Analysis Finds Nielsen Data Used By Modelers Undercounts TV's ROI By As Much As 20%

A first-of-its kind analysis of TV audience data used by modelers to measure the return on investment (ROI) for ad spending found that it is sorely lacking, and in some cases misrepresents TV’s ROI by as much as 20%.

The findings, which were unveiled Monday during the Advertising Research Foundation’s Audience Measurement Conference in Jersey City, NJ, are based on an exhaustive analysis conducted by Sequent Partners and Nielsen for the Council for Research Excellence (CRE).

The analysis, which is summed up in a white paper, presented by Sequent partners Jim Spaeth and Alice Sylvester, found that a combination of legacy factors in the way Nielsen data is processed for third-party modelers contributes to inaccurate and/or weak data distorting their outputs.

In 90% of the cases that Sequent and Nielsen analyzed, the effect was relatively small, distorting TV’s contribution to advertising ROI by less than 10%.



But in 10% of the cases, the effect was relatively significant, reducing TV’s ROI by 20%.

The analysis examined a variety of factors, but the No. 1 contributor to the distortion was the inability of modelers to align the actual audience delivery with the commercials that ran in the advertisers’ schedules, because the data supplied to modelers by Nielsen does not represent exact minute ratings, but are based on average quarter hour audience estimates. 

The findings are significant, says Sequent’s Spaeth, because many big brands rely on mix models as the basis for allocating advertising budgets to media. He said an underestimate in the range of 20% might be enough to knock TV off an advertiser’s plan altogether.

Spaeth likened the phenomenon the the kind of “last click attribution” analysis used by modelers in digital media, which gave too much weight to the digital media that had the last point of contact with a user before they clicked through to take an action.

Spaeth said more work needs to be done to figure out better ways to align Nielsen’s audience data with actual advertising exposure, but he said Nielsen has already begun taking steps to fix some of those problems.

2 comments about "Analysis Finds Nielsen Data Used By Modelers Undercounts TV's ROI By As Much As 20%".
Check to receive email when comments are posted.
  1. Peter Rosenwald from Consult Partners, June 13, 2017 at 10:59 a.m.

    This is a tricky but extremely interesting issue.

    In a piece written for 'Target Marketing' (which will be published in tghe next week or so) I posit that the ROMI (Return on Marketing Investment) must be a composite of measurable and unmeasurable data: the value of the measurable revenue generation must have added to it the value of prospects who are exposed to the brand but do not act immediately.

    Someone speculated that 'perfect' is the enemy of 'good'. With all due respect to Sequent and Neilsen, the idea that these metrics are ever going to get down to decimal point accuracy is perhaps rather over-ambitious and unrealistic. 

  2. Ed Papazian from Media Dynamics Inc, June 13, 2017 at 12:40 p.m.

    Quite right, Peter. The problem is that many of these so-called ROI evaluations don't even bother with audience data and rely only on ad spending. Even when some attempt is made to account for GRPs, no adjustment is made to reflect actual ad exposure variables, nor reach attainment on a short term basis---like week by week. Moreover, there are many other variables, not so easily quantified---like new vs old ad campaign, new vs old product, brand image and long term marketplace and brand sales momentum, ad awareness tracking, sales point registration, promotions,pricing discounts, product distribution,overall product category dynamics, what competitive brands are doing, brand loyalty aspects, etc. etc. As a result the findings, while usually noting some correlation between promotional activities and sales, fail to offer the types of sensitivity and detail that might prove useful to CMOs---especially where assessing the contribution of various media mixes and/or individual components of the media plan are concerned.

Next story loading loading..