Rebuild for the future
Retailers don’t necessarily need to hire more data scientists, but they would be wise to look to the ad tech space, said company president Dave Simon. vibenomics and an in-store marketplace.
“Very few people in this industry understand how these systems communicate with each other.…If you want to take market share and create products in a cost-effective way, [and] Let the entire ecosystem work better together to take ad tech talent out of the CTV business. ” he said.
“CTV is a very crowded space, but all of these people know how to take one data set keyed off from location IDs and combine it with a data set keyed off from cookie IDs. There are a lot of professionals out there. The real key is not generating insights, but knowing what questions to ask.”
Epsilon’s campaign echoed this, noting that the people who bring the greatest added value are those who wear both the business and technical hats to ensure data scientists have the information they need for media purposes.
“We need business translators who not only understand the technical world, but also understand the business and can help each other with translations,” Campaign said.
There are good reasons to take on the heavy lifting required for a modern data strategy. Looking to the future of in-store retail media, Simon sees great potential in machine learning. As someone who has spent years working in mobile app advertising, where deep learning models have revolutionized targeting, he’s bullish on the potential to reimagine in-store retail media.
[Related: BP’s Derek Gaskins will talk about their new retail media strategy at P2PI Live in November]
But even though retailers store vast amounts of shopper data, their protectionist strategies make them more like broadcasters than app developers.
“When a consumer walks into a store, there’s so much value and intelligence that there’s so much data being put on the back burner. Think about where the store is located, where the aisles are located, how much time and money goes into the design of the store. All of this creates incremental sales value that can be tied to the individual user.”
Unfortunately, the data is not currently being used in sophisticated ways to feed machine learning models to better predict outcomes, he said. When in-store ads are treated like digital impressions and combined with transaction log data, machine learning reveals correlations that highlight opportunities to influence shopper behavior.
“This shows that you not only understand your business, but you understand the mind of the consumer who comes into your store,” Simon says.
Progress once again depends on a mature data strategy. Unlike the mobile app space, where transactions and data collection are easy, retail data sources are incredibly fragmented.
“It’s not easy to piece it together,” Simon stressed. “On the retailer side, they’re being asked to do something for their brand and answer questions that they might have been asked once a year or quarter, but now they’re being asked to answer those questions 10,000 times a second.”
While this optimization is part of many retail media strategies, building the complete infrastructure to do it at scale remains a large, top-down, long-term investment for both marketing and IT.
This article first appeared on our sister brand P2PI’s site.
