The methodology is based on Bayesian inference using hierarchical probabilistic models. It was built to address the limitations in the market of the existing MMM solutions relying on linear, logistic, general or autoregressions using the maximum likelihood method.

Blackwood Sevens modeling engine, HamiltonAI, builds probabilistic models relating the activities of advertisers with fluctuations in sales. Our models take all relevant data into account and model any interaction that is relevant, in order to understand what drives sales. In addition, the models quantify the uncertainty in their learnings which enables them to handle data from a large variety of sources with different quality levels.

Learn more here.

We do not just provide a classic reporting dashboard, but a fully integrated self-service interactive platform that enables you to tie detailed reporting to predictions, investments and campaign activations. 

Predictions are directly visible in the dashboard and you can run as many as you would like. 

The platform/Dashboard has four main areas: 

– Media Insights: Provides accurate historical attribution of all media activities.

– Business Insights: Measure and quantify the sales impact from all non-media related factors 

– Predictions: Optimize, forecast and simulate future sales 

– Activities: Keep track of all campaigns and forecasts 

Learn more here.

The quality of the models is assured at several levels:

– Bayesian Inference specific checks: The result of the model is checked to make sure the underlying algorithms performed optimally (e.g; results must have the right effective sample size and parameter estimates must be without unwanted correlations).

– Statistical checks: A good model must fit well to the data and be able to predict well into the future. It is important to achieve this without over-fitting. In all models, we use a holdout test to assess the statistical validity of the model. Additionally, R-squared and mean absolute error are used to check fits

– Subject experts: The learnings of the model are inspected and checked by subject experts. Media learnings are compared to industry benchmarks and any deviations are investigated.

The investment in marketing will give rise to short-, medium- and long-term effects. Blackwood Seven offers models that give deep insights into the dynamics describing the short- to medium-term effects, and brand models that cover the medium- to long-term effects. 

For online marketing, the effect of advertisement is usually seen in the immediate short term. The sales are affected either immediately or up to a week after the ad has been shown. For offline marketing, the period is longer, and the effect of an ad may last up to 3-4 weeks. 

Both short- to medium-term effects are captured in Blackwood Seven’s models. In addition, any long-term effects are accounted for by including brand data in the models. This gives advertisers the opportunity to understand whether changes in e.g; sales are caused by changes in the targeted advertisement or are attributed to the long-term effect of branding.

We define long-term effects of marketing as effects that last longer than the immediate period of up to approximately 4 weeks. 

The long-term effects are determined by activities that do not have a large direct impact on sales but instead impact other variables that can affect sales at a later stage. This can e.g. be marketing activities that slowly increase the consumer’s quality perception of a product and thereby increase their willingness to pay. 

Many of the long-term effects of marketing can be seen as “brand” effects. Effects that positively (or negatively) impact the perception of the brand – and through this can affect sales in the future. These brand effects evolve slowly since they fundamentally are depending on changes in the consumer’s perception: A perception that can be based on 5, 10 or 15 years of consumer experience with the brand.

Sales models include the medium- to long-term effect of marketing and are able to quantify the brand impact on sales. However, to understand the underlying drivers that affect brand perception we offer specific brand models. These models are specifically developed for understanding and optimizing longer-term effects.

As the produced models are highly nonlinear with multiple synergistic effects there is no naive way of isolating effects in an independent manner. Indeed this is a feature and not a limitation. Marketing, and the rest of the world for that matter, does not consist of a bunch of independent events. In reality, everything is interconnected and potentially dependent on each other. This is why the cause and effect have to be re-cast into a simulation scenario. 

The way our platform detects and quantifies causality is by comparing a small change to a variable of interest while keeping all other parts of the network fixed. In this way, we can see how a small change can propagate throughout the networks with all the synergistic and nonlinear effects present. This flexibility in our DSL allows us to model marketing drivers, both soft and hard, at multiple levels. 

Type of synergies accounted for:

1. Media-to-media synergies (efficiency sub-models)

– A display ad is twice as effective when combined with a TV campaign than without.

– A sale is attributed to paid search, but was this sale potentially initiated by a TV ad?

2. Media-to-non-media synergies (15 – 25%)

-In many MMM, website visits are identified as sales drivers. But what drove the visitors to the web page? 

3. Non-media synergies

– Competitors have a negative impact, but how do their impact change when you lower your price? 

In addition to paid media factors, our models account for all relevant non-media factors as well, which we divide into different classes:

– Marketing-specific factors

– Company-specific factors

– Industry-specific factors

– Systemic factors

By accounting for non-media factors, we are able to measure and optimize the effect of paid media with high accuracy and avoid underfitting. In the Bayesian Framework of Blackwood Seven, adding relevant non-media factors allows for lower uncertainty in the media learnings.

Adding relevant non-media variables gives insights into non-media sales drivers. For example, how are sales impacted by price changes, PR activities and competitor media spend? Additionally, having these variables in the model provides us with the opportunity to measure the synergistic effect between paid media and non-media factors.

– Exchange rates, Consumer Price Index, Interest Rate, Unemployment Rate, Consumer Confidence, Stock Index

– Max/Min Temperature, Sun Hours, Wind Speed M/S, Precipitation by multiple locations per market 

– Gasoline & Diesel Prices

– Local Holidays & Observances

– Competitive Media Spend