Home / Using Analytics To Improve Performance Of Mobile Campaigns
Key learnings from the session with Prof Anindya Ghose, NYU Stern School of Business
There is an enormous inefficiency between the time consumers are spending on various channels and the marketing spend on these channels. Mary Meeker’s reports have shown a year-on-year increase in the use of mobile but the increase in mobile marketing spends in lagging behind. Some marketers will tell you that mobile advertising doesn’t work. However research has shown it depends what you are trying to measure. When it comes to influence on purchase, mobile scores well above pretty much every other channel.
Traditional experiments with mobile advertising have been with geo-targeting. You use geo-location data to choose people within a certain pre-defined area of your store and serve them with a coupon. Just by doing this, there was twice the amount of conversion than email and thrice that of direct emails.
The next question then was if this is the success with push notifications, what can we achieve with pull notifications? Consumers don’t want to be interrupted on any channel, especially mobile. By getting real-time geo-location data and segmenting customers by their distance from a store, we provided customized coupons. So essentially the thinking was the further a customer is from a store, the more motivation they need and therefore the greater the discount. The coupons would have to be ‘pulled’ by customers. We found that by using discount as a function of distance there was a 3% higher online redemption than otherwise.
Okay, great. We can’t possibly stop here! What about travel patterns? Can we study the impact of travel patterns on mobile coupon redemptions? John travels a total of 7 miles every day from home-office-gym-home. On a couple of days in a week, he makes a detour, perhaps to meet friends or buy groceries. Can we use data to understand if he is more likely to redeem a coupon on his standard travel days or when he makes a deviation? For the specific client we answered this question for, we found that he was more likely to redeem it on a standard route day. The mean of users’ commuting patterns is 1.9x stronger predictor of their m-commerce activities compared to the variance.
Another interesting study is when we looked at consumers’ mobile purchase behavior on subways. Are they more likely to click on a mobile ad when they are in a crowded subway or in an empty one? In this specific instance the answer turned out to be ‘crowded’. As a matter of fact, purchase rates increased steadily till we reached 4.97 persons/ sq mt. Obviously there is a cultural factor to how many people we can pack in before purchase rates start dropping again, but it was an interesting insight into the human psyche.
All the above cases have been using static location snapshots. The real cutting-edge work that’s happening in the mobile analytics space is in using shopping trajectory data. Shopping trajectory data is sophisticated enough to tell you what consumers are going to in a mall. It can even tell you what aisle within a specific store. What we set out to do using shopping trajectory data was to understand consumers’ real-time social context, devise a mobile advertising strategy and measure the impact of trajectory based mobile advertising on shopping behaviour as well as revenues. This was an enormous, complex exercise.
For a complete transcript of this session, please email us at firstname.lastname@example.org.