Stay marketing-savvy and tech-savvy. Get the latest in martech by subscribing to MarTech Today.
Making complementary media strategies work: The recipe for cross-channel attribution
Your marketing efforts across various channels will have a greater impact when they are aligned with one another, and columnist Kevin O'Reilly discusses how attribution is the key to achieving this alignment.
While much has been said about complementary media strategies, the secret to making it truly work is cross-channel attribution. And to make cross-channel attribution successful, marketers must think about two things: 1) the technology used to make it happen; and 2) the role of business-critical elements (media, marketing and so on) and how they work together.
First, let’s look at multi-touch attribution (MTA) and marketing mix modeling (MMM) and what they can tell marketers.
MTA vs. MMM: What about both?
Recently, there’s been a lot of noise around cross-channel attribution, with vendors launching new products or features — whether it’s marketing mix modeling (MMM) on top of a multi-touch attribution (MTA) platform or linear TV attribution based on traditional econometric modeling.
MTA and MMM models are critical components to making cross-channel attribution happen, but there are pros and cons to each.
An MTA model answers the question, “What is the true contribution of a measurable marketing action?” It can tell you whether a specific action changed the outcome of a customer conversion and, if so, by how much.
Where MTA is constrained is in providing a full view of the customer journey, especially in relation to how offline impacts online.
An MMM model provides a longer-term view of the online and offline impact of marketing actions. It does this by looking at the statistical relationship between spend/aggregate actions and business outcomes (typically sales). For MMMs to be successful, an extensive time-series of granular data is needed — usually two to three years’ worth.
So, you can get the granular view from MTAs and the high-level picture from MMMs. Both models have value, and both have limitations.
This is why a growing number of marketers are adopting a unified approach. From the MMM, they get high-level recommendations based on a full view of the customer journey. They can then use that information to “steer” the weaknesses in the MTA and improve day-to-day performance.
How do business elements work together?
Now that you understand the technology needed to enable cross-channel attribution, you have to understand how certain business-critical components work together. This is essential to inform your cross-channel attribution strategy.
When I say “business-critical components,” I mean the marketing, media and base elements that drive a brand.
Component 1: Marketing
Attribution and marketing must work hand-in-hand rather than in silos; and all those involved have to agree on why attribution is being used. Is it to spend less budget and prioritize areas seeing the most success? Or to find the channel that works best for a brand?
This type of collaboration is imperative and will enable a brand to:
- Give credit where credit’s due. Perhaps you’ve scheduled a big marketing push for next quarter, which means you need to link it back to the attribution platform so that other touch points are not unfairly credited for an effect.
- Test hypotheses. Attribution can help confirm or nix strategies for more effective marketing. For example, a gym brand may want to see if it can increase response rates by buying spots weighted toward mornings (on the assumption that active people are out in the afternoon). A quick-service restaurant (QSR) might do the opposite, targeting early afternoon or evening when people start to think about dinner plans.
Component 2: Media
Yes, you already have an idea of the media channels that work for you, but it’s important to let your attribution platform inform your strategy — and as close to real-time as possible.
Many companies don’t look at attribution until after a campaign has run, and that is too late! Constant review and on-the-fly changes are what turn a good campaign into a great one.
Component 3: Base
The base is what you’re left with once you strip away external influencing factors, including seasonality, clashing spots, competitor activity or high volatility.
Make it all work together
OK, we’ve talked about the technology and the importance of understanding business-critical elements. Now we can move on to modeling — the way to make them work together.
Regressive modeling (Bayesian or traditional) forms the backbone of attribution. It looks for how changes in X (marketing, weather, seasonality, product shortages and so on) drive Y (sales). It uncovers the relationship between variables and shows how they impact business outcomes/sales.
When building a regressive model, it’s important to keep a few things in mind:
- Cadence. All Xs (your explanatory variables) need to be on the same cadence — whether it’s weekly, hourly or minutely. In practice, this means that the variable with the lowest cadence determines the overall cadence. For example, if your TV data is weekly, your digital traffic has to be weekly, too.
- Granularity. The same holds true for the granularity of data. If one data set is only available at a national level, it restrains the impact of the rest of the data to that same level of granularity (unless you start making some spurious assumptions).
Ideally, marketers want to see this “attributed” view close to the cadence and granularity in which they buy media or conduct marketing activities. In our view, that’s weekly — or daily, at the DMA (designated market area) level — for all marketing and media variables. For base variables, some extrapolations or normalizations may be necessary based on source data, but the marketer is viewing attribution in as close to real time as possible.
Once you’ve cracked the data and have been able to attribute values to each part of the customer journey, you can start to play around with complementary media strategies — increasing and decreasing time and budget on various channels, for example — until you start to see changes in results.
Think of it as a sliding scale. If you increase your TV budget, you might want to increase your social budget as well to take advantage of second-screening viewers. Perhaps you’ll reduce spend on OOH (out-of-home) to compensate.
These are generalizations because, in reality, it isn’t always that simple. But using attribution properly will provide the data behind what is and isn’t working and show how consumers behave cross-channel.
Of course, this isn’t the be-all-end-all solution. As consumer habits change, you’ll need to redress the balance between media channels in an agile manner to keep up. But if you’re able to understand and accurately attribute how these different channels intersect and, ultimately, impact your bottom line, it makes it far simpler to optimize on the fly and tweak accordingly.
Opinions expressed in this article are those of the guest author and not necessarily MarTech Today. Staff authors are listed here.