Sramana Mitra: I’m particularly interested in understanding the predictive part of it. Let’s understand how you predict stuff based on a use case.
Stephanie Newby: The best way to think of that is to start tracking things and look for correlation between things. Then test them out. Start with some hypothesis, test that out and adjust them based on other data that you’re getting. I find that this is more valuable to have than just one source of data.
One source of data might be predicting something but you’re not sure for how long it will apply. If you bring in other data, it enables you to refine that forecasting capability more accurately because now you can start to see the cause and effect more easily. If I’m tracking an advertising campaign for a new product launch, a new cereal, for example. I’ll be tracking the social media conversation. It could be just volume. It could also be sentiment or finding out what’s positive in the conversation versus what’s negative in the conversation.
It might be more specific. It might be intent to purchase, which you can track as well. If you start to see that trend alongside your point of sale data, then you can start to see patterns emerge. What we’ve been doing is experimenting in bringing in different datasets. We’ve got a relationship with a Silicon Valley data company that has lots of different types of structured data that they enable their customer to view on a platform with a dashboard that they’ve created. What we can do is bring social media alongside customers of these data sources. When you see those alongside each other, you become better at forecasting.
Sramana Mitra: This is interesting from an advertising use case. Do you have a use case that you would like to discuss from product marketing? You said there were usage and use cases in product marketing as well. Is there one that you can think of that would be interesting?
Stephanie Newby: Abercrombie & Fitch, for example, use us for merchandise and looking at what would be the best placement of different articles in the store as well as the timing of putting those articles out in big events like Black Friday. That’s based on not just the uplift that they see, but also the buzz that’s created on social media.
Sramana Mitra: Are we talking about people putting those clothes on and putting pictures on Instagram kind of buzz or are we talking about people tweeting. I guess what I’m trying to understand is if we’re talking about the use case of a clothing retailer and how your technology would help in merchandising, how does the consumer in social media today give a system like yours signals?
Stephanie Newby: It’s based on them either taking photographs trying things on or tweeting about a article that they’re just bought. That can inform the merchandiser because based on volume, they will see that the placement of that article in the store enabled them to get a lift in the social conversation.