Lifting this up from Clickz:
Microsoft's Young-Bean Song began a speech at the MIXX conference in New York yesterday with a beer analogy: Consider a scenario, he said, in which Corona determined to put its whole ad spend toward whatever consumer contact point immediately preceded most of its sales. Its obvious move, according to Song, would be to "put all its marketing budget into neon signs at pubs."
Ridiculous idea, but Song, who is VP of analytics in Microsoft's Advertiser and Publisher Solutions group, argues most digital marketers assign credit in just that fashion -- with search, affiliate marketing, and other sources of lower-funnel clicks grabbing more recognition than they rightly should.
Microsoft wants them to think more holistically. A new study based on advertisers' use of the company's Engagement Mapping product argues for a model where more credit goes to high-funnel channels like special interest sites and ad networks. The data, drawn from 500 advertisers that are using the fledgling tool, do indeed show that marketers given the choice to assign greater value to those channels do so.
Broken down by channel, Microsoft found that under its attribution regime, vertical content sites and ad networks gain credit for conversions. Meanwhile, the median search buy loses 7 percent credit for conversions among those using Engagement Mapping.
Furthermore, Microsoft research indicates navigational search is responsible for a large portion of click behavior. Sixty percent of total sponsored search clicks specify the precise name of a brand advertiser, or even an exact URL. And search clickers exposed to display advertising were 22 percent more likely to produce a sale than those who weren't exposed.
Engagement is one of those words that somehow get me going. Not in a good way. Really.
For some reason, the "word" just rubs me the wrong way. It is so fuzzy a concept - and because the concept is fuzzy, it cannot be measured. Not yet, at least.
I have seen presentations from clients directing their creative and media company-partners to "engage" the consumers, or create ads and plans that "engage" consumers. But what does that really mean? Does that mean "surprise" consumers? "Delight" consumers? "Amuse" consumers? "Entertain" consumers? "Transform" consumers into brand recallers - or brand ambassadors?
I have also seen presentations from media companies - both from the planning and the ad-space side - defining "engagement" in various ways. Some see it as "time-spent" with a medium. Some define it as "compo" - i.e., % rating against a target group divided by the % rating against the entire population (that's affinity to some).
Some have come up with more complicated ways.
(Well, I did try coming up with one for magazines in Singapore a few years ago, when I included several variables - AIRs, time-spent per issue, average % of the issue read, average number of times re-read, response indication, and magazine-brand image associations - and rolled it into one score using some *secret sauce* I learned from IRT and psych-testing in the university; I was told that it was too complex. But I am not giving up hope. Was I measuring "engagement"? Not sure. The goal was to come up with a new currency - tick that box. But did it get accepted? Hmm. Leave that tickbox blank. Haha.)
So is Microsoft finally on to something with Engagement Mapping? Its methodology is unclear to me just yet - here's what the article said about the Microsoft methodology:
It assigns more or less weight to different ad impressions based on factors like frequency, recency, and ad size. Larger ad units get more credit, as do video and rich media. More frequently and recently delivered ads are also given greater influence. Advertisers can also tweak the dials on Atlas's interface to assign greater influence to those Web publishers and ad formats it deems most powerful.
I think they may be on the right path - but if it is indeed Engagement, then there should be a component of a response, right? It may be circular, I know - using 'behavior' as part of the engagement construct in order to predict behavior. But that just might work.
Perhaps, if they added more variables: frequency, recency, ad quality (not just ad-size), ad version (say, is it more creative than the rest?), context and environment synergies, and historical response - it just might work.
But then again, the question remains: Exactly what is Engagement? And as a media planner, or an advertiser, or a brand manager, what do I do with it? Can I buy it and plan it? Can I relate it with sales and put a dollar value on it?
Hmm.
I think Engagement will remain to be one of my most favorite, fuzzy words.
A Flickr Photo from Rubyran