Big data can help brands tailor more effective offers to consumers, but how?

Imagine when you walk in a shopping mall, a mobile advertisement pops up on your phone, giving you a coupon on exactly what you planned to buy.

Imagine speaking to your friend about an interesting ad you saw on Facebook then discovering, to your surprise, that your friend is also interested in buying that exact product—that’s the beauty of well-designed marketing, thanks to big data.

Anindya Ghose, author of Tap, Forces Shaping the Mobile Economy, is a Professor of Technology and Professor of Marketing at New York University’s Leonard N. Stern School of Business. In this interview with CKGSB Knowledge, he analyzes what consumers do with their smartphones and how businesses can tailor effective offers that occur at the optimal time, while also ensuring that information exchange is a healthy two-way street.

Q. Among the aspects that drive purchases, such as crowdedness, location, social dynamics and other forces, which are decisive?

A. They are all about equally important, it’s hard to categorize one as being more or less important. What we are seeing is that while each of the forces are important on their own, what also happens is there is a combination affect, or an integration affect if multiple forces can be designed in an given context, the effectiveness increases dramatically. For example, if you take weather as a force, and location as a force, you can combine the weather with location data and increase the effectiveness of your mobile marketing. If it’s sunny and humid you will give a certain set of offers; if it’s rainy and foggy you will show them different kinds of offers. Consumer psychology plays a big role in our studies, and even in the actual execution of these projects, we’ve been very careful in using consumer psychology to incentivize people to the right incentives, to motivate them to act accordingly.

Q. In an NYU speech you mentioned people tend to buy a product the day they got the mobile ad, and some buy it later. What’s the difference between the two different decisions? Does the one promotion you buy right away always have to do with discount or coupon?

A. There are two kinds of behavior that people exhibit with coupons, the first is an immediate purchase based on the received offer coupon, this is often because it’s an impulse purchase, or it’s because of the fact that you are able to carefully predict people’s preferences and give them the coupon at the right time, at the right place and the right location; the other kind is people who show a delayed effect to redeeming coupons, and sometimes the effect can be as long as two weeks, in this case it’s not because people are not interested in an impulse purchase or that we didn’t predict their preferences, it could be a variety of psychological reasons that prompt people to keep searching, and only after they exhausted the search for all the prices, they realize that the offer they have is the best.

Q. You also talked about how different people get different discounts, can you elaborate?

A. There are many ways to customize coupons and offers on smartphones, one is based on location, for example if a customer is a hundred meters away from a store, another customer is 300 meters away from a store, then the customer who is further away will actually get a higher discount, and those who are closer to a store will get a lower discount, because if you are closer to a store then your cost of traveling to a store is less, so you have a higher incentive to come; if you are further away, you need bigger incentive to come to the store, so we can customize coupons accordingly.

Q. Apart from location, what are other factors?

A. We can do it based on demographics, are you a man or a woman, what is your age, income bracket, we can do it based on weather patterns, are you a fast walker or slow walker, are you shopping with a spouse or a partner or shopping with friends.

Q. Would you offer higher discounts to women, for example?

A. Not necessarily, in some cases yes, it makes sense to give a higher discount to women, but it’s a function of their past historical transactions, it’s a function of who are they shopping with. One example we have seen, people tend to spend more money when they are shopping with their friends as opposed to when they are shopping with family, and the real psychological reason is that your family is well aware of your net worth and your inc ome, so you don’t have to show much you are able to afford. But with friends, they don’t know your exact net worth or income, so there is a tendency for shoppers to spend a little more when they are with their friends.

Q. What are some unique facts about Chinese customers you have found from big data?

A. One of the things we found is that mobile coupons actually have the highest redemptions from the highest income group. So when we looked at the shopping malls in China in many different cities, what we see is when we send coupons to shopping mall customers, and we have their prior demographic and purchase data so we know the income levels, we saw that the highest income levels are the ones who have the highest propensity to redeem coupons. This is very interesting because mobile marketing has often received some flak that it’s not very profitable because it only attracts the lower income customers, but what we are seeing is that the highest income customers have the highest redemption probability.

The second thing is we have seen age is positively correlated with redemption, so it’s not that only young people are redeeming coupons, even the older people are redeeming coupons sometimes much more than younger people. So studying the Chinese population in this country across multiple cities has given us a lot of insights.

Q. The China market and Chinese customers change fast, can big data help business to adapt to changes?

A. It can. In fact we have done studies with Alibaba and Taobao using the purchasing data of customers to make useful predictions. In fact if you deploy appropriate statistical and data mining algorithms and run some carefully designed field experiments, it’s very much possible to predict the next big wave that’s coming in customer preference information and we can fine-tune and tailor our offers based on those predictions.

Q. Are there any differences between the users and data collection processes in the US and Asia?

A. There are regulations that are different, laws that are different about how companies can reach you to give offers. For example in many countries, brands can send you a text message with an ad or with an offer, but in other countries they can only show you an ad within an app, so these are some of these differences. And I believe consumers are fundamentally similar everywhere, I’m seeing more and more evidence that people are willing to give their data in exchange for a relevant offer, because they understand the simple economics of this transaction.

