Measuring Marketing Impact Beyond Your DTC Website

Quantifying the “halo effect” in marketing

The customer journey is no longer linear.

With the rise of marketplaces, the awareness you build can now lead to conversions on your own direct-to-consumer (DTC) site, Amazon, TikTok Shop or other retailer sites - often all split across multiple devices.

This behaviour is known as the halo effect (also known as the spillover effect) - when marketing in one channel creates positive effects in a completely separate channel. In many ways, it's a net positive for your bottom line: with more options for customers to find a marketplace that suits them, gross conversions are often up as a result.

However, it creates a major challenge for marketers trying to understand the performance of their activity because traditional attribution models only measure conversions across DTC and other owned properties. Measuring only .com conversions can lead to massive under-reporting of Return on Ad Spend (ROAS). For brands with significant marketplace presence, 40% or more of true incremental sales may be missed in last-click reporting.

This impact is felt worst by upper-funnel campaigns that are already under-attributed in many measurement approaches, potentially leading to budget optimizations based on flawed data.

So, how do you tackle this massive visibility gap?

The answer is shifting measurement away from isolated website tracking and toward a single, holistic metric that captures business impact across all retail endpoints: Unified ROAS (uROAS).

‍

Achieving Unified ROAS (uROAS)

Since you can't track an individual user from a Meta ad to an Amazon purchase, MTA-based solutions won't work for achieving uROAS.

Instead, you need statistical or causal measurement approaches that can capture platform-to-platform effects.

Statistical — Media Mix Modeling (MMM) is the ideal tool for strategic uROAS measurement because it uses aggregated datasets that aren't affected by pixel or cookie limitations, and delivers cross-channel uROAS estimates that also factor in external factors like marketplace sales.

Incrementality Tests (Causal) — For more granular, campaign-level measurement, incrementality tests provide direct causal proof of how activity on one platform lifts outcomes on another. A common approach is Geo-Lift testing, where you run campaigns in specific regions and measure the resulting lift in marketplace sales in those areas.

Either of these approaches can work, but if you've read our previous lessons on the pros and cons of different measurement tools, you'll know that traditionally they both suffer from slow reporting cadences that make them unsuitable for day-to-day optimizations.

This leads to a problem we've returned to multiple times across the lesson content -inconsistent outputs between tools and the need for a single source of truth. The halo effect is measured through an MMM or incrementality test, but day-to-day optimization is built from MTA insights that are completely blind to cross-platform effects.

In practice, this need for a frequently updated, single source of truth is driving a fundamental shift in the design and use of MMMs.

‍

Why marketplace measurement is the next frontier in Media Mix Modeling (MMM)

The most advanced MMMs are now designed to do more than measure your total business - they specifically model the halo effect between channels. This ensures budget is allocated based on total, unified business value, not just owned website performance.

This shift is elevating incrementality testing from a useful audit to a mission-critical calibration input for MMMs. Marketers now use test results to calibrate the MMM itself, ensuring the long-term model learns from and adapts to short-term halo effect measurements.

A core innovation enabling this change is the rise of MMMs built on Bayesian statistics.

We've covered Bayesian MMMs in detail in previous lessons - they're a subject dear to our hearts at Fospha - but they're worth revisiting from the perspective of halo effect measurement.

They can measure the incremental halo effect — Traditional MMMs are excellent at finding correlations (e.g., "When our Amazon sales go up, so does our Facebook spend"), but Bayesian MMMs are significantly better at isolating and measuring causal relationships (e.g., "An increase in Facebook spend causes a 4% lift in Amazon sales"). They achieve this by incorporating incrementality test results (like Geo-Lift data) as "informative strong priors." This means the model starts with the proven causal truth of the halo effect from an incrementality test before running any calculations, ensuring its ultimate forecast for uROAS is built on confirmed fact, not just historical patterns.

They remove the measurement lag — The biggest drawback of using incrementality tests or traditional MMMs for measuring the halo effect is slow reporting. While a traditional MMM can be calibrated using halo effect findings from a test, it typically requires a complete re-run (often quarterly) to incorporate that finding, at which point the test results may no longer be valid. Bayesian MMMs solve this because they don't require a full rebuild. Instead, test results can be instantly incorporated to guide the model's learning process while the results remain valid, making them suitable for daily, agile reporting.

They combine data for strategic breadth — By combining the depth of causal testing with the breadth of aggregated data, Bayesian MMM provides the intelligence needed for omnichannel strategy. The model sees the entire business: all channels, all marketplaces, and all external factors, providing the holistic view needed to define uROAS grounded in causal truth through test results.

In essence, Bayesian MMM turns the slow, expensive process of testing the halo effect into a continuous feedback loop, making every test a permanent, high-value upgrade to your measurement system.

‍

Summing up

In the marketplace era, the halo effect is the biggest gap in attribution, and Unified ROAS (uROAS) is the metric needed to fill it.

Failing to measure it can entrench you further in bottom-of-funnel bias and lead to strategic budgeting errors - basing decisions on only a fraction of true performance.

Incrementality tests and traditional MMMs have roles to play, but modern MMM designs across the industry are shifting to address this need. At Fospha, we believe Bayesian MMMs offer the best approach to halo effect measurement: frequent, broad, and grounded in causal truth.

In the final lesson of this series, we'll zero in on the other innovations and industry trends you need to know to stay ahead of the curve and future-proof your measurement.