How is building a Market Mix Model like playing Cluedo?

How do you explain a dense concept like MMM in a simple yet meaningful way? According to our own Will Marks, by likening it to the classic murder mystery game Cluedo.

How is building a Market Mix Model like playing Cluedo?
By Alex Neighbour | 06 Aug 2024

Our very own Head of Marketing Science, Will Marks, had a dream last night.  A literal dream. 

In a sign he may be working too hard, instead of enjoying a restful slumber, Will dreamt up an analogy for simplifying the concept of Market Mix Models (MMM) and how they’re built. Because of course he did.

Explaining MMM in a way that’s colourful and memorable, while still imbuing it with the right level of detail to make it meaningful, can be a tall task. 

Let’s listen to Will deploy the classic murder mystery game Cluedo (‘Clue’ to some) to frame up what can be a dense concept.

How exactly is building an MMM like playing Cluedo? 

In Cluedo, the aim of the game is to deduce how a victim was murdered and by whom

In many ways the way this game is played is similar to building an MMM, where the objective is to identify the drivers of a sale.

In Cluedo (and MMM), there are 3 basic steps to discovering ‘whodunit’:

1) You must first consider the key factors involved in the murder

In Cluedo, we need to discover who the murderer was, where the murder took place, and what the murder weapon was.

In an MMM, we want to consider all major sales drivers – media, pricing, and externals etc.

2) You need to know how to interpret the evidence you’ve collected

In Cluedo it’s simple, if Colonel Mustard was seen down the pub when the murder took place it couldn’t have been him. 

An MMM also needs to know how to consider the information it receives. This is why we have Priors, so the model can consider adstocks, saturation curves, channel synergies etc.

3) You need to collect enough evidence to achieve a result

In Cluedo, you don’t find all your clues in one place. You need to wait your turn, roll the dice, collect evidence, and explore the board. 

An MMM also needs time. In this case, it may need 2+ years of data to amass evidence across a number of inputs, and over a stretch of time.

You don’t need all the evidence to form a conclusion

The winner of Cluedo probably won’t have needed to collect every shred of evidence before figuring out who the culprit was.

Instead, players use logic and all available clues to make a highly informed decision.

Similarly, Market Mix Models will also never contain every possible piece of information. Instead, by working through the 3 steps above, you can confidently draw firm conclusions.

Hazarding a guess in Cluedo is just like ‘last click’ attribution

Sometimes, in Cluedo, you can ignore proper strategy and simply hazard a wild guess by blaming whoever was last seen with the victim, and placing suspicion on whatever implement was found lying around. 

You may have heard your younger siblings deploying such ill informed guesswork.

This guess, based on scant evidence, would be the equivalent of solely using a ‘last click’ attribution strategy to determine sales drivers. It’s not usually a very thorough tactic, it avoids compiling much of the evidence, and it seldom gives you a complete or correct result.