Q&A: The Building Blocks of Attribution – Moving Beyond the Models of Yesterday

As competition heightens, technology evolves and budgets tighten, the ability to pinpoint which touchpoints and events are driving sales in your customer journey is an evermore crucial element of online marketing.

Impact Radius knows the importance of this, and to such an extent that the company is built around it; assisting clients in achieving a top-down view of their customer journeys, helping them decide where to concentrate their efforts.

Next week at Performance Marketing Insights: Europe, the group’s chief revenue officer, Tijs van Santen is on hand to explain developments in this area, how we’re moving beyond the models of yesterday, and how companies can start to integrate these strategies.

As digital marketing evolves, the importance of attribution only seems to get greater. Why is that?

Tijs van Santen: Just think about it – significant advertising budgets, increasing scrutiny on spending and more and more data-driven decision-making requires digital marketers and their peers to know exactly what aspects of their marketing programmes are yielding the intended results. This is why attribution data and insights have become so important.

Attribution begins with advanced tracking technology across devices and channels that exposes the entire consumer journey from the first time a consumer engages with, or is exposed to, a brand’s advertising, through to each and every conversion that customer completes. For omni-channel brands, there is added complexity with online and offline engagements and buying.

It is important to note that the tracking should be able to eliminate non-human traffic and impressions so that the consumer journey is properly represented.

The insights we can pull from the Attribution data finally solve the age old John Wannamaker problem –  “Half the money I spend on advertising is wasted; the trouble is, I don’t know which half.”  Now we know.

Is the ability to move forward with attribution something that’s afforded to most marketers, and if so, to what degree?

TVS: Some attribution solutions offer a basic level of low cost or free models and reporting to any advertiser, regardless of size. This can be a great start, but the limitation is that the modelling is standard and the reporting is fairly basic, often lacking granular insights and cost data. It also limits the ability to look at specific customer journeys that are aligned to advertiser’s specific goals.

Think of new versus returning customers, high-margin orders or even insights at a product category level, for instance.

Smaller brands are commonly challenged as they may not have the budgets to hire outside help; nor do they have the bandwidth to really use attribution data and insights on their own. If they decide to move forward with an attribution solution, they need to commit to the process and learn how to gain insights from this new data set in order to increase their ROI on ad spend – enough to justify the time, effort and cost they are putting into it.

For enterprise advertisers, the insights will only need to improve their bottom line by a small percent and the technology and analytics services will pay for themselves. Much stronger results are necessary for smaller advertisers to justify the effort and expense.

Without giving too much away, what are the things you’d first look to get right when developing an attribution model?

TVS: As mentioned, everything starts with being able to track the consumer’s journey across devices and eliminating non-human traffic and impressions, so an advanced tracking solution is a must. From what we have seen, though, this is not the biggest challenge.

To get true value from attribution you need buy-in from each of your channel managers, and that can only happen if the company agrees on what their ‘source of truth’ is, and rewards their marketing team on overall results versus channel results.

We can’t expect to gain full benefit from such a holistic technology with the antiquated methods of managing our budgets. The old “use it or lose it” mentality won’t work. Channel managers need to be willing to hand over the underperforming dollars in their budget to another channel manager who can return a higher ROI – and the manager handing over his budget should be rewarded for doing so. Collaboration vs. competition – extract the synergy from the marketing team and get them excited to work together for the bigger picture.

Once you’ve got the data, governance model and collaboration, you can begin to consider your model. Ultimately, attribution modelling helps cross-channel managers to shift budgets from one channel to another. And within each channel, it helps managers to shift their spend among their media buy points, down to the ad- or keyword level.

We always recommend to run a few models in parallel to compare insights. Make shifts to the marketing spend based on the chosen model’s results and notice whether it gives you the outcome you desire. The more comfortable the team becomes with how the data is telling them to shift their spend, the more money they can inject into each test, until they are using the model completely for every dollar spent.

We won’t be surprised to hear the validity of certain models being brought into question during PMI: Europe. What do you make of the general consensus around last click in 2016?

TVS: Every [human-driven] model has its flaws. The traditional last touch model is just more flawed than most, because it only looks at a single data point and can easily be manipulated or skewed. With the multi-touch models mentioned earlier – linear, position-based, time decay – you’re addressing every engagement a customer has with the brand’s advertising – not just the most recent.

Let’s use an oversimplified football scenario as an analogy. You have 11 players on the field – 22 in total. The goalkeeper kicks the ball more than halfway across the field to a team member. The midfielder gains a quick eight yards by passing it very quickly to the next player, that player gains another 10 yards by passing it onto the next, and the next. The striker scores the goal from just outside the box. Should that last kick get all the credit? Of course not. Every team member that touched the ball played a part in that win, and probably some players that didn’t even touch the ball. Attribution modelling helps you uncover how much each player contributed to that win.

What’s the most viable alternative, in terms of recognising influence, that could be widely adopted in the affiliate industry?

TVS: The most viable model will ultimately depend advertiser to advertiser. The end goal is not to measure the influence of one channel, rather to understand the influence and collaborative impact of all channels holistically.

A data-driven [algorithmic] model will suit advertisers that are ready to adopt alternative KPIs for their teams and transition to such a model. Like the common multi-touch models, the algorithm will consider all media involved in a consumer journey, as well as external factors such as seasonality, promotions, macroeconomic effects and the saturation effect, or law of diminishing returns. If an advertiser is not ready to move to an algorithmic model for whatever reason, then the importance of each channel and tactic becomes significant in selecting a rules-based model that best fits the business.