Q. We often see ads on social media, how effective are they, and how do you measure the effectiveness?

A. That’s a tough problem. We have something called ‘attribution’, and there’s a concept in digital marketing that customers can get exposed to multiple touch points, multiple ads in the path to purchase, and it’s incredibly difficult to isolate the effect of any given touch point on the final sale, but at the same time it’s also incredible important for us to do so, because when you know how much more or less influential a given ad was, you can allocate your budget accordingly. So within the world of social media, we have similar opportunity and challenges, unless you have a carefully designed field experiment or you have a natural experiment, it is not easy to be able to trace back the effectiveness of social media ad on sales. So we are working on statistical and economical models to help companies uncover the effectiveness of social media ads.

Q. For new customers, I’m guessing a certain brand’s ad probably has to show up a couple of times before the potential customers actually take actions to buy it. How many times does it take?

A. This is also related to digital attribution. It’s not just a number of unique touch points that a person faces in the path of purchase, it is also a question of how many exposures to the ad is optimal, because too few exposures may not trigger any interest awareness or desire in the customer, too many exposures can lead to a lot of annoyance and confusion. My co-author Vilma Todri and I have been working in attribution for a number of years, and we now have models in place to be able to extract that optimal number of ad exposures, and depending on the context or the state of the customer, it could be between four and five ad exposures per customer, which turns out to be an optimal number of ad exposures.

Q. Wechat reaches half a million people in China and overseas, it has company ads on its ‘moments’ feed, and the way they ‘customize’ ads is very interesting: if your friend interacted with that ad, there is a 95% chance that ad will show up on your Wechat feed, and if they didn’t click it, the chance drops to 20%. What are your thoughts on this?

A. On Facebook we see a similar phenomenon, this is referred to as socialization of advertising. And the core-theory is that my friend and I might have a similar innate preference for the product, so if he or she likes a certain brand, the chances of me liking the same brand are higher. This has now been quite widely used in different countries and different platforms, and science has shown that social ads tend to be more effective than just regular ads.

Q. You collaborated with Shanghai public transportation on a digital marketing program, how did that go?

A. One of the nine forces is “crowdedness”, meaning that the effectiveness of advertising on a smartphone is a function of how crowded your immediate context is. If you are situated with lot of strangers, you tend to focus more on the phone and that’s a good time for the brand to send you an offer. We worked with the Shanghai subway, using variations of crowdedness on subway trains, to design an experiment and see how customers that had been sent coupons react in different immediate contexts. What we saw is that as the level of crowdedness increases from one person per square meter to five people per square meter, the purchase rates on mobile phones keep going up, and sometimes in a non-linear way. So crowdedness definitely plays a big role, and that’s why I have an entire force dedicated to that in my book.

Q. Big data can help tailor some good offers, but data collection can be used for illegal things as well, such as identity theft and other forms of fraud. What can be done to avoid unintended data collection?

A. This is a very important point and I made an argument in my book that an eco-system is emerging where customers are willing to share their data as a currency to get a more relevant and targeted offers from brands. This is a two way street, it creates a lot of intimacy, but both parties have to act responsibly. Brands can get access to our information and it is their responsibility to design an offer or an advertisement that is perfectly relevant and well-targeted for me, if they do not do accordingly, then they will lose the customer. As customers see more and more relevant offers being sent to their mobile devices, they are more and more willing to exchange their data as a currency. I believe that’s a healthy eco-system that’s emerging with give-and-take between consumers and brands. Consumers have to be careful about how they share their data and where they share it, because they have to be cognizant about potential risk, and the flip side is companies have to be careful about how they use our data, but I believe that this eco-system will become more and more mature in the coming years and mobile will play an important role.

Q. What’s the bigger picture of big data in future?

A. I think one of the most fascinating technologies is going to be the Internet of things (IOT), an eco-system where multiple devices are stitched together electronically. This IOT system such as a smart home or smart car is going to create even more data, it’s going to tell companies more and more about consumer behavior. So I personally believe that the biggest source of data in the coming years is going to be from IOT as a gigantic source of data.

Q. In future digital devices might become wearables and chips, how will this transformation impact data collection?

A. Together with my co-author Professor Beibei Li in Carnegie Mellon University and Professor Xitong Guo in Harbin Institute of Technology, we are doing a project in China and we are essentially looking at how the adoption of wearable devices by consumers is going to make them change their lifestyle, we can track the effect of lifestyle changes before and after they adopt the device. We collaborated with many hospitals in China and from there we can have the information about the patients before and after the adoption of devices and that lets us evaluate how consumers and patients are changing their personal health and wellness behavior. I think it’s going to be a very exciting project, because if we find any evidence that wearable devices do affect healthcare and wellness, it’s a big finding.