Multi-touch attribution (MTA) has been the marketing measurement holy grail to performance marketers for years, promising to unravel the intricate web of consumer interactions and determine each touchpoint's genuine value. MTA is often messier than promised.
MTA models are complex animals that require expertise. Implementing and interpreting them requires a dedicated analyst (or team) with strong statistics and data science skills. It requires a lot of resources, making it unattainable for small businesses.
Elusive Accuracy: Expert implementation doesn't make MTA a crystal ball. It relies on data, which is fragmented and incomplete in today's privacy-focused environment. Apple's privacy rules, GDPR, and California's data protection restrictions limit information, requiring MTA models to estimate. This "modelling" can account for 60% of the data, causing errors.
Walled Gardens and Browser Limits: Open platforms like Chrome allow MTA to track more thoroughly. However, Facebook, Google, and TikTok's walled gardens disrupt everything. These platforms closely protect their data, making cross-channel user interactions difficult to understand. Frequency capping and cross-platform optimisation are nearly impossible.
The world of MTA is full of models—first-click, last-click, linear, temporal decay, and more. Each has pros and cons, making it difficult to choose the perfect one for your organisation. Most models are rules-based and require human changes. MTA is slowly using AI, but human assistance is still needed.
Cost/Return: The expense of MTA may be its biggest issue. A complex MTA system is expensive to implement and operate, and the ROI is unclear. The costs may outweigh the benefits, especially for SMBs with restricted budgets. Redirecting MTA funds to other marketing may improve results.
Ironically, enclosed gardens provide good ecosystem attribution but hinder extensive MTA. Facebook, Google, and TikTok have extensive user data to accurately analyse campaign performance, but cross channel understanding of your audience is difficult.
The Future of MTA: Digital attribution has been present since about 2003, but AI and machine learning have not solved its problems. The future of MTA is doubtful because to rising privacy rules and gated gardens. Optimism is high in some AI circles to replace the rules based human, but it is still early days.
For small and mid-market businesses, simpler optimization models that focus on important touch-points and actionable insights may be the answer. Perhaps a new measurement method that accounts for the intricacies of the modern customer journey without compromising accuracy or cost is needed. What do you think?