NEWS20 June 2013

IPA: Marketers should not rely solely on big data

Features UK

UK — Despite the potential for big data, there is still the need for first party and proprietary data, according to a study from trade body the Institute of Practitioners in Advertising (IPA).

‘The Big Opportunity – Audience Research Meets Big Data’ found that although big data can “greatly enhance industry currencies”, its growing availability did not remove the need for audience-centric data that provided a bigger contextual picture of the market.

In addition, despite the potential to be more accurate than traditional claimed behaviour surveys, the report said that it was “essential to understand the universe that a data set represents and as a result whether or not it is valid to project that data onto a wider universe”.

It added: “There is no safety in big numbers alone. Bigger is indeed typically better when it comes to sample size, but size it not the only key variable when it comes to assessing the reliability of a dataset.”

The study added that it was important to know where a dataset came from and what checks and balances were used in quality control.

In addition, the report warned: “The danger of relying solely on big data is that this will potentially limit our wider understanding of the impact of media exposure.”

It added: “Big data presents a quantum leap forward for media research. If used in conjunction with relevant industry survey data, it will amplify the strengths of both to create a whole bigger than the sum of its parts.”


1 Comment

11 years ago

A perfect example is data that relates to tagged internet pages. They provide a good count of the traffic, but poor counts of page views (as opposed to page impressions) and a poor count of the audience. A valid page impression may have been auto-refreshed, or sent to a browser page that is not the focus, so a 'duplicate view' or no view at all may have occurred. Further, with cookie deletion unique browsers over longer time frames greatly overstate the true audience. By 'super-imposing' key audience metrics derived from panels we can start to adjust that traffic data to represent meaningful audience data.

Like Report