Agencies: Data Visualization Still 'More Flash Than Substance'

Data VisualizationAn array of visualization products have emerged to help marketers turn "big data" into smart data.

This week has brought announcements around visualization from the likes of Tag Man, Placeable and SumAll. Meanwhile, plenty of other data and analytics vendors -- including Tealium, Adobe, IBM, Acxiom and Domo – are investing in this area.

But while senior agency execs welcome advances in making data actionable, many are unimpressed. They say today's data viz products commonly confuse accessibility of data and how to make it actionable.

"Everyone wants big data, but when you look at adoption of big data solutions, it's lacking because agency people aren't seeing the relevance," said Elliott Trice, VivaKi's global experience officer, whose job involves developing the Publicis media operation's tailored data-visualization products. "A structural shift needs to happen. You need to have a team working with users in various parts of the agency on what their specific needs are."

One problem is that agency staffers aren't always clear on what their data needs are, apart from discrete metrics. Few media industry staffers have an MBA in analytics, which would help them parse the data and understand which custom visualization package would benefit them.

The big need now is not so much technological. It's actually as simple as step-by-step guidance. But that can be costly, and requires trust on the part of agencies with the data analytics providers they employ.

Visualization's Limited Role

This week, TagMan produced an upgrade to its Visual Insights Suite, which is intended to give brands an "intuitive" look at how their search marketing has led to purchases. Meanwhile, analytics startup SumAll has integrated Foursquare check-ins for its clients, offering to provide a clearer sense of how promotions have been working for restaurants and events. But for the moment, these "improvements" are about making data more quickly readable than actionable, agency executives say.

"The early iterations of data visualization are typically more  flash than substance," said Michael Dowd, chief technologist for WPP's GroupM Next, when asked about the kinds of visualization services available. "Dashboards collect a lot of attention, but unless those data visualizations are directly facilitating the implementation of better strategies, they are not particularly valuable to us as marketers. Some marketing concepts are simply too complex to represent cleanly within a dashboard."

That said, the visualization dashboards capably serve two purposes, Dowd said: representing data for those who are not actively making strategic and tactical decisions, and guiding deeper analysis for those who are.

Dowd said he sees these products as "a supplement to existing marketing strategies, not a replacement. And ultimately, tools like the ones from TagMan, Placeable and SumAll can become an asset to marketers and businesses alike."

Function Over Aesthetics

The main virtue of data visualization is that it helps agency executives and marketers make sense of the reams of spreadsheets and numbers through immediately understood patterns. It's more than just pie charts, but it doesn't have to be more complicated than that. But, more than aesthetics, the point is to give structure to vast collections of consumer data from numerous online and offline channels, noted Anna Nicanorova, data scientist at Omnicom's Annalect. Even though media buying is the flip side of the creative part of the agency,  media plans and results need to be communicated in a narrative fashion.

When choosing a data-visualization provider, Nicanorova said she investigates the functionality of dealing with data, rather than the final visual output.

"A lot of applications will claim that they were created for big data, but under closer examination we constantly find a lot of restrictions on the size, types and file formats," Nicanorova said. "We always closely examine the back-end infrastructure that handles data ingestion, to make sure that the specific visualization tool provides a proper pipeline to get the data into the application in the first place.

While there is a lot of attention being paid to chart types and what colors to use, Nicanorova said she is noticing more practicality from clients these days. In particular, there is a greater demand for "data interactivity." In essence, it's more valuable to have all the information filtered into a single chart, rather than having to sift through a pictorial version of different data sets. The way Nicanorova would like to see the final visualization is through the ability to hover, zoom and filter the visual information she receives.

The big things are control, synthesis and customization.

"A beautiful data visualization that doesn’t help achieve a client’s KPI is nothing more than a pretty screensaver," she said. "You can’t have an all-encompassing view because it’s going to result in information overload. … If a client is looking to increase ROI and their dashboard isn’t helping them to do that, then they should look for a better solution because quantifiable results are what good data visualizations should provide them."

1 Comment

  1. Ironically, the most challenging aspects of data visualization is what you don't see. To wit:

    * Data acquisition and cleansing (in industry parlance sometimes called 'ETL' for 'extract-transform-load') can account for as much as 80% of the costs in managing the varieties of digital exhaust given off by ad servers -- according to agency CIOs

    * True interactivity -- the slicing, dicing, zooming, & filtering that Nicanorova discusses -- requires a uniquely powerful database, especially when the data points number in the billions.

    A clean, fast spreadsheet is always preferable to a pretty, slow data visualization.

    Reply

Add a comment

XHTML: You can use these tags: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>