The Good, Bad, and Ugly of Privacy and Tracking: A Business Perspective

April 27, 2012           https://www.x.com/devzone/articles/good-bad-and-ugly-privacy-and-tracking-business-perspective
by Brenna Roth
Much like the antagonist from a hard-boiled detective novel, our every move is being tracked—at least with regard to our online activity on ecommerce sites. It's not by Big Brother, but by Big Business, which is watching, recording, and analyzing our every move, click, and product view. Many site visitors are blissfully unaware of the pervasive tracking. For the minority of savvy users who are aware, most feel "creeped out" or violated. Though there's a compelling reason why ecommerce sites want to track and collect all that user data, there's also some caveats that ecommerce companies (and websites in general) should be aware of. We'll take a look at these issues—the good, the bad, and the ugly of privacy and user tracking from a business perspective.

The Good

Our initial look will focus on the good uses of visitor tracking. We'll examine some of the interesting techniques that ecommerce companies are using to create a compelling experience for their users based on tracking data from previous interactions. These will include some machine-learning techniques as well as hands-on techniques by marketing professionals.
The amount of data that ecommerce sites (and most websites in general) have access to is almost incomprehensibly large. It's too much for any individual (or even any reasonably sized group of people) to look through. Instead of using generic heuristics to have people sort and organize data, many companies are relying instead on computers and machine learning. Computers can sort and organize data in a way that helps add business value. Let's take a look at a couple ways that ecommerce companies are using this now.

Recommendations

The most well-known version of a recommendation engine is run by Netflix, which suggests movies to customers based on their past preferences. At a high level, the company is pulling in lots of preference information from all its customers and then using that data to find similar items to recommend to similar customers. This recommendation system offers customers a compelling reason to keep coming back by helping them discover new movies they might like. The power comes from leveraging lots of preference data from lots of customers. Intelligently sorting through that data allows Netflix to recommend movies that customers actually want to see.
Recommendation engines don't need to be limited to content sites like Netflix, however. In fact, ecommerce sites can use such a system to recommend items to customers based on past shopping data. Knowing what an individual has purchased in the past is a strong indicator of what she'll want in the future. Companies that can leverage that data to help customers discover new items they would like but wouldn't otherwise know about can create a win-win situation where sales increase and customers have a unique experience tailored to what they like.
There are several near "out-of-the-box" solutions for creating a recommendation engine. They are no longer limited to the domain of "Big Data" experts, but can now be easily used by a range of different businesses. The most popular recommendation engines run on Hadoop, with Mahout being a popular implementation. Setup isn't trivial, but it's quickly becoming a practical option for many companies.
The business case here is simple. Customers are leaving a trail of data when they make purchases. That trail can add value to future visits by the customer as well as add to the company's bottom line. Knowing what the customer has done in the past, and using that data to predict what he'll want in the future, allows the business to provide a uniquely tailored experience instead of a generic one-size-fits-all shopping experience.

Know Thy Customer

The value of tracking user behavior goes beyond just predicting what users will want and recommending that to them on a future visit. Seeing how customers interact and shop can also provide valuable information for marketing purposes. Companies that don't track who their customers are don't know who their customers are, and can't reap the benefits of informed marketing strategies.
Ecommerce sites don't just stop at tracking who their customers are—the successful ones go a step beyond and track how their customers are shopping. They can do this through a battery of A/B split tests as well as more complex multivariate testing. The idea here is that the shopping experience is constantly being tested and retested, with continuous experiments being run to see what layouts best enhance a customer's shopping experience. Generally, this is done on an aggregate level, but knowing who the customers are and how they are shopping plays a big role in creating worthwhile experiments.
This business case here is simple as well. Knowing who your customers are and how they shop on your site plays a big role in improving customer acquisition (getting more people to the site) and customer conversion/retention (getting them to buy something and keep coming back for more, respectively). The successful companies go beyond just tracking the trail of data and create situations to measure new data from user experiments.

The Bad

Collecting customer data isn't always sunshine and roses. In fact, in some cases, the bad parts can outweigh the good parts. We'll take a look at some of the notable pitfalls of collecting customer data. In many cases, these can be mitigated through proper planning, but they shouldn't be ignored.
There are a number of companies working to reduce the barrier to entry for large data processing, but there's not a single one-size-fits-all solution for all companies. Significant work is required to set up a tracking system. Domain expertise is required in a number of areas to get a system up and running. This domain expertise is not cheap, and generally is difficult to find. This is one of the biggest reasons why companies aren't tracking customer shopping behavior on ecommerce sites.
The tracking difficulty isn't just limited to getting a system up and running. A company also needs to know what to track. In fact, figuring out what to track may be significantly more difficult than setting up the tracking system. Understanding what to track and why requires a unique skill set that's hard to find. Each site has different key metrics, and experience with one site won't necessarily translate to another site.
The big problem here is that there isn't an easy way to get up and running with collecting and processing the data you need. It can be an expensive and time-consuming ordeal requiring a number of experts. It also requires constant attention and continuous improvement. This can be prohibitively expensive to smaller ecommerce sites just starting out.

The Ugly

There's a lot of potential value to be gained by tracking customer behavior online (along with tracking other customer data). The "bad" parts can, for the most part, be mitigated through proper planning and an adequate budget. There are, however, some ugly parts that may not seem obvious at first, but when they crop up, they can spell disaster for an ecommerce company. These are existential risks for a company, and must be properly accounted for by any company handling customer data.
A number of prominent hacker groups have been in the news lately, due in large part to their liberation of customer data. This data can be a high-value target for a number of reasons, and there's a very real chance that an outside group could steal it. The consequences of that obviously depend on what data has been collected and how it's protected, but it's a salient threat nonetheless. The more data that is collected, the bigger the threat of theft, and consequently the increased need for additional security measures.
Hackers aren't the only big threat that companies handling customer data face. Customers are becoming increasingly wary of infringements to their privacy, and feel "creeped out" by tracking measures of any kind. Many customers relish the feeling of anonymity they get from shopping online. These customers are especially prone to feeling violated when tracking measures become too obvious. A balance is necessary between offering no custom recommendations and a completely tailored experience. Overleveraging customer data can turn a lot of customers off and keep them from returning in the future. It's not just individuals, either, who are becoming more aware of privacy and tracking online; there are several large privacy rights groups that are raising awareness as well.

Conclusion

Collecting customer data can provide powerful information that can be used as a win-win for enhancing both the customer's experience and the business's bottom line, but with great power comes great responsibility. There are a number of potential pitfalls a company can face when tracking and collecting customer data. In many ways, the wealth of information that can be collected is unprecedented, so some growing pains are to be expected.