AiQx Universal Search
“I want to know about audience interested in BMW and are in London.”
“What kind of people watch HBO and are in New York and also visit eBay?”
“Who owns both a phone and a tablet and is interested in swimming?”
“Know more about CXOs or executives who read about SaaS.”
To answer these questions AiQx Universal Search was developed. As the name suggests, Universal Search enables users to search for a number of things when looking for a combination of demographics and products, or industries, categories, keywords, and brands.
Examples could be:
• cars24(WEBSITE) and age-30–45(DEMOGRAPHIC)
• cars24(WEBSITE) and (BMW(KEYWORD) or Audi(KEYWORD))
• magazineusa(WEBSITE) and reisefrage.net(WEBSITE) or (Fashion(KEYWORD) and shop(KEYWORD))
With the addition of data from multiple sources, users can also get information on first-party data, audience segments, and keywords. Combinations of various data points generate richer and more actionable insights, simplifying the entire campaign optimization process.
The tech behind AiQx Universal Search:
AiQx Universal Search is a web-based application that allows users to search for more than 4 million entities across multiple locations.
The crux of AiQx Universal Search is the data. Insights are generated from all processed cookie profile documents of ~200GB within a period of seven days. Approximately 40 TB of data from various sources is crunched and aggregated using Hadoop to make documents that represent the profile of each cookie.
For every search/query, the ~200 GB of documents are scanned to filter those user profile docs that satisfy the query. The scanning and filtering of documents is powered by Amazon Athena. These documents are then analyzed to generate insights to give a holistic picture of individuals who belong to the searched entity or are interested in it. Insights can also be converted and used as targeting strategies to deliver performance on digital advertising campaigns. The entire process of generation of insights is done within 10 secs despite the huge processing that takes place in the background.
The architecture is designed to be scalable and reusable. The backend is based on async non-blocking event-based architecture, built using a reactive toolkit (Vertx) on JVM.
To enable users to look for all these entities, the typeahead search/autocomplete service is powered by ELK(Elasticsearch, Logstash, Kibana) framework.
HOW IS IT HELPING CLIENTS?
THE CHALLENGE:
A digital marketing agency based out of the US, wanted to better understand our analytical capabilities. We saw this as a fantastic opportunity to open the doors to AiQx Universal Search and share with them the exciting insights they would have access to. What they saw ticked almost every box for the senior management of the brand.
Additionally, they were also interested in our ability to ingest their own data and build an ‘Affinity Dashboard’ using their CRM data. Although this was a first for MiQ, we saw this as an opportunity, which, if done well, could be replicated with other agencies in the future.
THE SOLUTION:
The solution came to life at Cannes Lions in June 2017, when MiQ joined forces with the client to pitch to a major CPG brand with a global audience using their customized version of AiQx Universal Search. Their aim was to show how best to reach brand audiences. This pitch was a great success and the team is now using AiQx Universal Search at all levels — from the senior leadership team to junior planners.
THE RESULTS:
Universal Search has opened conversations with a number of clients. The clients’ planning and strategy teams are able to see the benefits that the tool can bring to the table, further strengthening our relationship with the agencies.
A $15–25M fixed margin managed service agreement has been initiated with some clients, consolidating the majority of their managed media spend with MiQ. A fantastic example of how AiQx Universal Search has positively impacted agency relationships!
Authors: Manikanta Prasanth Kuriseti, Sreejit Choudhury, Anil Kumar Javvadula
