Growth Frameworks

Data Trends: The Evolution of Marketing Attribution

A focus on data privacy and consumer identity – including an increasing discomfort with consumer tracking and the advent of GDPR and other privacy legislation – has changed the playing field in marketing. Your approach to marketing attribution needs to change, and change quickly.

Marketing attribution models are designed to offer a data-driven way to tie spend to the conversion events – and, ultimately, the revenue dollars – that result from your marketing activities. Many companies develop a holistic attribution model that attempts to capture all of the touchpoints leading up to a conversion by including activity across all marketing channels. A holisitic model requires significant data investment – and typically includes data acquisition, cleaning and storage; statistical modeling; and serving the output into reports, dashboards and sometimes back into ad networks for campaign optimization.

‍

For many years, the trend in marketing attribution has moved towards an ever-increasing effort to identify and track customers and prospective customers. Advertisers and advertising vendors alike tapped into cookies, pixels, mobile device identifiers and location data all in an effort to tie together browsing and shopping behavior across sites, devices, and even across the online / offline divide. This approach to attribution has all begun to unravel, however, with increasing consumer awareness of and discomfort with being tracked, the advent of GDPR and other privacy legislation and, more recently, changes by Apple designed to give individuals the ability to restrict how and when their digital activity is tracked.

‍

However, despite these macro challenges, your teams still need to understand the ROI on your marketing spend. So how can your data team help?

Take a step back:

Take a step back from the team’s current development or maintenance of attribution models. Talk to the marketing team and listen for both their pain points and strategic goals. How does attribution fit into both their tactical decision-making and longer-term planning? Click-based attribution models have long under-valued offline channels and brand awareness campaigns. Branded keywords deliver clicks and sales, but how do consumers find the brand in the first place? Many marketers have embraced or are experimenting with influencer marketing, another channel that doesn’t fit neatly into attribution models. The real underlying needs of your marketing team may necessitate a new approach from your data team. The goal of data efforts should be to provide useful tools for marketers, and an unbiased lens on their performance.

Diversify your modeling approaches:

Click-based attribution is satisfyingly deterministic, but its apparent precision can obscure other shortcomings. You should also consider media mix modeling, which uses a regression model to find relationships between levels of channel spend and aggregate outcomes such as conversions or sales. This can be coupled with causal inference or Bayesian methods to overcome reservations around attributing causality from correlation. Another way to understand the customer journey prior to conversion is to simply ask your customers in post-purchase or other surveys. In our experience, this type of qualitative input has biases, since motivated customers are more likely to see and respond than are less motivated ones, but it can nonetheless be surprisingly informative when collected from a large enough sample and if trended over time. Adding these additional methods to triangulate marketing impact may require adding new skills to the team. An experienced marketing analyst with a strong statistical foundation can really super-power your marketers.

Embrace the power of experimentation:

Any data-driven culture is grounded in experimentation: testing changes and letting the data determine which option is best. Moving forward, we believe experimentation will play an even more important role. In our experience, it is critical for data leaders and teams to consult closely with marketing teams on design, analysis and interpretation of results. Reasonable controls are essential to determine whether a conversion is incremental or was likely to occur anyway.

Despite – or perhaps because of – all of the changes to tracking and privacy requirements, the need to understand the ROI on your marketing dollars has never been greater. In our view, these changes present an opportunity for innovation. As you prepare your team to tackle marketing attribution challenges in the year ahead, we encourage you to take a step back, diversify your attribution modeling approaches and embrace the power of data-driven experimentation.

‍

Growth Timeline

No items found.

Don't delete this element! Use it to style the player! :)

Cae Keys

Truemuzic

https://interests.summitpartners.com/assets/DHCP_EP9_FutureHealthCare_DarrenBlack-2.mp3

About the Authors

Cathy Tanimura

Cathy Tanimura is Vice President of Analytics & Data Science at Summit Partners. She works closely with the Summit team and our portfolio companies to help create and execute strategies that effectively use data and analytics to drive better decisions and build better products. She is the author of SQL for Data Analysis: Advanced Techniques for Transforming Data into Insights. Prior to Summit, Cathy worked at Strava, where she built an analytics and data science team focused on product, marketing and business development. Previously, she led analytics teams at Okta and Zynga.