The end of attribution
Sep 1, 2023
Introduction
Touch-based attribution is a useful tool for digital marketers. It lets them see how customers engage with and flow through digital channels. But, after two decades, its time to acknowledge that touch-based attribution isn't a way to answer the marketing ROI question.
In this article, we'll explore the history and limitations of touch-based attribution. We'll start by looking at different types of touch-based attribution models, the inherent limitations in these models, the implications of privacy changes, and most importantly, how to distinguish between attribution and incrementality.
We'll also discuss why it's important for marketers to be aware of these pitfalls and how to incorporate touch-based attribution into your measurement process.
The history of attribution
In its simplest form, touch-based attribution assigns credit to digital marketing channels for a customer's journey. This means marketers can quantify the impact of each digital touchpoint on the path to conversion or other desired outcomes.
Touch-based attribution has been around since digital advertising platforms emerged as an alternative to offline ads in the early 2000s.
By now, three different touch-based attribution models have emerged:
First-touch - attribute credit solely to the initial digital interaction
Last-touch - attribute credit solely to the final digital interaction before conversion
Multi-touch - attempt to measure the contribution of each digital channel across an entire sales funnel but end up being a mess of inaccurate estimates
First-touch and last-touch attribution models have limitations because they don't capture all touch points on the customer journey. Some vendors introduced multi-touch attribution models, which they claim provide a more holistic view of the sales funnel. In the last few years, new variations of multi-touch attribution have come to market.
Time decay models are similar to regular multi-touch models but assign a weighting that decreases over time as the customer's journey progresses.
U-shaped and W-shaped attribution models emphasize the first and last touchpoints, respectively, to give credit where it's due for influencing a customer's decisions.
Custom attribution — some software vendors promise models to match the specific needs and objectives of customers.
ML-based models — some vendors have used Markov Chains and Logistic Regression on top of these multi-touch models.
Despite advancements in this area, the traditional strategies used by touch-based attribution vendors are completely broken. Continue reading to uncover why.
Why is attribution dying?
There are several reasons that lead us to believe that the use of touch-based attribution to measure return on marketing investment is dying:
Touch-based attribution is structurally biased against indirect response digital channels.
Touch-based attribution, regardless of the flavor, can't handle offline channels.
Touch-based attribution is becoming even worse due to privacy changes.
Touch-based attribution requires unnecessary data engineering and collection.
Finally, touch-based attribution doesn't address the key question on every marketing (and business) leader's mind — incrementality.
Let's take a look at each of these issues.
1. It's biased against some media
Direct response marketing is a type of marketing strategy where the goal is to encourage an immediate response from buyers. In contrast, indirect marketing attempts to build relationships with potential customers through various channels. Often, those indirect marketing activities lead to customers seeking out products and services through direct response channels or directly through the vendor's website or mobile app.
In digital marketing, direct response refers to channels like Google Search, Retail Ads, LinkedIn Conversation Ads, etc. Indirect response marketing can happen on several channels. Social and display ads. Organic social. Events. Partnerships. The list goes on.
Touch-based attribution can be somewhat effective at tracking the performance of direct response digital channels. However, it can't measure the impact of indirect channels. This limitation is substantial and discourages many marketers from exploring indirect response marketing, which is unfortunate because 95 percent of your buyers aren't currently in the market.
2. It can't measure offline media
Fast-growing businesses often seek to explore multiple channels to capitalize on growth opportunities. As digital channels become increasingly saturated, offline channels offer a viable alternative. However, except in rare cases, it's impossible for touch-based attribution to track offline channels like TV or radio! Even when special codes or URLs are promoted, buyers rarely follow them if and when they buy.
As the digital landscape continues to evolve, new channels such as Podcasts and Connected TV are providing opportunities for marketers, but they behave more like traditional offline channels. Attribution in these channels remains difficult, as they aren't designed for direct response, posing a challenge for measuring performance.
If offline channels are a significant part of your marketing strategy, you're likely relying on inaccurate touch-based attribution that doesn't give these channels the credit they deserve. Alternatively, you may be underinvesting in these channels due to the lack of measurement capabilities.
3. It's not privacy friendly
There are several privacy-related changes that make touch-based attribution worse.
The demise of the third-party cookie has had a major impact on companies and their software vendors, who rely on cross-site tracking to create touch-based attribution models. Chrome, Safari, Edge, and Firefox have all implemented policies to restrict using such cookies.
Apple recently announced that they are eliminating UTM codes from URLs in Mail, Messages, and Private Browsing, which will have a considerable impact on marketers who rely on UTM parameters to track clicks. It's likely that other browsers will follow suit soon.
The implementation of Apple's App Tracking Transparency changes in iOS 14.5, prohibiting apps from collecting data without user permission, and the recent Privacy Sandbox changes in Android, have rendered touch-based mobile attribution practically obsolete.
Recent years have seen a noticeable effect of privacy laws, such as the General Data Protection Regulation (GDPR) and the California Consumer Privacy Act (CCPA), on touch-based attribution. Companies must obtain consent before collecting any data related to customers, or face the risk of hefty fines and other penalties.
Privacy challenges combine with other pitfalls to make it difficult to obtain actionable insights from touch-based attribution. To account for errors, some marketing teams have implemented hacks on top of touch-based attribution. These hacks involve adding multipliers to their touch-based attribution models for each channel, based on either intuition or, in rare cases, experiment data.
4. It requires unnecessary effort
Many teams become disorganized after a few years of following this course of action. They begin with first-touch or last-touch attribution, then progress to multi-touch. They require their teams to manually tag every campaign with the appropriate UTM codes, a process that's prone to numerous human mistakes.
Finally, they attempt to reconcile the discrepancies between their touch-based attribution vendor's figures, those provided by UTM code tracking, and their own intuition. Chasing a false sense of accuracy, they apply multipliers to each channel to adjust the model for indirect and offline channels.
Is the effort worth the reward? Rarely. CMOs often express doubts about the accuracy of their attribution models.
5. It measures the wrong thing
Rather than tracking individual touches, company leaders care more about understanding how much incremental value marketing drives, both in total and through each channel. When conducting research with marketing and finance executives, we concluded that the real issue wasn't attribution, but rather, incrementality. This was true across Technology, Ecommerce & Retail, SaaS, and Consumer categories.
Touch-based attribution is a simplistic way to map the customer journey based on touches that are detectable, regardless of the causal relationship between a touch and a subsequent conversion.
Incrementality gets at the causal effect of marketing's impact on sales. Said another way, it answers the following question — what's the loss in sales when marketing investments are reduced to zero? It looks beyond the digital user journey and considers a broader set of factors.
Marketing leaders are being asked to demonstrate incremental value. Using touch-based attribution to answer this question is setting the industry up for failure.
So, what are marketers supposed to do?
Touch-based attribution is ridden with errors when used to measure return on marketing investments. Both in the underlying assumptions used to promote this methodology and in the incomplete and error-prone data collection.
Marketers have known forever that "clicks" aren't the end-all be all. Not by a long shot. Many marketing channels and tactics influence buyers without ever generating a click. But, they've just not had an alternative measurement methodology.
We get that. We've been in your shoes. Regardless of where you are in your journey with attribution and incrementality, the sooner you acknowledge the gaps and incorporate alternative methodologies, the better.
We've written a detailed guide on what to do instead. Read it here!
Solutions