Find the other parts in this series here:
- Part 1: The state of attribution in 2023 (this post)
- Part 2: How we do attribution
- Part 3: Attribution outputs & actions
Navigating the Multi-Touch Attribution Landscape: From Cookies to a Cookieless Future
In the ever-evolving world of digital marketing, attributing customer actions to specific marketing touchpoints has become increasingly complex.
Multi-touch attribution (MTA) emerged as a sophisticated approach to overcome the limitations of last-click attribution, providing a more accurate understanding of the customer journey and the effectiveness of various marketing channels. However, the impending demise of third-party cookies has thrown MTA into jeopardy, necessitating new strategies and methodologies to accurately track and measure marketing performance.
The cookie-driven era of MTA relied heavily on persistent identifiers to track user behaviour across websites and devices. Cookies allowed marketers to create detailed customer profiles, enabling them to identify and attribute conversions to specific touchpoints. While this approach provided valuable insights, it also raised privacy concerns, leading to stricter data regulations and the eventual phaseout of third-party cookies.
In the absence of cookies, MTA is undergoing a paradigm shift, focusing on privacy-preserving methods and leveraging first-party data. Marketers are exploring identity solutions that balance user privacy with effective attribution, such as contextual targeting, probabilistic matching, and machine learning algorithms. These methods rely on contextual signals, anonymised user data, and predictive models to infer customer journeys and attribute conversions to different marketing touchpoints.
Despite the challenges, the cookieless era presents an opportunity to refine MTA practices and focus on more holistic marketing measurement. Marketers can now delve deeper into customer interactions, understanding the nuanced relationships between touchpoints and their collective impact on conversions. This shift towards understanding the overall customer journey rather than relying solely on click-based attribution will lead to more effective marketing strategies and optimised resource allocation.
The future of MTA lies in a hybrid approach that combines privacy-preserving technologies with advanced data analytics. As marketers adapt to the evolving digital landscape, MTA will continue to play a crucial role in understanding customer behaviour, measuring marketing effectiveness, and optimising campaigns for success.
So, what can we do?
The answer to these problems is, al often, to combine a few different techniques and tools. Analysts (or “Marketing Economists” as Eric Seufert calls them) can effectively measure marketing performance and optimise their strategies for success in the cookieless era:
- First-party data-driven attribution: Companies are increasingly relying on their own first-party data, such as customer email addresses, website browsing behaviour, and CRM data, to gain insights into customer behaviour and attribute conversions. This approach offers greater control over data privacy and enables more personalised marketing strategies.
- Contextual targeting and attribution: Contextual targeting involves delivering ads to users based on the content and context of the webpage or app they are viewing, rather than tracking their individual behaviour across the web. This privacy-first approach allows marketers to attribute conversions based on the relevance of the ad placement.
- Probabilistic matching and predictive modelling: Probabilistic matching utilises statistical techniques to match anonymised user data to known identities, enabling marketers to attribute conversions to specific advertising campaigns without using persistent identifiers. Machine learning algorithms can further enhance attribution by analysing large datasets of user behaviour to identify patterns and trends, providing more accurate attribution insights.
- Multi-source attribution: This approach combines data from multiple sources, such as website analytics, email marketing, social media analytics, and CRM data, to provide a more comprehensive understanding of the customer journey and attribute conversions across various touchpoints.
- Cookieless identity solutions: Several companies are developing cookieless identity solutions that aim to balance user privacy with effective attribution. These solutions typically use a combination of contextual targeting, probabilistic matching, and machine learning algorithms to create unique identifiers for users without compromising their privacy.
- Attribution modelling: Attribution modelling involves assigning weights to different touchpoints based on their perceived contribution to the conversion. Various attribution models exist, such as last-click attribution, first-click attribution, and W-shaped attribution, each with its own strengths and limitations.
How do we do this at Tasman?
At Tasman, we work with early stage company to get to a valuable data platform as quickly as possible. That means we make trade-offs in platform complexity versus the amount of actionable insight we can get out of it. The most suitable attribution approach for your company will depend on several factors, including your industry, growth strategy, and the type of data you have been collecting. We assess these factors and recommend the most effective approach(es) for the specific needs.
By working closely with you, we can tailor an attribution strategy that aligns with your unique business goals and unlocks valuable insights that may have been missing from you. Our expertise in attribution modelling, data analytics, and privacy-preserving methods will ensure that you gain a comprehensive understanding of your customer journey and optimise your marketing campaigns for maximum impact.
For instance, for consumer goods (and retail) companies we would always look at first-party data-driven attribtion first. We’d leverage customer loyalty programs, purchase history, and website analytics to gain insights into customer behaviours and attribute conversions. Simple example: get people to self-identify early on your website by giving them a discount code in return for signing up to a newsletter (and then using that code at checkout, even if it is a different session).
In service companies, conversion cycls are typically more complex and we need to get more creative. We’d look at using multi-source attribution primarily — combining data from CRM systems, email marketing, social media analytics, and call / customer support tracking to understand the multi-touch customer journey. This is also where it becomes very clear that attribution is a business analysis exercise — not exact science! Different attribution pathways will give you different results; context here is vital.
For B2B Companies the complexity of the conversion chain is so large that we have no other option than to utilise more Probabilistic matching and predictive modelling techniques. We’d use anonymised user data and machine learning algorithms to accurately attribute conversions in complex B2B sales cycles. This is hard but with a centralised data warehouse, it can be done.
Lastly, when we work with App Companies we are very much dependent on the state of the App Store ecosystem. That is quite complex, and with various degrees of app tracking limitations, we have to approach this leveraging a combination of the approaches above. E.g. we’d use first party data from the app itself (so tracking must be really good) as well as a statistical models to understand the impact of media sources (MMMs), as well as leveraging 3rd party data which has been already modelled from MMPs to create a full view.
In all, attribution nowadays is a lot more complex than it used to be. A great data platform is an indispensable asset here.