Attribution modeling: Q&A with Dustin Engel of Range Online Media
Over the past five years, the client summit has become a popular, annual event among search engine marketing agencies. I have had the privilege to participate at a number of these events; most recently, Range Online Media's Client Summit and Fall Fest, which was held last week in Dallas. While I initially intended to report on a social media panel that I moderated, I felt compelled to relay that almost half of the afternoon was not spent on SEO, PPC, social media or even creative development — rather, it was spent on attribution models that effectively move measurement away from the last click before conversion.
Now that search engine marketing is maturing, marketers strive to understand the effect other media have on the ultimate conversion. While multiple data sources suggest that there is a clear correlation between media touchpoints, there are multiple barriers to solving this equation. Earlier this year, Jupiter Research released a study to this effect, noting that 70% of online advertisers seek greater campaign measurement beyond the last click, yet only 13% have implemented attribution models that do so.
Dustin Engel, director of strategy at Range Online Media, moderated a panel to sum up the afternoon's attribution discussion. Later, he sat down with SearchBuzz to share some additional thoughts on the topic.
Q: Range does well within categories that have an online conversion. How active are your clients in exploring or adhering to an attribution model?
A: Range's clients adhere particularly well to an attribution model with varying degrees of complexity. Range believes in the concept of unified tracking, and we have worked with our retail, luxury and travel clients to find more sophisticated ways to track marketing activities.
Q: What does unified tracking provide?
A: Utilizing a unified tracking solution allows an advertiser peace of mind that click data and post-impression data are de-duplicated. At the basic level, the majority of our clients that we work with on cross-media campaigns are attributing view-based conversions. We currently work with many of our clients to get more advanced — and therefore efficient — in their attribution modeling. No one model is right for every advertiser, but by identifying activities that create relationships between the advertiser and the consumer, and weighting and cross-measuring those activities, you can determine which model is best for each unique business.
Q: What are some typical examples of unified tracking systems?
A: While there will always be disparate data out there, and not all items will be covered in the unified tracking solution — for example, e-mail tracking is a typical offender — unified tracking provides strong directional data. This is typically an ad-serving solution such as Atlas or DART, or an analytics solution such as Omniture or Coremetrics. To be clear, this is the first step in attribution. The positive is that advertisers utilize post-impression data and are not tied to a consumer clicking on an ad to show value. The opportunity is that this is still a “last ad” model and the value of the impressions of a display campaign in creating an emotional connection, creating a more educated and intent consumer, and displacing competition is overwritten by a click. Hence, there is much more actionable data to see and utilize.
Q: Can an attribution model be a source of competitive advantage?
A: When looking at the whole consumer path and identifying which activities are the most impactful to the business in a variety of areas, such as creating a more educated and intent consumer and creating actionable rules, it can.
Q: How do you create this type of consumer?
A: As brand advertisements typically carry more weight than positioning and differentiation (outside of price), the effect will typically be a more educated and intent consumer once they fall into the sales process. Therefore, identifying metric changes such as conversion rate — because need is established prior to arriving at the site — retention and repeat purchase — if more properly educated in the brand and product/service features, there is less of a need to win with price and perceived value in purchasing direct with the advertising brand — and average order value. Take the case of Apple's Mac campaign. A higher focus on positioning as a lifestyle choice enables them to displace competition and be the “cool choice”. Thereby, it is rare that they have to compete on price, even in a down economy. Very few Apple ads showcase price point and that is a media-driven reason why they are able to sell at a premium.
Q: Can you tell us more about creating actionable rules?
A: By having more visibility, an advertiser will be able to create more actionable rule sets by putting value in the right places. As Young [Song] from Atlas outlined [earlier in the day], the end result is that all marketing activities surface in a dashboard with a common barometer of success. On the back end, the weighting of value on each activity changes based on rules. This becomes advantageous, because you can tune these rule sets based on market and economy dynamics. For example, a retailer may know that in January or February, consumers are less likely to buy. However, it is a great time to build the database and educate on the brand so they could invest in brand media and social media with a “non-selling” messaging strategy. The rules for their media mix could change from a sale being the primary success criteria to e-mail signups. The top-performing media under that methodology can — and will — look entirely different, but will be just as strategic.
Q: What are the barriers to creating such an attribution model?
A: Toolset challenges, a lack of standards and the economic cycle. There are many toolsets that cover pieces of the equation, but there is still no end-to-end solution. Ad servers are, in my mind, the leaders as far as providing a solution — but they still have incremental tracking costs associated with them. Therefore, an advertiser must be heavily invested in the exercise and be willing to tag every online marketing channel. Analytics solutions typically fall short in understanding post-impression data. They rely on click-based data only — although you typically can import post-impression data into them from your ad server. The plus of an analytics solution is that the providers have created intuitive dashboards outlining effect of multiple clicks. However, without post-impression data, there is a major gap. There are also providers of individual solutions out there, but critical mass of these solutions is yet to be realized.
Q: Can you comment on the lack of standards?
A: The majority of advertisers need a standard in order to be comfortable moving forward on an attribution strategy. However, every advertiser brings different dynamics to the equation. For example, a retailer with thousands of retail stores that consumers pass by on their way to work (an underappreciated advertising medium in its own right) brings a different brand equity equation to the ecommerce world than an online-only retailer. This exemplifies a case of two extremes. But the reality is that each advertiser needs to make different tweaks and tuning to their attribution strategy. This requires effort and resources, and therefore, inhibits mass adoption.
Q: And last, but not least, the economic cycle?
A: Advertisers right now are heavily affected by the economy. While downstream channels aren't necessarily affected, upstream channels and also a desire to experiment with new strategies fall by the wayside. I do expect that once the economy stabilizes and toolsets/solutions have put some finishing touches on their solutions, then the game will change and change fast.