Back to list | Previous article | Next article
'Sales Uncovered': sales uplift and return on investment
This article by Guy Consterdine was published in the December 2005 edition of Admap. The research study was short-listed for the Media Week 2005 Awards for 'Research project of the year'.
How effective is magazine advertising? The Periodical Publishers Association (PPA), the UK’s professional body for magazine publishers, wanted fresh evidence on this. In particular, the PPA wished to use the new National Readership Survey accumulation data for distributing across time, week by week, the exposures generated by each magazine insertion, so that exposures could be related more accurately to purchases.
The PPA commissioned an analysis of TNS Superpanel data, and drew the conclusions that:
• Magazine advertising was associated with an additional sales uplift of 11.6% on average, in terms of sales value
• There were also significant uplifts in terms of sales volume, market share and market penetration
• In mixed-media campaigns it is often the case that too much is spent in television, when it would be more effective to allocate some of the TV money to another medium.
Analysis of the Superpanel database was conducted by TNS, led by Andrew Roberts, Tracy Waring and Imogen Coote. The PPA’s project is titled ‘Magazines Uncovered’ (1).
How the analysis was done
Superpanel’s 15,000 homes record their take-home purchases via bar-code readers and keypads, on a daily basis. The analysis examined purchasing records during the 18-month period August 2002 to February 2004. Panellists’ media exposure was measured through a self-completion questionnaire called mediaSPAN.
20 fmcg brands were selected for analysis according to a set of detailed criteria. One criterion was that magazines must account for at least 10% of the brand’s total advertising expenditure. Another was to focus on brands whose sales had increased during the campaign period, in order to examine the keys to success: how far was the sales uplift associated with the use of magazines. There also needed to be a distinct campaign period, and a pre-campaign period of similar length with no magazine advertising. The 20 brands were those which met the criteria and which spent the largest amounts on magazines. The cut-off point turned out to be a magazine expenditure of £325,000 or higher.
The NRS readership accumulation data (2) were used for distributing across time, week by week, the exposures generated by each magazine insertion. This meant that a more realistic comparison of week by week exposures and purchases could be made, than in previous studies prior to accumulation data being available.
Taking each of the 20 fmcg brands’ campaigns in turn, Superpanel main shoppers were ranked according to the weight of their exposure to the magazine campaign. The top 40% of main shoppers were defined as the ‘exposed’ group; in general, they accounted for about 90% of total magazine exposures. The bottom 40% of main shoppers in the ranking were defined as the ‘non-exposed’ control group; they only accounted for around 2% of total magazine exposures.
For each brand, purchases were analysed among the exposed group and the non-exposed control group, for each week during the pre-campaign period, and during the campaign period. Purchases during the campaign period were then compared with purchases during the pre-campaign period, separately for the exposed and non-exposed groups. The analysis was therefore based on tracking the purchases of the same individuals (the exposed group, and the non-exposed control group) through time. External events in the marketplace applied to both groups, and any differences in composition between the two groups were constant through time.
Differences between the two groups in terms of sales uplift (in the campaign period, compared with the pre-campaign period) were therefore associated with exposure to magazine advertising.
Further technical details of the analysis can be found in ‘Sales Uncovered’ (3).
Magazine advertising’s effect on sales
Aggregating the results of all 20 brands, there was an average sales increase of 10.0% among those not exposed to the magazine campaign – the increase being due to other activities than magazine advertising. However among those exposed to the magazine campaign, the average sales increase was 21.6%. Thus the magazine advertising was associated with an extra 11.6 percentage points increase in sales (in terms of value) – Figure 1.
Other criteria of performance
Other criteria of performance were also examined. Magazines were linked with an average increase of 18.1 percentage points in volume sales. Similarly, magazine advertising was linked to an uplift of 6.7 percentage points in market share on sales value and 8.6 percentage points in market share on sales volume.
It was also found that magazine advertising was winning new customers for brands and at the same time increasing the average weekly weight of purchase. Across the 20 brands, brand penetration of the market rose by 7.0% in the campaign period among people not exposed to magazine advertising, but rose by 15.5% among those who had seen the magazine ads – an uplift of 8.5 percentage points.
