Under the lid with Lifesight
October 2024
Welcome to our 5 part series on how causal attribution is shaping performance marketing.
Causal attribution is not just about crunching numbers but uncovering the "why" behind the data. Our Director of Performance Media, Fahmi Mohammed, teamed up with Lifesight's Co-Founder, Rajeev Nair, to explore how causal attribution can reveal the true impact of your marketing efforts. No more vague reports—now it's all about measuring incremental revenue and understanding the real effect of every decision, past, present, and future.
Part 1: Introduction to causal attribution
...
Traditional data modelling often misses the mark when it comes to understanding complex relationships. That’s where causal attribution steps in. Instead of just analysing past data, it digs into the ‘why’ behind it, revealing the real impact of marketing on outcomes like revenue. It’s like running a science experiment – using methods like marketing mix modelling (MMM) and controlled tests to uncover true cause-and-effect, not just correlations. This gives brands actionable insights, helping them measure incremental revenue and make smarter, data-driven decisions.
Part 2: The science of incrementality
...
In today’s data-driven world, causal attribution provides a robust framework for measuring marketing effectiveness by blending Marketing Mix Modelling (MMM) with experiments to give brands actionable insights.MMM is ideal when brands have at least two years of historical data, using machine learning algorithms to analyse how different marketing channels drive revenue, taking into account non-linear relationships and time lags.
For brands with limited data, running controlled experiments offers a quicker alternative, such as segmenting users by ad exposure or using geo-targeted methods, like ramping up Facebook ads in specific locations to measure impact. To improve accuracy, synthetic control methods create "synthetic" clusters of cities or user groups for comparison, minimising real-world variables. Ultimately, combining MMM with experimental approaches gives a comprehensive view of incrementality, balancing precision with flexibility, and helping brands optimise their marketing spend for maximum return.
Part 3: Uncovering the truth in attribution
...
In today’s data-driven marketing landscape, retailers face growing pressure to accurately measure the impact of their marketing efforts while optimising budgets to fuel business growth. Effective decision-making hinges on two critical metrics: incrementality, which identifies the true contribution of marketing activities to business outcomes, and marginal outcomes, which predict the value of additional investment in specific channels.
However, challenges arise from fragmented data and inconsistent measurement methods, particularly when comparing platforms like Facebook and TV. To address this, marketers need a unified framework that consolidates and analyses data across all channels, offering clarity and actionable insights. Such a system enables predictive forecasting and ensures that marketing spend is directed toward driving the highest returns, even in a continually evolving environment. Achieving this requires not just data but also robust tools to interpret it, empowering brands to navigate complexity and maintain agility in an increasingly specialised marketing ecosystem.
Part 4: Navigating the future of measurement with AI
...
Causal AI is revolutionising marketing analytics, moving beyond the limitations of traditional machine learning. By employing advanced models like Directed Acyclic Graphs (DAGs), brands can simulate "what if" scenarios—such as altering ad spend or adjusting pricing—to predict cascading effects across their ecosystem. This shift integrates marketing analytics with decision intelligence, enabling brands to make holistic, data-driven choices. While traditional multi-touch attribution assumes complete visibility of user journeys, it often falls short due to fragmented or untrackable touchpoints. Data-driven attribution addresses this by treating observed journeys as representative samples, applying projections to account for missing data. Together, these innovations are paving the way for more precise, actionable insights in marketing.
Part 5: Integrating causal factors
...
Causal AI is transforming how brands optimise marketing spend and predict outcomes, but it requires significant input of business domain knowledge to define relationships between causes and effects. With causal systems, brands can tackle three key questions:
How to allocate budgets to meet revenue targets efficiently, shifting spend from low-performing channels to maximise returns
Predicting the revenue impact of a given spend, helping validate confidence in the system; and (
Making granular decisions, such as scaling individual campaigns or ads, supported by metrics like incrementality and Marketing Mix Modelling (MMM).
Beyond straightforward attribution, causal AI also considers complex effects like synergy, halo impacts (e.g., Amazon ads boosting offline sales), and cannibalism between channels. By analysing interactions and effects across platforms, causal AI enables brands to achieve more precise optimisation and make decisions that account for both direct and indirect contributions to revenue.
