By Brian Lewis
Just a decade ago, conversion rate optimization (CRO) was the playground for early adopters who demonstrated that website traffic means nothing if visitors do not convert into customers.
Today, CRO has become an important staple for all digital marketers, whose success often hinges on the quality and quantity of data they have to drive their decisions. To satisfy this need, many companies have found that adding a data scientist to their conversion team is the next frontier.
A data scientist is a hybrid between a data analyst, a statistician and a business manager. They harness big data from a variety of different sources, bring it together to produce a clear summary of what is going on in a given business and make predictions about what will happen next. They help to guide businesses by finding trends and insights that might have otherwise been overlooked, making the necessary distinctions between causation and correlation. A data scientist is curious - questioning previous conclusions and looking at information in new ways in order to get the most accurate picture possible for the company. Above all, a data scientist is action-oriented - not looking to create reports but to help direct more intelligent business decisions.
Although many marketing organizations talk about the importance of mining big data for driving both tactical and strategic decisions, most have not been able to integrate the technique into their daily operations. This is where the data scientist comes into play.
Unshackled from the constraints of pre-packaged analytics reports, data scientists can take raw data about website traffic, combine it with customer purchase data, geographic and demographic information, and even PPC and email tracking/click data, to create a profile of customers most likely to respond to certain marketing messages and promotions, at what time of day, in which season, and for what products or services. These insights can literally open up a new world of optimization opportunities for a robust conversion optimization team.
Here are some of the ways data scientists are taking conversion optimization to the next level:
Using a variety of information sources, a data scientist can identify the characteristics and behavior of a company's best customers, and create a model that allows the optimization team to streamline the path to purchase for future customers. Companies like Groupon have entire teams dedicated to mining data to create customer models for the targeting and personalization of offers. A data scientist can build a profile that includes information about what time of day the customer arrives, where they are geographically located, which device(s) they use, whether they have visited the site before, what pages they visited and other touch points they might have had with the company (for example, if they received a direct mail piece or called customer service).
A data scientist knows that a considerable amount of human behavior follows a pattern. Netflix famously used predictive analytics when it examined the viewing patterns of millions of subscribers and used the information to plan and execute its hit series "House of Cards." Predictive analytics essentially gives companies a crystal ball that reveals what types of promotions will work best on which people, which products a specific visitor is most likely to be interested in and when to send targeted emails to certain customers based on the past click and purchase behavior of similar customers. All of this, of course, needs to be tested and refined by the conversion optimization team, but the opportunities are endless with a good data scientist constantly uncovering new trends and forming new hypotheses.
With all the testing tools available today (explore Website Magazine's "Master List of A/B Testing Solutions" at wsm.co/abtools15), one might wonder what added value a data scientist could bring to the discipline of A/B testing. After all, the heavy statistical lifting is all done by the testing tool, right? Well yes, and no. A testing tool can run a test, but it can't determine what to test. There are plenty of "obvious" test opportunities that can be determined by Web analytics data and/or usability tests, but a data scientist can help uncover additional areas for testing, because he or she has access to a broader array of data and can make correlations - and hypotheses - that may be outside the normal line of vision for the rest of the CRO team. This might include optimizing for customers with higher lifetime values or other business success metrics that simply aren't evident when looking at Web analytics alone.
Check out six must-take steps toward a career in data science at wsm.co/datasci6
Armed with the right collection of information, a data scientist can develop algorithms for personalized recommendations that will lead to higher conversions. This is particularly powerful for companies like eHarmony. Its very business model relies on its ability to deliver the best possible "match" for each user. Even websites that are not in the matchmaking business understand that serving the right content to the right user at the right time is central to streamlining the path to purchase. Data scientists can offer advice about which types of content and offers are most likely to have a favorable impact with specific customer segments.
There's no question that a data scientist can expand the capabilities of a conversion optimization team. But for all the added value that they bring, data scientists still cannot replace a strong visual/UX designer, a conversion-focused copywriter or even a conversion analyst. The "science" of conversion rate optimization crosses over many disciplines - including psychology and neuroscience - and many of the most successful optimization professionals understand there is also an art to applying all these scientific insights to a specific website. The data scientist can look in the rearview mirror to report what has happened in the past, can look in the crystal ball to make predictions for the future and can develop hypotheses about potential areas of optimization. The data is almost meaningless, however, without the rest of the CRO team to turn that hypothesis into a meaningful test (or sequence of tests).
Before expanding an in-house conversion team to include a data scientist, a business should do a thorough self-evaluation of their current CRO efforts. If the conversion team is still working on the basics like form and content optimization, removing visual distractions and improving the user experience, it's probably premature to take on a data scientist. The team probably won't have the bandwidth to test and implement all the new opportunities that the data scientist uncovers.
There's no shame in not being ready or able to take the plunge into big data. Conversion optimizers were making significant improvements in website efficiency long before the advent of big data, and those techniques are still effective. What's important is that a company embraces conversion optimization as a critical business function, and be willing to evaluate the opportunity that data science offers down the road.
Brian Lewis is the director of optimization at SiteTuners, where he works with clients to diagnose conversion barriers, streamline conversion paths and support test-planning efforts.