LONDON: Machine learning is becoming increasingly important to sift through huge volumes of data, but humans will still be required to answer the nuanced questions that arise from the inevitable gaps in the data, according to a leading planner.
Writing in the current issue of Admap, Ruth Zohrer, Head of connections planning and marketing technology at Mindshare UK, cautions that digital data is no “silver bullet” to understanding and defining audiences.
“Just like surveys, [it] is equally prone to delivering flawed insight and therefore requires questioning, interpretation and validation before being taken as dogma,” she writes.
She outlines three critical steps in this process, starting with the need to define data inputs.
Planners now have a multiplicity of data sets to work and early prioritisation of pertinent data sources is essential to avoid getting lost in a sea of irrelevant data points.
“Most briefs today have more than one starting point to reaching a solution, yet some will be more effective than others,” Zohrer observes.
“Because of this, the order in which different data sets are prioritised and applied during planning stages is paramount to delivering desired outcomes efficiently.”
Having chosen the data inputs, it is then necessary to combine them, supplementing survey or CRM data, for example, with digital data from tracked actions online to understand people as holistically as possible.
“Layering data when defining audiences makes it possible to redefine segments of people across the ‘who, what, when and where’ dimensions to stress-test broad generalisations that naturally occur when working with large audiences,” Zohrer advises.
Planners can then generate nuances in targeting and messaging that are better aligned with decision journeys.
The final step is to validate the initial assumptions made as part of the plan, exploring whether the fragmented data the planner is working with risks delivering false errors in informing, targeting and reaching an audience.
“The role of the planner … becomes that of experimental designer to test for and around these data gaps through the different elements of the brief,” says Zohrer. “The test design is crucial to delivering confidence in the results that will eventually prescribe media decisions moving forward.”
Sourced from Admap