Commentary
How Advertisers Give Credit Via Multi-touch Attribution
- by Danielle Shamos , Op-Ed Contributor, May 9, 2017
Marketing Mix Modeling (MMM), statistical analyses first introduced in 1990 and still used today by many global marketers, uses historical data and makes assumptions about the impact of marketing tactics on overall business growth.
Brands, in turn, have been using this data to optimize cross-channel media spend.
The growth of nonstandardized, log-in platforms, like Facebook, Twitter and Pinterest, which provide unparalleled insight into their individual users — has given way to a clearer understanding of how media affects online and offline sales.
While ad viewability has been part of the chatter around transparency in the industry, the focus for advertisers has shifted to how different channels deliver on actual business objectives and how they work together to drive efficiencies.
Brands want to know which digital channels are most effective along the consumer journey, and how they’re able to increase not only favorability, but also conversion.
The path to purchase has become more complex, since consumers engage with brands at several touch points before making purchases. As behavior has evolved beyond a simple search followed by a purchase, the last-click marketing model has faded in popularity; it can undervalue ads that deserve credit.
For instance, a 2015 study by Datalicious, commissioned by Facebook, revealed that last-click discounts the effectiveness of Facebook ads by 911%. Barriers have taken longer to break.
In order for walled gardens, like Facebook, to demonstrate their value alongside traditional models, they have to share information to contribute to the greater good of holistic marketing measurement.
In the past, much of Facebook measurement data had been evaluated in silos, and it was not easily comparable within the multi-touch attribution model. Facebook has made significant progress recently toward giving marketers a more complete picture. It has partnering with Neustar MarketShare and Visual IQ to share aggregated data across ad formats, including reach and impression volume by week and geographic area.
Integrations with these third-party companies allow ads across Facebook, Instagram and the Audience Network to demonstrate their full value in a multi-touch attribution model. They give appropriate credit across multiple impressions and clicks that live within the consumer's path to purchase.
Facebook is also offering Advanced Measurement. With the solution, advertisers that are not currently leveraging Neustar MarketShare or Visual IQ can gain an understanding of how ads across Facebook properties drive reach and sales — and how they stack up against campaigns on other channels.
These are just the first steps in platforms providing a more defined breakdown of ROI for advertisers.
When approaching the performance of omnichannel campaigns, there are four important things advertisers should do:
1. Consider the complex path to conversion and how that goes beyond last-click attribution.
2. Leverage a multi-touch attribution solution to get insight into how to optimize the media mix and use it to impact planning and buying.
3. Take advantage of the instant feedback on paid social to iterate along the way to continue driving improved results.
4. Get comfortable with the difference between media objectives and business objectives.
One final point to remember: While viewability is important and should be evaluated, behavioral change and tangible business objectives are most crucial.
Danielle, I agree that it's important to evaluate an ad campaign's results based on tangible results---to the extent that these can be defined channel by channel, without considering the effects of audience duplication, prior advertising, and many other variables. However, no campaign is going to be effective if its reach and/or frequency is compromised because 50% or more of its served ads can't be seen and, of those that make it to the uisers' screens, many are truncated or only "visible" for a second or two.
Danielle, thanks for sharing. Just wanted to quickly chime in to add to Ed's comment about viewability. Completely agree with his point and would also ad Invalid Traffic as an ongoing concern amongst brands, agencies and media providers. Combined, non-viewable and fraudulent traffic, if not detected, can adversely affect even the best attribution modeling systems. Marketers have gone from cross-device targerting as the hot trend in 2016 straight to MTA (Multi-Touch Attribution) in 2017. This is a tremendous leap and an exciting one; however, there still exists that challenge of ensuring the data being used for modeling is "clean" data. Brand marketers can truly harness the promise of MTA if they are thoughtful and thorough in their collection, filtering and unification of the various marketing touchpoints.
The article was going strong until this sentence ... "Brands want to know which digital channels are most effective along the consumer journey".
Surely "Brands want to know which (omit digital) channels are most effective along the consumer journey".
I see that there is nary a mention of offline marketing activity (still the bulk of consumer influence at a brand level), nor a mention of competitors activity. Oh, and did I mention factors like weather, season, the economy etc.
Oh for the bliss of such a simplistic model.
Thanks, Ed. I agree that campaigns with compromised reach and frequency due to viewability will not be effective. The rise of viewability solutions offers more ways to measure performance consistently across channels. I think it’s good to evaluate alongside tangible business results. If you’re using log-in platforms and targeting based on user data and setting the optimal reach and frequency levels to achieve desired business outcomes but you’re falling short, it’s important to evaluate how viewability may have had an impact. Then determine which platforms are delivering against business objectives and observe the correlation to viewability. In addition, consider the targeting, creative, and timing of the cross-channel campaigns.
Thanks for the comment, Omer. I also agree that this is an exciting leap forward but it’s still a highly complex environment that requires meticulous and comprehensive data collection to capture the full value of MTA models. Since the inventory on log-in platforms like Facebook, Twitter, Pinterest, and Snapchat is primarily native with a focus on high-impact newsfeed or timeline placement plus targeting is based on user data, the fraudulent traffic concern for brands is less than if they’re buying across 100s or 1,000s of sites within a display network.
Danielle, it might not just viewability.
Duplication of contacts may also be an issue. As the reach of a campaign increases (i.e. from higher weight), having de-duplicate audience data is increasingly imperative (something TV does very well).
I agree, John. Brands are not only evaluating digital but also offline efforts. The focus of this particular analysis is how brands evaluate paid social and other digital channels utilizing multiple touch points along the consumer journey. This is not to say that there are not other valuable offline data points within brands' omni-channel efforts that influence consumers along their purchase journey. I also agree that competition, weather, seasonality, and the economy are constant factors that impact all brands and their digital and offline marketing as well as business outcomes.