clusters). The processed data is stored in AWS’s NoSQL database (DynamoDB) and Elasticsearch for aggregation and insights.
Trapyz currently processes 56,000 location scans in a minute, which adds up to more than 24 billion location scans in a month with over 92 million visits identified across 640+ cities.
Seamless deployment and continuous support
For Trapyz, the deployment on AWS was seamless.
“It was done in a phased manner while testing the platform for different workloads and throughput. We had started acquiring and onboarding some of our earliest clients even as our architecture was evolving and we are very proud to say that we haven’t had a single day of outage or interruption in operations over the last two and a half years,” says Ranga.
AWS helped Trapyz with a complete review of their architecture and provided suggestions on optimising for scale and cost-efficiency.
“AWS credits during the initial scale-up period were extremely useful in helping us test and design our platform for scale. AWS technical support teams were extremely responsive and helpful when we needed their assistance,” he says.
Customer satisfaction and scale
With AWS, Trapyz was able to onboard new clients with workloads of 80 million logs per day in a few days, resulting in faster monetisation cycles. For their clients, this flexibility in transforming and analysing data sped up customer segment creation time by 300 percent as compared to conventional processes.
“We have also been able to expand our product portfolio with minimal changes to our architecture and keep our investment in DevOps relatively low. Not only has this freed up resources, but also brought in a tremendous amount of confidence in being able to scale our operations seamlessly, when needed,” says Ranga.
The breadth of the AWS platform with services like DynamoDB, Lambda, Kinesis helped them design their architecture to handle maximum throughput while keeping costs in check with practically zero downtime and minimal maintenance and monitoring.
Additionally, the AWS Activate program has given Trapyz phenomenal opportunities, through its community events, to network, exchange stories and explore collaboration to expand their presence.
Market overview and future plans
Trapyz is currently partnered with some of the largest app publishers in the country in the media and commerce space for mining insights from their audience data. They are also working with leading brands across the food delivery, personal care, apparel and travel segments. They have been recognised by the industry and government and won several accolades including The Economic Times – Martech of the Year 2020, SBI Innovation Excellence Award 2019, GITEX Future Stars 2018 Dubai Finalist, among others.
Girish says that they are at a critical inflexion point in their business where they are looking at growing their business outside India this year. This is not only going to increase their scale exponentially but also the complexity that comes with handling multi-country datasets. “We look forward to growing our business in new regions with AWS and collaborating with them to build a platform with a truly global scale,” he says.