Meanwhile there were increases of 2.1% and 3.7% in average weight of purchase, among the non-exposed and exposed respectively – an uplift of 1.6%. Thus the sales uplift from magazines was achieved mainly by bringing new buyers to the advertised brands (i.e. increase in penetration), and only to a lesser extent by increasing the average weight of purchase – a much harder factor for advertising to affect.
Another finding was that in-store promotions work better when combined with magazine advertising.
Mixed-media campaigns: TV+magazines
Among the 20 brands analysed, seven were TV+magazines campaigns whose impact was assessed by medium. The average budget split across these brands was that 70% of the budget was spent in television, 22% in magazines, and 8% in other media.
For each of these seven campaigns, Superpanel main shoppers were ranked according to their weight of exposure to the magazine advertising (as described) and, separately, to the TV advertising.
For television, a medium with a more diffused audience than magazines, the 40% most heavily exposed to the television advertising were defined as ‘heavy viewers’; they accounted for about 72% of all television exposures. The bottom 40% accounted for about 13% of all television exposures, so they were described as ‘light/non viewers’ rather than ‘non viewers’.
Aggregating across the seven campaigns, main shoppers who had seen none or very little of either the television or magazine advertising showed only a small increase in sales during the campaign period: 3.9% (Figure 2).
By contrast, those exposed to the magazine advertising but who were only lightly or not exposed to TV, showed a much higher increase in sales. The same was true of those heavily exposed to TV but not exposed to magazines. For the two groups, the sales increase was 26%-29%. Clearly, advertising is effective in increasing sales.
Examination of the sales value figures showed that these two groups had similar purchasing levels during the pre-campaign period, and identical absolute increases in sales during the campaign period (hence the very similar percentage increases in sales).
The relative cost-efficiency of the two media may be examined by looking across all four groups shown in Figure 3. To do this, the absolute increases in sales value (£) are profiled across the four groups as represented in the pie chart, Figure 3.
Television advertising was closely linked to 71% of the sales increase: that is, the 46% among the TV+magazines exposure group, plus the 25% from the TV-only group. There may also have been some smaller effect among the group heavily exposed to magazine advertising and only lightly or not at all to TV.
Magazine advertising was also closely linked to 71% of the sales increase: the 46%, plus the 25% from the heavy magazines & light/non TV exposure group.
This can be compared with the profile of advertising expenditure: 70% on TV, 22% on magazines. While magazine advertising appears to have achieved something approaching the effect of television advertising, it did so at less than a third of the cost.
This does not mean that magazines are two or three times more cost-effective than television in all circumstances. What it indicates is that, pound for pound, magazines are more cost-effective at the relative levels of expenditure in these seven campaigns.
The reason, I suspect, is that television had been allocated too much of the budget and magazines and other media too little. In general (and with some exceptions in particular circumstances) the law of diminishing marginal returns applies to the effectiveness of advertising. Beyond a certain point, the more the money spent in a medium the less effective each successive tranche. The position can be reached where extra money is yielding very little return. In the hypothetical and simplified Figure 4 television is in this situation.
In view of the relative spend in the seven mixed-media campaigns (TV heavily outspending magazines and other media), I suspect that television is on a shallow part of the curve for these campaigns, and magazines are on a steep part (as in Figure 4). If TV expenditure was reduced, it would be the least productive part of the expenditure which was being removed, thus raising the average effectiveness of the remaining TV money. Average return on investment (ROI) would rise. If the money saved was placed in magazines, it would be yielding returns that were greater than it would have achieved if spent on television, though lower than the first tranche of expenditure in magazines. Thus magazines’ average return on investment would begin to fall, and the ROI of the two media would come closer (Figure 5). Eventually it could reach an equilibrium position where the ROI of both media were the same.
I draw the conclusion that if one medium is seen to be two or three times more cost-effective than another, it implies that the allocation of money between the media could profitably be modified.
(1) PPA: Magazines Uncovered, 2005. Results and interpretation are published in the form of a series of three brochures: Sales Uncovered, Planning Uncovered, and Communications Uncovered. Downloadable from www.ppamarketing.net
(2) Guy Consterdine: Distributing Print Exposure – A New Planning Tool, Admap, November 2004
(3) PPA: Sales Uncovered, 2005, page 18. Downloadable from www.ppamarketing.net
Admap is a monthly journal devoted to making advertising, marketing and research more effective (visit www.warc.com). The above is the unedited version of the article, slightly modified to appear on this website.