MTA: Marketers Talk Attribution – Feb. 6, 2019 – Presented by: The ARF Cross-Platform Measurement Council
It’s one thing when marketers talk about doing multi-touch attribution (MTA). It’s another when they share what leads to – or detracts from – successful implementation.
AIG, Coca Cola, HSBC, Johnson & Johnson and Miller Coors discussed working within their organizations to generate and apply actionable marketing attribution results. MTA expert Alice Sylvester of Sequent Partners will lead them in discussion, focused on how they have evolved people, processes and technology to achieve their goals.
Following up on ARF’s Cross-Platform Measurement Council Attribution Working Group’s Attribution Showcase in June 2018, conversation covered relevant topics, such as:
• Motivating their organizations to take on a complex analytical solution like MTA
• Driving adoption of MTA within their organizations
• Integrating the best of what has been working with new thinking
• Designing and activating new workflows
• What it takes to break down legacy silos
• Developing new systems to implement MTA results
• Evolving processes, based on learning
Alice Sylvester (Moderator) – Partner, Sequent Partners
Abha Dawesar – Head of Marketing Strategy, Global Banking & Markets Americas, HSBC
Newcombe Clark – Global Director, Rapid Learning Lab, AIG
Alice Sylvester (Moderator) – Partner, Sequent Partners
Greg Pharo – Global Director of Advertising Research & Media Analytics, The Coca-Cola Company
Igor Levin – Senior Director of Digital and Personalized Marketing Analytics, Johnson & Johnson
Bill Cramblit – Media Analytics and Optimization Manager, MillerCoors
READ THE KEY TAKEAWAYS
The panelists shared many common views on MTA, despite dealing with different situations —such as more regulatory oversight and first party data availability for financial vs. CPG companies — and dissimilar corporate needs, cultures and structures. Highlights of their comments by topic area include:
How they described multi-touch attribution
- For these marketers, multi-touch attribution extends beyond digital media to include offline media, too.
- MTA may involve non-media advertising elements such as creative executions or non-advertising factors like branch locations for a bank.
- Although some panelists considered MTA more forward-looking vs. marketing mix modeling (MMM), which they considered more backward-looking, they saw benefits to both.
Using MTA alongside other types of models
- MMM is still being used alongside MTA, e.g., MMM is often used for longer term, strategic questions. However, some believed that, as TV begins to offer more granular data (e.g., via addressable TV) and with the push to personalized marketing, MMM may become less important.
- MTA is seen as faster and offering greater granularity than MMM, allowing for more tactical analyses, e.g., in-flight optimization of specific media.
- There is interest in having single, unified models that can integrate marketplace data (i.e., MMM-like) and individual (i.e., MTA-like) data, but development of such models is in the early stages.
- Companies should start small, e.g., looking only at small problems or beginning with a limited number of brands. This will provide helpful lessons to use in expanding usage.
- Plan for a long internal sell-in process. There may be less interest in MTA if business is good, more interest if management has heard about MTA in the marketplace.
- MTA may result in new workflows that affect team interactions. It requires cooperation from all stakeholders, e.g., brand marketing, finance, legal, sales, and even HR.
In-house vs. working with outsider modelers
- It is not unusual for CPG companies to work with outside suppliers in addition to in-house experts. They cited these advantages:
- Outside vendors can help ease the burden on an in-house group, thereby lessening the group’s need to expand, which might raise internal staffing concerns.
- Outside vendors can provide companies with data they would not have otherwise.
Still, having some services in-house can make it easier to engage the range of stakeholders who may bring a valuable understanding of marketplace issues and customers.
- The financial services companies have so far avoided working with external vendors, though one company is exploring the possibility of working with an outside company it “can trust.” Their reasons for keeping MTA in-house include:
- Concerns about privacy of their customer data
- Ability to answer ongoing questions —as much as a year later after an initial analysis
- Opportunity to leverage internal resources to “better bring truth to our own power”
Best practices/lessons learned
- With so much available data and many factors that can be analyzed, it is important to determine what is important and to stay focused on the most critical elements.
- Attribution must relate to business needs; it is easy to get enamored of the analytical capabilities, but that will not be productive.
- Be sure you share the same understanding as others, for example:
- Is someone really referencing MTA — or something else like last touch attribution?
- What are the key media/consumer touch points?
- What outcomes/success metrics should be tracked?
- Is the goal to assess a unified, holistic approach or something more tactical or limited, e.g., assessing only one media type?
- Success relies on working well with different groups across the organization and keeping them focused on the goal of the analyses.
- There are a number of ways to set up attribution. More than one panelist cited the advantages of looking at test vs. control analyses.
- Try to use the most granular data possible, recognizing what is available. CPG brands, for example, may have to use household data vs. personal data.
- Don’t fall into the trap of looking at each analysis as being a success or a failure. You can learn from both. The key is to keep increasing your understanding of how “things work” to make better decisions.
- Don’t rush to get results. It is important to bring in comprehensive, high quality data.
- Use legal support to consider what data inputs will – or will not – trigger privacy concerns.
- New workflows that require more automated processes may bring in unanticipated hurdles.
- Walled gardens may not release data that would help with attribution.
- Some important data (such as non-addressable media data) cannot be incorporated into MTA. One possible solution is linking MTA and MMM, although doing that right isn’t easy.
- Programmatic efforts within digital can be tricky to incorporate.
- Brand equity cannot be easily accounted for within MTA models.
Areas of interest going forward
- How might the industry develop and adopt standards for attribution and share more examples of best practices?
- How can data from walled gardens be more readily available? What services might be able to work with them to access the data?
- How can current experimentation with AI in the models be expanded? An example: can you predict how an ad developed in one part of the world will perform in a different region?
- How will attribution spread to other markets, if you are a global company?
- What impact will new forms of TV have on marketing?
- How might MTA models assess creative impact beyond limited A/B comparisons?