Wednesday, November 16, 2016

How Marketing Analytics Is Helping Atlanta Brands Compete

On Wednesday, November 9 the Atlanta Tech Village hosted Navin Sharma, VP Brand Insights & Analytics at Arby’s, and Elizabeth Colyer, VP Strategic Planning & Insights at Sharecare, Inc.

I, and around 40 others, arrived ready to get insight on how to use marketing analytics to inform business decisions. While I initially thought this would be a panel discussion, with a moderator asking questions, the time was actually divided between the two professionals, with each giving a presentation on what they believed would be the most helpful during in the short time they had.

Navin began by giving a brief overview of his position as VP of Brand Insights and Analytics. One of his first pieces of advice centered around testing.

“It’s important to empirically test everything you do,” said Navin.

AIMA Nov 9 .jpg

Navin then said that Arby’s now uses analytics to inform business decisions almost daily. He then expanded into the three rules that he and his team use to ensure that every marketing initiative shows value:

1.     Tie every metric to sales/profit

2.     If the metric can’t be tied to sales/profit, come as close as possible

3.     If it’s impossible, deprioritize it

By ensuring that metrics can be tied to sales, you are better able to analyze the impact of marketing programs including media, pricing, LTO Program, competitive messaging, brand strength.

For example, Arby’s now knows the sales/profit impact of every different type of coupon issued. They used to measure the success of a coupon simply by redemption rate but are now digging deeper and looking at other items that are purchased when the coupon is redeemed such as total order value as well as how the overall success of the restaurant is impacted.

On the back of every receipt is a customer survey. By combining and analyzing the results of these surveys over time, Navin’s team has identified the attributes of a restaurant that are considered “must-haves,” “implicit expectations,” and “potential delighters.” New restaurants that score well on certain factors have a much higher likelihood of succeeding.

Once Navin was done speaking, the moderator transitioned to Elizabeth, who began her talk by discussing her overarching goal to maximize the convergence between customer engagement trends and known goals.

“We constantly make sure we are thinking about how company goals and user goals overlap,” said Elizabeth.

She elaborated by saying that the value provided to the customer and the value provided to the organization must overlap. She followed up with a crude but relevant example posed in the form of a question: Should users who viewed websites discussing the symptoms of the most popular sexually transmitted diseases be retargeted with ads from those sites? This could be an example where the value to the user would not overlap with the value of the organization given that STDs tend to be a fairly sensitive topic.

Elizabeth then dove into 3 different stages that her company, Sharecare, has gone through as an organization. These included Foundation, Automation & Testing and Advanced Sciences & Predictive.

Foundation is one of the most critical pillars to success. This includes team- making sure you have the right people with the right skill sets who you can count on consistently for implementation. She said don’t be afraid to hire technical. Having team members who can feed the development team existing code helps processes to run more efficiently.

She then transitioned to Automation & Testing. Ongoing testing is critical. She cited one example where they had a call to action on one of their main conversion paths to make an appointment. They tested changing this to a phone number and immediately increased clicks to the button by 22%.

Finally, she discussed Advanced Sciences and Predictive Analytics. This step is simply looking at current data to make predictions about the future. How you accomplish this depends on the type of data sets available. With point in time data sets you’re focusing on cohort analysis. In other words, how these segments are currently interacting on a website. Longitudinal data sets are long term and allow you to look at more valid statistical methods of predicting future outcomes. Those data sets, that measure past user behavior, can then be used to predict the likelihood of future engagement on particular platforms.

Blog by John Rhinehart, Search Marketing Analyst at Relevance Advisors