top of page

Forecasting a Better Future: Why Curves, Transparency, and Trust Matter

by John Wallace

July 19, 2023

“Because of our confidence in the model and the experiments, we have become the trusted conduit for optimizing marketing spend across the organization. “ – Devin Griswold, THRIVE


I want to tip my hat to Devin Griswold at Thrive - the relationship between optimizing growth and trusted analytics is critical. The most common message we hear from marketers is that while they have access to loads of performance data, they lack trust in using it to forecast and optimize performance. Making great decisions with confidence requires highly relevant, trusted data. However, recent IAB research suggests that while marketers acknowledge the need to update their measurement strategy, 66% are not taking any meaningful action to do so. Trust is missing.


The journey we are taking our customers on is the leap from Measurement to Marketing Effectiveness.

Measurement

Marketing Effectiveness

Key Business Question

"How did my media perform last week?"

"How do I grow revenue and profit next week?"

Deliverable

A scorecard

A forecast

Analytics Confidence

Better than last click

Trusted to forecast and optimize

Outcome

Sometimes a better understanding of past performance

Grow revenue and profit

Our design philosophy for the LiftLab platform is simple: to deliver a system that helps grow revenue and profit for our clients. We knew that to accomplish this, our analytics needed to shift the focus from arguing about measurement to using trusted analytics to forecast growth. Trusted analytics are essential to encourage customer adoption of AI-driven forecasting.

Curves, Not Lines: Forecasting Performance Requires Stronger Analytics

Marketing operates according to the law of diminishing returns, as taught in Econ 101. These are curves, not lines. Incrementality represents just a single point on the curve. Relying solely on incrementality leads to significant information loss, making meaningful forecasting challenging. We built a system that leverages the full diminishing return curve to understand elasticities for every spend level across the media plan. This approach provides a more complete understanding of how spending affects outcomes, enabling our clients to make more accurate forecasts and plans.

Combining Sciences for a Trusted Approach

Forecasting requires data you can trust; a measurement approach that's just "better than last click" will fall short. At LiftLab, we built a system that blends modeling and experimentation. We took a holistic approach that fosters organizational trust through built-in model validation and the ability to quickly understand the performance of new channels and tactics. Our Agile Mix Model (AMM) provides coverage across the media plan. By combining modeling with the science of randomized control testing through specialized media experiments, we enable our customers to promptly forecast performance for new channels and tactics.


Experiments validate and improve the AMM quality by introducing new performance data from experiments solving two age-old problems with any modeling approach, including Marketing Mix Modeling (MMM). First, what do I do when I don’t have enough data yet (often called the cold start problem)? Second, how do I circumvent the garbage-in/garbage-out? Better data yields better models, and our blended approach provides data that pure MMM approaches typically lack.


Transparency Drives Trust

We have always been more open about our methodology compared to our peers. Transparency is embedded in our data science DNA. We realized we could go further by providing every customer with a robust under-the-hood view of our analytics and are driven by the question, "What our data science team would want to see if they were our clients?" Our robust 'under-the-hood' view of our analytics is provided to every customer. And, to take it one step further, we are piloting tools to enable them to integrate new experimental data, refresh models, compare performance across models, and promote new models to "production" themselves. Ultimately, we believe that some data scientists will want to model, and we embrace that too!


We deliver Marketing Effectiveness so our customers can forecast performance at different spend levels to grow revenue and profit. Job 1 is to establish a trusted analytics foundation. More robust analytics, a holistic approach that combines sciences, and a new standard for transparency are the building blocks of this trusted foundation.


And this trust gives customers, like Devin Griswold of THRIVE, the ability to forecast with confidence.


Happy to jump into this further. Feel free to contact us, and we'll walk you through it.









Kommentarer


bottom of page