Digital Communication, Unstructured Data, and Customer Outcomes
When we shop or go online, we leave behind a huge number of digital traces. These include which ads we walk past, what we write in online reviews and which livestreams we watch. Most of this data is ‘messy’: text, video and activity logs rather than neat spreadsheets. As a result, companies rarely use it to make better decisions.
My dissertation asks how this data can answer three concrete questions: first, is advertising on digital screens appearing in supermarket aisles actually make people buy the advertised products? Second, why do some shopping malls thrive while similar malls close, and can online reviews tell us why? Third, how can beginning gamers who livestream on Twitch build an audience when thousands of others play exactly the same games?
The motivation is that retailers, mall managers and creators make costly decisions that are often based on gut feeling or slow, expensive surveys.
Important insights
Seeing an ad on a supermarket screen makes shoppers about 8% more likely to buy that product. The screens work best for hedonic products, new products and cheaper items. They also work better on weekends, later in the day, in sunny weather, in busy stores, and with emotional rather than informational ads. Shoppers don't stockpile because of them. Instead, they also buy more from the same brand and product category. This offers very important insights into how effective this tool is, for what products, when and with what content.
For shopping malls, what matters most to visitors is what the mall itself controls: its atmosphere and its mix of tenants. Prices and the offer inside individual stores matter less. Malls with less satisfied visitors are in the end also more likely to close. This offers an important insight in how shopping malls could be saved and made future proof.
For streamers, sticking to similar types of game portfolio attracts the most viewers, and chasing the most popular games doesn't help creators to grow against popular belief. However, some tactics that boost viewers quickly also make streamers more likely to quit later.
More information about the dissertation