While signal resilience in the form of Meta’s Conversions API, Google's Enhanced Conversions, and server-to-server integration have become crucial for measurement within a single channel, following users across platforms is harder.
Marketers face the challenge of assessing the real impact of their ad spend, especially as privacy regulations continuously change.
With a perfect way to understand ad performance across channels gone, advertisers have three options:
- Continue collecting data and hope for a new universally accepted tracking method: ‘waiting for biscuits’
- Develop an identity-tracking system that adapts to regulatory changes
- Embrace probabilistic measurement: using machine-learning methods to measure ad performance, focusing on continual improvement
In 2019, Performics invested heavily in option three, developing privacy-first techniques to regain the valuable insights once taken for granted.
The opportunity for privacy-centric measurement
While the digital landscape prioritises privacy, the need for measurement that mirrors this presents an exciting opportunity for advertisers to rethink strategies. Several core principles define this new approach.
- Data diversity fuels smarter algorithms Instead of focusing on one data type, pulling in diverse sources helps algorithms learn faster, become smarter, and make better decisions.
- Experiments and testing as a benchmark Lift testing is a trusted way to measure how individual channels perform. Using lift tests, advertisers can see how each channel drives incremental growth and improve their media planning accordingly.
- Probabilistic trumps deterministic Maintaining a clear picture of performance across channels requires a lot of work. Many turn to probabilistic measurement instead of striving for a perfect view, using lift test insights to estimate performance. Combined with marketing mix modelling (MMM), we get a fuller view of campaign performance, which helps with smart media decisions.
The most common 'new' statistical approach is Regression-Based Attribution (RBA). This method uses machine learning to link media efforts with sales, assigning weights to each tactic. The great thing about RBA is that it doesn't rely on individual user data, making it privacy-friendly. But how do advertisers implement this methodology?
We developed and launched Publicis Regression Based Attribution (PRBA), an end-to-end tool to simplify data ingestion and automate analysis, providing accurate, privacy-first attribution across multiple channels. Inspired by and partially based on Meta Open Source Project Robyn, PRBA helps marketers assess each marketing touchpoint’s contribution fairly, meaning real-time performance measurement compliant with evolving privacy standards.
Measuring actual platform value requires a thoughtful approach. Before feeding data into the PRBA, we must make key decisions, including selecting the appropriate variables to analyse, assigning weight to each tactic, and ensuring proper model setup for accurate results.
Performics' collaboration with various advanced multi-channel digital advertisers highlights the effectiveness of PRBA, accurately measuring the value of not just overall digital platforms, but also two-to-four tactics within each. This level of detail goes beyond traditional MMM and into space Attribution historically occupies. While probabilistic measurement can’t completely replace a deterministic measurement solution, it strikes the balance between consistency and application.
“At Performics, we believe the future of digital media lies in performance orchestration. The agencies that will lead the way are those who truly understand the levers at their disposal and can expertly activate the insights within the various algorithm-driven platforms to maximise growth for their clients.
Publicis Regression Based Attribution (PRBA) provides advertisers with powerful tools and consistent measurement to understand the true impact of their digital investments. This consistency empowers smarter, data-driven decision-making, ensuring businesses can optimise strategies and allocate resources more effectively.” Sam Holt, CEO, Performics
Based on our extensive experience deploying PRBA for many of the largest UK advertisers, here’s what’s important to consider before starting:
Choose the right metrics for success
For each media channel, PRBA identifies the metrics that best reflect performance. For example, while a platform might drive thousands of clicks, we know the total influence on people seeing ads is far greater than the direct traffic. PRBA regularly determines that reach or even impressions offer a clearer view of a channel’s intended impact providing a comprehensive measure of each touchpoint.
Decoding cross-channel trends
PRBA uses available lift tests across channels to identify the ground truth and make the necessary adjustments for a more precise evaluation of each tactic's effectiveness. A good example is for large retail advertisers whose branded paid search is likely closely aligned to sales, making this extrication difficult. Using targeted lift tests, not only improves the model, but encourages evaluation of each tactic to drive a cross-channel measurement agenda.
Aligning PRBA with MMM
PRBA aims to complement rather than duplicate econometric analysis but must be in line with practical business data. In some cases, clients are already working with an old attribution read or an MMM that does not provide the actionable granularity needed for digital. Rather than create confusion, PRBA helps to accommodate other data sources, by aligning our overall numbers or accommodating the wealth of knowledge used by internal teams to improve models.
Turning privacy challenges into opportunities: from delivering results to application
After implementing PRBA, advertisers can access weekly cross-channel reporting within just eight weeks. This new reporting approach provides a clearer picture of each channel's contribution. As a result, advertisers can make informed investment decisions and have a roadmap for testing of misattributed tactics, improving campaign performance.
Privacy-focused measurement has its share of challenges but also creates exciting possibilities. It encourages the development of a broader range of skills and fosters creative problem-solving. As seen in our industry collaborations with Meta and other partners, overcoming these challenges creates more robust, impactful strategies that drive measurable success for advertisers.