Customers who engage with brands on multiple levels across channels are producing a lot of data. As channels and data production increases, so does the complexity to which brands must navigate to make sense of this data and make it work for them. Couple these challenges with growing data privacy and security regulations, and brands must leap over them to reap benefits.
For brands, the ability to use ML-led-analytics has allowed them to detect the future propensity of their customers. For example, the ability to detect those customers who are more susceptible to up- and cross-selling opportunities, as well as those who would most likely become a customer. This precision and added customer insight have allowed brands to increase ROI on enhanced strategies.
Sign up for your weekly dose of what's up in emerging technology.
With such a vision, Berlin-based Zeotap is utilising AI and machine learning to provide 360-degree customer understanding and increase ROI on their marketing budgets. For this week’s feature, we caught up with Projjol Banerjea from Zeotap to gain deeper insights on how the startup is using emerging technologies to drive growth.
Zeotap is a customer intelligence platform, and its unique capabilities include a global identity resolution and unified data asset, enabling its clients to achieve better results through precision at scale and intelligent technology. The overall data asset at Zeotap spans more than 1 billion consumer profiles across North America, Latin America, Europe, and India.
How Flagship Product Is Different From Others In The Market
To this question, Banerjea answered that the stack of the company is interoperable and modular to make it customisable for varying use-cases (from customer activation to cross/up-selling to customer retention). It is well-integrated with the entire Martech ecosystem and has seamless workflows across different pre-existing systems among publishers, sell- or buy-side platforms, and other data platforms.
The flagship products of Zeotap are outlined below:
- Connect: This product allows a brand to digitise and consolidate a client’s first-party data for a unified customer view. This entails merging the offline and online data worlds, namely emails and phone numbers from CRMs and POSs with data from website/app and advertising channels.
- Enrich: This product enables brands to unlock the full power of their first-party data by combining it with high-quality third-party data so that customer trigger points can be better understood.
- Activate: Activate allows brands to increase ROI through better prospecting, churn prevention and re-targeting.
How Zeotap Uses AI & ML
Zeotap Uses AI and ML to increase the scale of certain segments through lookalike models. It also increases the quality of the data by taking multiple signals into account, removing selection bias in analytics and increasing the accuracy of estimations and relevancy of searches. The company uses machine learning techniques in all of its products. In fact, many products use a combination of these techniques.
Some of the uses of machine learning are mentioned below:
- Improving Data Quality: Zeotap uses machine learning to benchmark the quality of the data partner’s data. They use a combination of MCMC, Bayesian Inference, and Deep Learning to achieve this and have recently filed a patent regarding this combination.
- Increasing Scale: The company uses Graph-based machine learning to build lookalike models based on multiple signals.
- Correcting Selection Bias In The Database: Zeotap uses a combination of Heckman correction, satisfied sampling, and non-linear curve fitting to produce true analytics from the data concerning the country population.
- Search: The company uses search in their products, for which they use machine learning methods to understand and explore the data-catalogue taxonomy and find relationships between query terms with a set of taxonomy candidates and their dependencies.
Core Tech Stack
According to Banerjea, Zeotap’s customer intelligence platform is an end-to-end solution built in a modular fashion for maximum use-case scenarios. Technically, the data pipelines are largely based on Apache Spark, and the backend is a microservices-based architecture with the services developed overplay framework, dockerized and deployed on Kubernetes. The frontend is developed over Angular 8 framework while the charting and dashboarding uses D3.js.
Tackling Hiring Phase
To this question, Banerjea said, “Hiring top talent is a priority for us as we put our product and good customer experience at the centre of what we do. Currently, we’re looking for talent who are motivated, innovative, trustworthy, and agile to join our growing team in ten offices around the world.”
He added, “In Bangalore, we have a team of more than 60 data scientists and engineers, and we’re always on the lookout for new hires who will add to our product expansion and development.”
In the next few years, the goal of Zeotap is to become the global leader in customer intelligence. The company has recently launched a robust software layer that supplements the former product. They are now working on building out the third tier, which is an analytics layer that complements the software and data layers as well as a universal ID solution that sits vertically across all three layers of the stack.