Teemo: Getting more customers through the door with big data analysis

The rise of online shopping may have drawn shoppers away from the traditional stores, but businesses with physical locations are still looking for ways to entice customers to visit their brick-and-mortar stores, as well as their virtual ones. Marketing platform Teemo leverages big data to target, drive, and measure in-store visits in a uniquely precise way. Founded in 2014, the French platform now has offices in the United States and has worked with major retail brands including Volkswagen, Leader Price, and Intersport.

“Thanks to our use of geolocation data, companies can target shoppers and deliver ads relevant to their location history, and then accurately measure how many physical visits a particular campaign has generated,” says Guillaume Charhon, Co-founder and CTO at Teemo. “To do that, we collect large quantities of anonymized data through our app partnerships, and use algorithms to offer relevant key analyses to our customers, such as the number of visits per store, length of visit, and so on."

In order to process the large volumes of data required to feed its algorithms, Teemo looked for a platform that was straightforward to use, powerful, highly secure, and scalable. The company found that Google Cloudwas the perfect fit.

“Both retailers and customers want ads that are high-quality, well-targeted, and relevant,” says Guillaume. “It is our big data approach that enables us to answer those needs and stay ahead of competitors. We process billions of data points per day and answer to more than 50,000 advertising opportunities a second, in real time.”

Building a scalable platform

For any start-up, being able to scale from day one is key to quickly building a customer base and growing the business. Teemo needed a platform that could immediately be trusted to handle large quantities of data, in order to benefit from big data scalability from the beginning. “We wanted to start feeding data into the platform very early on, before we were even certain exactly how we would use it,” says Guillaume. “It needed to work from the launch of the very first application partnership.”

To do that, Teemo built its platform on Google Cloud, creating a data pipeline using Compute Engine with Cloud Pub/Sub and Cloud Dataflow to feed data into BigQuery and Cloud Bigtable. “Implementing the infrastructure was very straightforward, it took only a few months to develop,” says Guillaume. “It’s thanks to BigQuery that we are able to match the advertisement to a particular person. We index data by day in BigQuery, and by geographical location in Cloud Bigtable. That means we can locate relevant data much more quickly. We then run our smart algorithms using on-demand Cloud Dataflow clusters based on this data to prove a customer has visited a store.”

“With Google Cloud, managed services like Cloud Dataflow and Cloud Dataproc really take the burden away from the operations team,” says Mikael Gibert, DevOps Lead at Teemo. “There’s very little maintenance from our side so we can focus on optimizing the platform and the accuracy of the information we provide for our clients.”

Delivering data insights

With the analytics offered through BigQuery, Teemo’s clients are able to optimize their advertising campaigns and gain valuable actionable data insights. “For example, we carried out a project for the French supermarket chain Carrefour,” explains Guillaume. “Our data is gathered using software integrated with our mobile app partners, and it is completely anonymized and GDPR-compliant. Using this anonymized geolocation data, we were able to track which clients frequented only Carrefour, and which frequented numerous supermarket brands. It’s those kinds of insights that enable our clients to improve their targeting of particular customers.”

“Another customer, the French DIY chain BricoDepot, reported that every €1 spent with Teemo was worth €15 in revenue gained, thanks to the increased traffic in the shop,” adds Guillaume.

Staying flexible

Thanks to Google Cloud, Teemo’s platform is able to process data in close to real time. “Maintaining low latencies is really important for mobile advertising. Even with 50,000 requests per second, we are able to process the requests we receive in under 40 milliseconds,” says Guillaume. “We are using four different Google data centers, one in Europe and three in the United States, so we’re close to our clients. And with Cloud CDN, we can cache banners as close to the ad viewer as possible. We’re really happy with the performance of the platform.”

The flexibility offered by Google Cloud is another key advantage. “Because of the flexible pricing structure, we can be much more agile,” says Mikael. “If we want to switch from Compute Engine to Cloud Functions from one day to the next, we can do so. We don’t have to wait for an annual subscription to end.”

Now, Teemo wants to add even more managed elements to its infrastructure, using Cloud Functions to execute code automatically, App Engine flexible environment to automatically scale up and down, and Cloud Composer to orchestrate workflows. “We also want to start using Cloud Memorystore, which has only just become available, for our Redis database,” says Mikael. “All of this should reduce the operations burden even further, so we can concentrate on code, not infrastructure.”

“We are currently focusing on the United States, and then we want to take the platform global,” says Guillaume. “With Google Cloud, we know we can scale up without any constraints, so we can continue to take in even more data and deliver even more store visits for our customers.”