| ||||||||||||
PRINT
COMMENT
SUBSCRIBE
RSSGlobal warming and consumers warming up to a brand or product: While it might not be obvious at first glance, there are certainly parallels between the two. First, both climate changes and consumers' responses to advertising are significantly more measurable today than in the past. And, second, in both areas new patterns, emerging from newly acquired data, routinely challenge our convenient ways of thinking and render our customary methods of going about our business inadequate.
Traditionally, the industry viewed advertising effects as a one-way street: advertiser launches a campaign, consumer responds to its ads -- and advertising was attributed as the factor driving consumer attitudes and behaviors. This was a convenience approach, as in many cases we only knew how to quantify the beginning of the process (reach and frequency of exposure) and the end of it (sales). The industry only occasionally considered effectiveness measures, such as branding, that happen in between.
Search and consumer-generated media introduced so much complexity that predicting advertising effects indeed became similar to predicting climate change. In digital ecosystems, just like in the atmosphere, feedback loops amplify the ways a consumer warms up or cools off to a brand. For example, both traditional and digital media campaigns generate online word-of-mouth. That buzz, however, remains after the campaign is over, keeps on reappearing in search results, and acts like a perpetual media placement in itself, continuing to influence consumer behaviors and their responses to consequent campaigns.
Online advertising enjoys steady increases in spending, primarily because it's the most accountable. Unlike print ads, digital banners come with a full repertoire of relevant performance measures: impressions served, clickstream data, direct and indirect (view-through) conversions and rich-media interactions. Digital video is far more measurable than TV commercials, as it yields useful branding metrics such as interaction time, re-plays and pauses -- and, as with any online ad, we can track video influence to subsequent online brand interactions. In addition, we can routinely measure paid search, site visits, blogging, message boards' postings and brand terms' search volume, and correlate them to media exposure.
However, do all these new metrics really translate into more effective campaigns? If, after "having invented the Internet," Al Gore had not moved on, he would have discovered a different inconvenient truth: Online advertisers really only use a fraction of insights contained in their data.
Data analysis is only valuable when it shows what causes what. Global warming opponents often point out that despite the plentiful data, it is still hard to separate the effects of human activity from the planetary cycle or other natural trends. Online advertisers face a similar challenge when trying to establish what proportion of desired consumer behaviors is actually caused by advertising and what would happen anyway, regardless of exposure. It is actually possible to answer that question today, but it requires control ad placements and processing vast amounts of data. So, often online marketers rely on assumptions and shortcuts instead.
Proper attribution becomes particularly relevant when addressing tracking view-throughs (when a user saw an ad, did not click on it, but subsequently performed a purchase). Advertisers typically track such conversions back to the Web sites or networks that delivered the ads and credit the most recent online marketing exposure. However, the fact that a consumer saw an ad prior to making a purchase does not prove that that ad prompted the purchase.
As a result, placements with higher reach often get credit for leading to sales, because future consumers, influenced by an array of media, are more likely to also encounter these high-reach placements prior to purchase.
The methodology exists to isolate valid view-throughs from those that would occur even without media. But we still lack industrywide statistics and studies that determine the instrumental factors for generating the valid view-through response.
A similar challenge exists with sponsored search, to which advertisers typically attribute all online conversions occurring within one month after consumers click on sponsored links. It does not happen in a vacuum, which leads to the billion dollar question: To what extent does search drive conversions independently and to what extent do consumers use it just as a navigation tool to find brands for which they previously saw a digital display ad or TV commercial?
In advertising, as with the atmosphere, we can only establish causality through constant observation and testing. Marketers should abandon outdated linear models along with search and display tracking silos and crediting the most recent online exposure for purchases. We need to take into account more touch points, both online and offline, to identify what's truly working.
24 people recommend this article.
RSS FEEDS
Because part way into it, this appears: "If, after "having invented the Internet," Al Gore had not moved on, he would have discovered a different inconvenient truth: Online advertisers really only use a fraction of insights contained in their data."
Any author who feels the need to keep that old and thoroughly disproven Gore "quote" alive doesn't deserve serious consideration. Until the writer catches up on his homework, uses more than a fraction of HIS insights, and leaves the political agenda out of his writing, he's on ignore.
While the digital sphere is 'metric-rich', and traditional media are comparatively poor, those same traditional media seem to have done a great job over the years building the great brands that we all know and love. As a matter of fact, I struggle to name a brand that has been built in the digital sphere - apart from digital brands per se such as Google, Facebook et. al.
The issue with said metrics is the "attribution of success" that is subsequently attached to them as you correctly point out. To me, the issue is not the linearity of the models, but the bivariate nature of these models. By this I mean that "the model" ends up with only two factors - the dependent variable (the sale) and the independent variable (the "cause" of the sale). It is naive in the extreme to think that a single marketing action results in the sale - but that is the direction that digital metrics are (incorrectly) taking us.
We need to look at the reason why. The digital sphere track their metrics in all their richness while remainingly completely 'agnostic' of other marketing inputs. The same happens in traditional media (though with less rich metrics). That is, the TV network 'claims' the sales success, or the magazine does likewise. Why? Because they don't have access to each of the other medium's data. Only occassionally do you see studies like "The Media Multiplier Effect".
So, whose problem is this? It is the marketers problem. They own the brands - but rely on the media to tell them what worked and what didn't. So, it should come as no shock that the digital world tells them that THEY were responsible for the brand success, while at the same time television claims the success. Who is right? Well, hey BOTH are.
The issue here is that we have to move to multi-variate non-linear models of brand success. Further, marketers need to realise that no single medium can do this. That is, THEY (or their agents) need to be solely responsible for establishing how their marketing communications work and create successful brands.
To my thinking there are two ways this can be done. The advertiser can 'seize the day' and themselve employ extremely intelligent analysts specialising in multivariate non-linear modelling. The other way is to 'outsource' this.
Who is best placed to do so? The sector of the business that needs to step up to the plate is the media agencies. It is the media agencies that need to have sophisticated modelling and analytics divisions that the advertisers engage to do this work for them. Why the media agencies? Because they understand the metrics. The metrics of each media segment are unique beasts and cannot be simply bundled together in a marketing model - does 100 GRPs equate to 2 x DPS, x number of page impressions, y number of billboards etc etc. Media agencies know this because they spend millions of dollars a year subscribing to the data and continually using it on a daily basis. Outsourcing this function to a 'specialist audit' firm is dumbing down your model and prone to very expensive errors until they understand the nuances. Every one of them will say - "hey, we're a specialist - sure we can do it". And as soon as the models run off the rails, they will go straight to the media agency to ask why.
So why isn't this happening on a large scale? In my experience it is mainly because the advertisers want this to be "thrown in" as "part of the service". Analysing brand success is VERY different to planning and buying media and must be remunerated separately.
In Australia, most media agencies do have analytics divisions capable of such analysis. The challenge is getting the advertiser to pay for the analysis, and not to expect the media agency to include this as part of their 3% commission.
John Grono GAP Research Sydney Australia