3 min read
19 Aug


If you are expecting a formula or even a ballpark prepare to be disappointed. This is genuinely a question that falls in the ‘how long is a piece of string?’ box. I have worked on marketing mix modelling engagements where the fees were between £5k and £500k.

I will instead propose that cost is perhaps not the best way to think about it. And run through some factors that do not correspond well to the overall fee.

Investment, not cost

Marketing mix modelling is an approach that generates business recommendations. Recommendations that if implemented grow your business, save you money, or both.

Set aside concerns that you are purchasing models and re-frame as an investment in your business that improves your marketing effectiveness and/or efficiency.

While it is not possible to guarantee the scale of these improvements, even before you start you can compare the modelling fee quote to your current spend on marketing.

Maybe it does not seem so steep in those terms? Now assume it unlocks a 1% saving.

If that saving is larger than the fee quotes the investment breaks even straight off.

Number of models is not a good guide

Any engagement uses a data set that needs to be collected, cleaned, and processed prior to modelling. A chunk of the overall fee quote is a fixed cost, regardless of how many models you build using that data.

But building two models does increase the cost but not proportionally. Some more inputs will be specific to the second model and need creating. Then there’s extra analysis time needed. But the fixed cost described above is shared over the two.

There are exceptions at extremes. One engagement I did needed two models, the first of which was an input in the second. In this situation, to answer the business question, it was two models or nothing. Reducing the scope to one model was not possible.

At the other end of the spectrum, incremental cost of more models to a multi-brand, multi-national engagement may be low. This does not mean that the fee quote is ‘cheap’ and therefore risky. Just that huge economies of scale can be passed on.

Finally, one model of one product for one FMCG brand is relatively straightforward to build compared to one model for national sales of a retail brand broken down by store. Industry can dictate question type that in turn leads to different modelling architectures.

Frame discussions at the outset around the questions that need to be answered rather than getting stuck on how many models need to be produced to do this.

Number of business questions is not a good guide

If a marketing mix modelling engagement only has one primary question, compared to another that has ten it does not follow that it should be cheaper. The same data collection and modelling process needs to be applied to each.

I have had the objection before that if only one question needs to be answered then less time is needed for the write up & recommendations. If there is only one question it is bound to be a big one. Do you really want to scrimp on how that is relayed back to you?

Frame discussions at the outset around the value to the business of the questions to be answered. This gives a way to prioritise if scope needs to be cut.

What price on expertise? 

Modelling is just a tool (my pet topic.) You are not in the market just to buy marketing mix models; you are buying expertise that can provide valuable recommendations.

You can load Upwork and find someone to build you a marketing mix model for much cheaper that I will. There is nothing wrong with that (I have a profile on Upwork) but can you be sure that the provider knows their technical stuff, really? Do some research

You can hire a first-class econometrics grad as an intern. That probably covers you in terms of technical expertise, but do they yet have the commercial experience or business acumen to make the most of a brilliant model? Perhaps collaborate with them

Even in the realm of experienced practitioners, you have an array of providers, from soloists/collaborators like me to global marketing effectiveness agencies. With different pricing models across this spectrum this is where it gets trickier for you.

Find out about who will be doing the modelling & developing recommendations. Commercial experience, technical/industry specialism, communication style 

The builder’s triangle

GOOD, FAST, CHEAP. What is most important to you? You can only have two. If you need a GOOD and FAST turnaround, expect the fee quote to go up – it is the only way that most providers will be able to guarantee the quality stays high enough.

What can I get for £x?

You have probably seen those memes that artist freelancers have created when their prospective clients object to their fee and ask them for what they can do for less. If you have not, look them up. It does not work quite that way for modelling.

But there are options. Revisit the business questions and see if the modelling can be scaled down by holding over the lower priority questions for later. Or maybe alternative approaches will give you good enough answers to use initially.



That is why I only give some “starting from £per day” on my website. It depends on so much that I only publish a minimum. Because as I hope that I have shown “per quantity” pricing just does not make sense when what you are buying is intangible.


Please ask before reproducing my material partially or wholly for commercial use.

 © Jo Gordon Consulting Ltd 2020

* The email will not be published on the website.