Can Old Media Enhance New Media? How Traditional Advertising Pays off for an Online Social Network

Back to the Peer Reviewed Journals List

Pfeiffer, Markus., Zinnbauer, Markus. (2010). Can Old Media Enhance New Media? How Traditional Advertising Pays off for an Online Social Network. Journal of Advertising Research. (50), 1

Reviewed by Christopher Berry, May 2010

Executive Summary

Pfeiffer and Zinnbauer run a media mix model study and find that old media is important in generating awareness while new media is important in pushing through to conversion.

Review

The authors used very granular data from a pure online company – a social network – to explore the relationship between traditional advertising and online advertising.

The firm in question spent ($USD) 12.4 million in TV, 3.4 million in Print, 2.7 million in Outdoor, 13.2 million in Internet, and 0.4 million in Other – for a total in 31.1 million. They drove 2.4 million registrations, 1.8 million activations, 92,000 in first sales, 150,000 repeat sales, and 24,000 in win back. The question is ‘what was the effect of advertising spend per channel on business success?’

They added additional factors to the mix, such as the existence of major holiday, a major event, hours of sunshine, and unemployment. They were able to derive a baseline, and, using their statistical method, could isolate the incremental impact of lift that each channel delivered. They express the contribution of each media channel in additional registrations.

Based on equal spending for each channel, they found that TV spend drove 2037 extra registrations per day on average, print drove 365, SEM drove 1432, billboards drove 279, while their radio spend had no discernable effect. They found that Christmas/New Years drove an extra 1240, the Olympics 127, and for a 0.1% increase in unemployment – an extra 75. Tellingly, they found that a competitors radio add detracted 683 and an increase by an hour of sunshine detracted 232. (Yes! Not all factors have positive effects!)

They performed additional analysis and found that different channels were more effective at different times along the conversion funnel. For instance, SEM was 3 times more effective in getting somebody to a paying membership than TV.

Their model was far more predictive at the top of the conversion funnel than at the bottom. (Their model has an R^2 of 0.93 at the top of awareness funnel, while down the funnel explanation was 0.67.) The authors go onto state that word of mouth and experiential effects might be very useful for improving the model, and that further research is needed on that front.

Why is this Important To You

This should inspire you. This is the first transparent multi-channel mix analysis that I’ve seen as applied to a pure online play. Data can’t hide in the brick-and-mortar. There is nothing to scapegoat. The central conversion events were tracked, and, data that analysts in particular rarely consider were incorporated: macroeconomics, sunlight, and competitive.

What’s notable is how important old media is in terms of getting people to the top of the funnel – and the importance of search to get them in deeper. The importance of the user experience and word of mouth and getting them all the way through, and that impact on repeat sales, were not covered. The user experience in particular should be of interest to web analysts, who, at the root, should be about optimizing the user experience.

Analysts are empowered to perform similar analyses based on the daily data that may be at their disposal. To perform the analysis, analysts require PASW, R, or, if really stretched – Excel. Depending on the specifics of what data you have at daily granularity, the nature of your business model and customer LifeCycle, try extracting the daily volume of traffic arriving by way of search engines, the days that email blasts are going out, direct traffic, paid search traffic, paid search cost, news events, word of mouth (however you benchmark that presently) and conversions. Run a correlation (and with PASW, you may time shift the correlation backwards or forwards in time to check for lag effects) and prepare to be quite pleased. The next question to ask is ‘why’ and ‘how can I exploit that insight to drive more money for my firm?’

Analysts really are not limited to analyzing what is just in front of them.

I recommend that interested members of the Web Analytics Association should read this journal.

A single copy of the full journal reviewed above is available to members of the Web Analytics Association. To request a copy, email Shannon Taylor.