Frequent "sanity checks" with clients are critical for DataSong, a San Francisco-based online marketing and advertising company. The meetings help marketers confirm that their analytic models are steering them in the right direction and delivering on clients' expectations.
Tess Nesbitt, director of analytics at DataSong, said it's important for online marketing agencies to stay on the same page with clients, particularly when the main product is based on the findings of analytic technology. Without this close connection, it's easy for one failure to cascade into many.
The marketing industry is increasingly relying on data-driven decision-making. What once was considered primarily a creative industry is now dominated by big data and analytics, as advertisers try to pinpoint their target audience in a fractured media environment. As the industry transitions to a more fact-based approach, agencies must learn new approaches.
The main analytic work at DataSong involves helping companies attribute sales to specific online advertisements. This means tracking customer clickstream data to see which advertisements they encountered prior to their purchase and judging which was most likely to have induced them to purchase. The company uses this information to identify future advertising opportunities. All of the work is done in analytic technology from Revolution Analytics.
When DataSong starts working with a new client, the client often gives them a large amount of customer data. This may include things like transaction history, demographic data, email addresses and Web history data. Nesbitt said figuring out what to do with all the data and determining its significance is the hardest part.
Tess Nesbittdirector of analytics at DataSong
"Ideally we like to take a customer and watch them over a period of time and watch what kind of marketing they're exposed to," Nesbitt said. "We take big data and make it even bigger."
But by working closely with the client, the job becomes easier. Nesbitt said the sanity checks, in which they go over analytic findings to make sure everything makes sense given the client's subject matter expertise, are just the beginning.
As DataSong works with clients, it explains its methodology to analyze client data. Nesbitt said some clients may just want to know the broad outline of how it works, while others will ask to see the specific algorithm being used.
Managing expectations is another key piece of the relationship, Nesbitt said. Developing models to predict which advertisements and which media outlets are worth advertising with can be a trial-and-error process. Any client expecting immediate results may be disappointed to learn that the process can take some time.
Considerations are slightly different for companies that want to run marketing analytics initiatives in-house. Greg Bucko, the manager of customer insights at Southern States Cooperative, said selecting the right system can be difficult. The livestock feed and seed manufacturer and wholesaler wanted to do more to understand customer behavior for the purposes of locating new stores and optimizing advertising campaigns.
When Bucko came to the company to lead the initiative two years ago, he was most familiar with traditional analytics products from companies like SAS Institute and IBM's SPSS. But he eventually settled on a system from Alteryx Analytics, which is part of a new generation of analytics platforms that allow users to build predictive models and visualize data from the same interface without knowing any coding language.
Bucko said the analytic technology market is changing so fast that it's hard to know where to make investments.
"It's at a pace right now that I've never seen before in my career," he said.