"On TV And Video" is a column exploring opportunities and challenges in advanced TV and video.
Today’s column is written by Matt Hultgren, vice president of analytics at Marketing Architects.
Complicated, expensive, unreliable and frustrating.
I have heard CMOs and marketing executives use these adjectives – and a few other unprintable ones – to describe their experiences with multichannel attribution.
To most, it’s a time-consuming exercise that’s filled with shaky data and unexplainable math, conducted using black-box methods that rival the CIA’s code of secrecy. Mere mortals have little chance of understanding its details.
My honest answer, when I hear these stories?
All attribution is wrong, but marketers need to do it anyway.
The ‘silver bullet’ does not exist
This may sound counterintuitive. Why pour time, effort and budget into a flawed application?
The answer gets to the central issue of attribution: the data model. All models have flaws, from missing data to incorrect assumptions. There is no silver bullet or single, perfect tool to measure each and every influence, action or result that a broadcast ad generates. Period.
Once marketing executives accept this fact, it’s possible to move forward with multichannel measurement programs that actually work. It’s possible to make decisions based on analytics where marketers are highly confident.
First, marketers must be be clear about what their multichannel attribution model can and can’t deliver. Who doesn’t want to link every customer interaction – online and offline – to a quantifiable sale? It’s the holy grail of marketing measurement. But a 360-degree view requires immense discipline across the enterprise to capture information completely and govern data for accuracy; most organizations are years away from this goal.
What a multichannel attribution model can deliver are actionable insights to help marketers hone their media budget, message, offer and brand strategy. With broadcast advertising, for example, marketers can expect their attribution model to show them:
- Which dayparts generate the strongest response
- Which stations and markets are most effective
- Which creative packs the biggest punch
Second, marketers must know what is and isn’t measurable. Television and other top-of-the-funnel activities create positive influences that optimize a brand in the minutes immediately following an ad, as well as in the days and weeks that follow. Tracking immediate actions, such as Google searches, phone calls and web orders, is relatively straightforward. Measuring improved brand awareness, consumer preference and word of mouth is harder.
Use multiple models to boost the trust factor
Lastly, marketers must tune up their data model. Whether they create their own or work with an agency or attribution partners, they must ensure the model has flexibility to accommodate the unique data points of their company, product and campaign. In other words, avoid one-size-fits-all models that take a cookie-cutter approach.
Whenever possible, marketers should run at least two attribution models side by side. In this way, they can inject more confidence into the analytics and insights that they receive. For example, when running an internal attribution test, it may be helpful to have an outside attribution firm simultaneously model the same data. When both models reveal similar results – using two slightly different methodologies – marketers would know they’re heading in the right direction.
Your attribution is wrong. But please, do it anyway.