At the IPA’s 2020 Effworks conference, Les Binet presented his analysis on the potential for Google search to predict market share (link here). In his presentation, Binet talks about the “clear effect that advertising has on share of search” and he goes on to explain how share of search is a “nice, clear way of measuring the effects of advertising”. Indeed, Binet explains how Google search provides “a rich source of data on what people are going to do, what they think, how they feel and what they’re going to buy”.
Having used Google search data ourselves for a number of years and given Binet’s use of it to measure the effectiveness of advertising, an obvious question for us was whether we could do the same for media coverage? We therefore decided to conduct a similar piece of research to that carried out by Binet, but for online media coverage of brands rather than for advertising.
What really drives business outcomes from PR is not volume of coverage, but the audience reached and influenced by the coverage. The problem here is that PR uses volume metrics, largely because press cutting companies, many of which now offer evaluation services, have always and continue to make money based on the amount of coverage they capture and analyse. Volume metrics are deeply flawed when it comes to measuring effectiveness, first, because including all coverage regardless of whether people are likely to see it is going to skew the data and give misleading results; and second, because the audiences reached by different media vary enormously.
Our system overcomes these problems by capturing media coverage most likely to be seen by an audience and then applying an algorithm, based on in-depth academic research, which determines the probability of someone being engaged by the coverage to the extent they are likely to react to it. Over the past four years the algorithm, which determines the ‘engaged audience’, has been rigorously tested and it performs extremely well.
Another factor that needs to be taken into account is how audiences react to different types of media. We know from our own experience and research that broadcast media generate the strongest reaction but only over very short time frames, online media generate lower reaction but over much longer time frames while print media generate little direct reaction at all.
Most crucial, however, we have also found that 80% of the effectiveness from coverage is driven by just 20% of the media that generate it. Fortunately, we are able to identify the media involved, but this is yet another reason why volume metrics should not be used to measure the effectiveness of coverage.
Figure 1 reveals the relationship between the engaged audience for car brands from online media coverage versus share of Google search, for the period used in our study between 2014 and 2021. The relationship is strong, with a correlation coefficient of 0.81 at a significance level of 99.999…% (note that the correlation coefficient is the square root of R squared shown on the charts). You can also see from figure 1 how a brand such as Ford generates significantly more searches than the more glamorous Porsche, despite the latter generating stronger media coverage. The reason for this is cost. Ford is much cheaper and therefore significantly more popular than Porsche, as a result of which it drives proportionally more search volumes over and above those generated by media coverage. Figure 1b shows the same data as figure 1 but with the high and low-cost brands removed in order to minimise the influence of price.
The implications of these finding should not be underestimated. Given the clear relationship between advertising effectiveness and Google search and the clear relationship between online media coverage and Google search, there must be a relationship between online media coverage and advertising. Given the strength of the relationship between online media coverage and search we believe it highly likely that online media coverage is having a significant amplification effect on advertising’s effectiveness in driving consideration.
Figure 1. Share of search for car brands vs engaged audience, from 2014 to 2021
Figure 1b. Share of search for car brands vs engaged audience, with high and low-cost brands removed in order to minimise the influence of price, from 2014 to 2021
TV advertising, in particular, is expensive. While the impact of PR on integrated campaigns has been known about and used for many years, our results provide definitive evidence that what might be called ‘day-to-day’ online media coverage can have a significant impact on advertising and should always be taken into account when determining advertising’s effectiveness.
Having established the link between online media coverage and audience behaviour at the top of the funnel we also wondered whether it would have a measurable link with sales. Fortunately, the SMMT publishes new car registrations, which enabled us to extend our research to the bottom of the funnel.
Figure 2. Share of sales for car brands vs engaged audience, from 2014 to 2021
Figure 2b. Share of sales for car brands vs engaged audience, with high and low-cost brands removed in order to minimise the influence of price, from 2014 to 2021
Figure 2 reveals that that the relationship between the engaged audience for brands generated by online media coverage versus sales, again for the period used in the research between 2014 and 2021, is also strong (the correlation coefficient is 0.79 at a significance level of 99.999999…%).
In addition, we can see how brands such as Ford and Vauxhall generate significantly more sales than Porsche and Jaguar, despite the more glamorous brands generating stronger media coverage. As with Google search, the reason for this is cost, with Ford and Vauxhall being much cheaper and therefore selling more than Porsche or Jaguar regardless of the media coverage they might generate. As with search, we can remove these high and low cost brands in order to minimise the influence of price, shown in figure 2b.
Figure 3. Share of sales for car brands vs engaged audience, with high and low-cost brands removed in order to minimise the influence of price, from 2014 to 2021
Figure 3 shows the same car sales versus engaged audience data averaged for the period between 2014 and 2021. The influence of cost can clearly be seen for Vauxhall and Ford at the lower end and Porsche and Jaguar at the higher end. Interestingly, VW sales remained strong over this period despite the ‘dieselgate’ scandal, revealing that even the most damaging revelations can be overcome if the brand is sufficiently robust.
We also decided to look at individual brands to understand more about the link between online and media coverage and sales. What we discovered here is the importance of different types of online media. Engaged audiences generated by consumer titles are significantly more inclined to purchase a niche brand such as Abarth, while the engaged audiences generated by coverage in national media are more inclined to drive sales of well-known brands, such as Mitsubishi. This should not be too surprising given that Abarth purchasers are far more likely to look at automotive consumer media than someone who is likely to purchase a Mitsubishi.
As with Google search, the implications of these findings are very significant. They provide firm evidence that online media coverage – and therefore PR – has a significant link with sales. In addition, as already mentioned above, the relationship means online media coverage must be playing a role in advertising effectiveness. While we have only looked at automotive sales data to date we have no reason to doubt the same will also be true for other sectors, especially those where information plays a key role in the journey to purchase, such as white goods and mobile phones among many others.
We do, however, look at search data a great deal and figure 4 shows the relationship between engaged audience and share of search for supermarket brands between July, 2020, and June, 2021. As with car brands, the relationship is strong and typical of what we see across numerous sectors.
Figure 4. Share of search for supermarket brands vs engaged audience for the period between July, 2020, and June, 2021
In conclusion, there can be no question over the importance of online media coverage in the sales and marketing process. We are also seeing a great deal of evidence that the impact of online media coverage is happening over very short time periods – in many cases immediate – and will be reporting more on this in the next few weeks.
Ultimately, brands that properly embrace online media coverage as key to their marketing activities are far more likely to see improvements through the funnel. Whether directly as a result of online media coverage or indirectly through the amplification effect on other marketing and advertising activities, this provides tremendous potential for sales and marketing as well as advertising and PR to deliver even greater ROI – and therefore achieve higher budgets – over the next few years.