1 Comment


Andreas Weigend, formerly chief scientist at Amazon, has some ideas for how businesses can do more with their customers’ data. A few come from his old employer, while others stem from personal experience.

andreas weigend photo
photo: Andreas Weigend

Advances in technology tend to lump people into three categories: the indifferent, the luddites and the trendsetters, who, by virtue of their behavior and the beliefs they share with others, influence the future. When it comes to how companies use the ever-growing supplies of data on consumers, one of those trendsetters is Andreas Weigend, once the chief scientist at Amazon.com and now a lecturer at Stanford University.

At a talk alongside other data scientists in San Francisco in April, he brought up the notion of a single place where consumers could see the data companies collect. This sort of thinking suggests that Weigend is part of a group of people defining what data sharing should look like in the years to come, and how both companies and consumers will have to adapt.

Perhaps another indication is that he’s got fans. After the April talk, he took a few students and friends out to dinner at a family-style Italian restaurant. Long after the meal, some students lingered and asked him questions, as if he were an oracle or celebrity. And it is easy for people to listen to him talk for hours. He frequently makes references to foreign people, places and companies and seems to take it for granted that you are just as worldly as he is. If you engage him in conversation, you will immediately receive a vigorous response, as if he is pre-programmed to share his views, so as to have the best shot at getting others on board. This is not a man who keeps his hunches to himself. He looks the part of an idea guy, with blond curls fizzing up from his head.

Andreas Weigend. Source: Flickr user alvy.

Andreas Weigend. Source: <a href="http://www.flickr.com/photos/48600106280@N01/16048144/">Flickr user alvy</a>.

When your data is no longer your data

When it comes to being transparent with data, Weigend thinks Amazon has done a pretty good job. “One of the things we worked for at Amazon was to make it trivially easy (to show) all of the things you clicked on,” he said. The site also lets customers see what they purchased. Those are key data points for Amazon’s recommendation engine, which Weigend describes as a grid — if you view or purchase one item and then another, Amazon can line up that performance with that of other users and then to serve up items you might like.

Amazon also uses customers’ purchase history to help improve the user experience for purchases that customers attempt to make in real time. “If you buy a book which you have bought before, Amazon tells you, ‘Are you sure? You bought this item already, on December 17, 2007,’” he said. “It’s trying to help you minimize regret. It’s trying to help you make a better decision. This is how we refine the raw data, the data you created, in order to help you make a better decision.”

Other companies are not so revealing. “Some airlines don’t remind you, ‘Look! Your miles are expiring in three months,’” he said. He has nothing against airlines. It’s just that he flies a lot — he splits his time between homes in San Francisco and Shanghai and attends many conferences each year — and has plenty of examples to share in the context of flight.

An airline customer-service representative won’t permit Weigend to hear about his previous customer-service calls over the phone or see that data on the screen behind the counter at an airport, even though Weigend was the one who helped the airline create that data. And flight attendants might use fake names on name badges, even though they could conceivably access customers’ names. Weigend has a word for this sort of peculiarity: asymmetry. He is trying to fight against it.

As a consultant, he tries to change the way companies generate, analyze and share data about users and customers, among other things. That might mean advocating for data symmetry. It also might mean motivating companies to assign costs to problems such as unresolved customer calls and then figure out ways to improve the situation. His ideas stem from experiences such as ensuring that particle-physics data wasn’t being thrown off with dirt on a photo plate at CERN and incorporating external data sets to arrive at new insights while reviewing financial data for Goldman Sachs and other companies.

Voice recordings, itineraries, maps

Still, the world is not yet as Weigend feels it should be. How does Weigend live in this imperfect world so lacking in data symmetry? He leads by example, in a sense.

On his personal website, Weigend lists his flight reservations. When he does call an airline customer-service representative, as soon as he hears someone say this call may be recorded, he retorts that he will most certainly be recording the call. He carries around a voice recorder, and he has a mic that can hide underneath his shirt.

recorderRecording customer-service calls might come across as a bit awkward. But should it really push our social buttons? One day it could be common for companies to share that sort of data with customers, and then it will not seem so surprising.

Weigend also uses mobile devices in combination with Google Latitude for keeping tabs on his whereabouts, going back several years. He makes current location data public on his website and shares it with friends.

When it comes to Latitude, he knows he is an “edge case.” But his father, Johann Weigend, spent years as a political prisoner in East Germany, where the government was convinced he was an American spy. “I believe in having people know where I am. If something happens to me, somebody at least knows where I am,” said Weigend, a native of Germany, no stranger to issues of personal privacy.

When data is just wrong

In using geolocation, Weigend has become aware of a problem he calls sketchy data. He believes users should be able to correct data, because it’s not always right. At least once, Google Latitude has shown Weigend was in one place (Weehawken, N.J.) when he was actually in another (the west side of Manhattan). Google might think Weigend is in a head shop when in fact he is visiting his friend who lives above the head shop. And it’s not unusual for county officials to enter real-estate data into computer systems wrong, he said.

Some websites permit users to change their data, such as Amazon.com, on which customers can remove items from their purchase history. And it asks if something is a gift, so the system won’t use gifts to modify its algorithms on users’ actual preferences. Weigend likes those options a lot.

Where personal and business meet

How do his personal patterns overlap with his perspectives about what companies should do? It might come down to the best way to help people and companies and engender trust among all.

“Getting people to think about the amazing world of big data, that’s more about Hadoop and all that poop,” he said. “It really is about the questions that we ask. What world do we want to create? And that’s, you know, my little part in this world.”

Feature image courtesy of Flickr user aweigend.

  1. Big Data, one doesn’t have to be a machine like Hadoop, but have an interest and capacity for it, my dream is to build a self-guided photocopier using unemployed and highly talented ex-professionals, cast off by the corporate “culture” that rules the copier industry of today. But one asks, why do companies reject real talent? if so, why is it considered “bad”? and if these professionals are depressed, and have become “incapacitated” by losing meaningful work, can they be rehabilitated? The job of a startup like this alone is risky, as no known company supports this kind of venture, so one must have both passion and also be aware of what consumers need/want, what the current technology offers, the “culture” of the existing industry, if the technology to execute this aim exists, and much more…How to research? Using the Internet, (obvious) the equipment in question, observing other users, how they interact with machines, using yourself (the expert) as a benchmark, ask them questions, listen to their “whinging” – a tell tale sign that your startup concept has a market. This involves many sources, both online, offline and in real-time.


Comments have been disabled for this post