Advertising effectiveness continues to be misunderstood, with the latest Freakonomics podcasts popularizing the low and uncertain individual-level effects of TV Advertising and Digital Advertising. As I blogged 7 years ago, this question is relevant to the academic, but not to the brand manager, CMO, CFO or CEO. What matters is not whether we can prove beyond reasonable doubt that our ads have changed any individual’s behavior. What matters is whether we can prove beyond reasonable doubt that our advertising investment has increased our sales performance, and by how much. Published academic research is full of such proof.
Above table 6 from ‘The effectiveness of different forms of advertising’ shows how the % sales increase for a 1% increase in advertising (‘advertising elasticity’). We grouped the many ad forms into 3 categories:
(1) Firm-initiated (FIC) are paid contacts without consumer action, such as all offline advertising and emails.
(2) Consumer-initiated (CIC) are paid contacts triggered by a consumer action, such as search or visiting a website.
Content-integrated CICs are an integral part of the medium’s editorial content. For example, consumers specifically access price comparison sites to obtain information on the searched-for items. In contrast, “content-separated” advertising means that the ad is not part of the site’s editorial content. Retargeting messages, traditional banners, and paid search advertising (typically appearing at the right, top, or bottom of the loaded search engine’s site) are examples of content-separated activities.
The data come from an online retailer managing the different categories of (1) fashion; (2) electronics, entertainment, and hardware (EE&H); (3) home & gardening (H&G); (4) sports and leisure (S&L); and (5) beauty and wellness (B&W). Across these very different categories, our findings are consistent: offline advertising and emails only have an elasticity of about 0.003, which means that doubling TV ads only increases sales by 0.3%. For the electronics category, there was no significant effect. That is indeed a lot lower than the retailer expected, echoing today’s reports about disappointing offline ad effectiveness.
Online advertising triggered by consumer action (CICs) is 10 times more effective than offline advertising. In every of the five studied categories, online advertising is more effective than offline advertising. However, we see a substantial benefit of content-integrated online ads: they are over 4 times more effective than content-separated ads. Doubling spending on content-integrated ads gives you over 13% sales increase.
Why is this the case, and why hasn’t ‘performance marketing’ picked up on this performance difference yet? Looking at ad effectiveness in different parts of the online conversion funnel, we see that content-separated ads are equally strong in driving traffic to the retailer’s website. However, this traffic is much less likely than that from content-integrated ads to convert to page views, checkout and sales order.
Mangers found these findings to have face validity, as the direction was often in line with their intuition, but our analysis added by pinpointing how much the retailer’s allocation should change. Applying our reallocation saw revenues go up by 28% and helped managers show how wrong last click methods are.
What does this research mean to your budget allocation today? Carefully consider which websites you are spending your online ad dollars on. Is the consumer on the site to browse and purchase products? Expect four times larger ROAS than on websites consumer flock to for other reasons, such as catching up with friends or searching for general information. Or at least test it out, and let us know what happens for your brand and category!
What does this mean to your marketing research? I fully agree with Andrew Willshire, who wrote 3 years ago in Adweek: ‘Most brands should forget about individual-level attribution and focus on the aggregate response. Chasing individuals around the internet produces more data than can be managed, yet not enough to solve the problem. Ten million impressions should have an effect on total sales or other conversion metrics, which can be quantified through marketing mix modelling. This should control for environmental effects and take offline media into account.’ If that was true when ‘chasing individuals around the internet’ was still possible, and it was, then it is even more true in our move toward a cookie-less future.
DIVE DEEPER: Our analysis is based on daily data and does not capture brand-building effects of advertising. Specifically, it does not consider how advertising reduces price sensitivity, allowing you to increase or maintain prices – we do so elsewhere for 350 brands in 39 fast moving consumer good categories. Moreover, we don’t consider beyond-customer benefits of advertising, such as the impact on investors (higher market cap), business partners and attracting talent.
Technically, our analysis is based on aggregate data and temporal causality, as measured in Granger Causality tests to show at the 95% confidence level that spending on certain ad forms goes up before sales (advertising to sales causality), while others go up at the same time as sales (activity bias). Accounting for other influences in structural vector autoregression (SVAR) models, we then estimated that paid search and retargeting were successful at getting people to the website, but not at converting them to paying customers. The published paper has more details on the estimation method.