Marketing Mix Modelling vs Conversion Modelling: What’s the difference?
Table of contents
- Marketing Mix Modelling VS Conversion Modelling
- Strategic VS tactical approach
- Do you need a lot of historical data?
- How fast can you start taking action on the results?
- How granular my analysis can get?
- What about complex customer journeys and ITP?
Let’s continue our data-driven tool comparison series. This time we’ll focus on two modelling solutions: Marketing Mix Modelling (MMM) and Conversion Modelling.
What do we know just from the naming? Both tools are related to marketing, both have modelling at the core… Don’t you assess your marketing mix with Conversion Modelling too? Is it just two different names for the same solution? (Spoiler: no! Otherwise, we probably wouldn’t have written this article).
Dissimilar to Multi-Touch Attribution, MMM is designed to solve a different challenge altogether. Let’s figure out what these approaches actually do.
Marketing Mix Modelling is a complex statistical analysis, most often regression-based, that looks back at sales over a long period of time (from several months to multiple years) and determines what caused those sales. It looks over significant budget allocations to digital and offline channels at macro level (think of TV and Radio ads, Digital Marketing, Events Sponsorship, Billboards, Printed Ads and other categories) and takes into account market conditions, competitors, special promotions, inventory levels, product prices, seasonality and even weather. Based on this information MMM would further look for spikes in sales, determine whether they are related to any of these factors and calculate the impact of any future activities within these macro channels.
Ex.: A large multinational CPG company needs a fundamental solution to revise all their digital and offline marketing activities, their CMO runs MMM once per year to make data-driven decisions upon budget allocation across brands, markets and channels.
Conversion Modelling is focused on digital marketing activities and operates on a website visit level. It leverages Machine Learning to assess the impact of each single visit and predict conversion probability based on the variety of visitor’s actions and characteristics. These prediction data can be further used to solve various performance marketing challenges from getting extra signals for smart bidding to strategic media planning and budget allocation.
Ex.: Performance marketing and analytics teams of a large global Retailer are willing to understand the true value of their digital marketing activities but lose lots of attribution data due to privacy and cookie limitations, they’re using Conversion Modelling on a day-to-today basis to make real-time budget reallocation decisions on channel and campaign levels and optimise their ads performance.
Not quite so, MMM and Conversion Modelling do overlap when it comes to High-level use cases. Both can be used to make better decisions on channel level budget allocation for digital marketing activities. However, if you need to assess the holistic marketing strategy across online and offline for a long period of time, MMM would be better suited to do so.
On the other hand, due to its macro nature MMM doesn’t allow for tactical media planning, ads optimisation and smart bidding strategies. Conversion Modelling is perfectly designed for granular hands-on tactical optimisations.
Both solutions have an element of modelling and prediction, and as a result you do need historical data to run them.
Conversions Modelling, however, only requires a few weeks of data and once the ML model is built you can analyse even freshly launched campaigns. MMM is very different – it would require from several months up to multiple years of data as this approach would assess various correlations across the marketing channel spend and conversion contributions as well as seasonality and other external factors.
MMM is sophisticated and takes around 3-6 months before you can start the analysis. This is why it is usually only performed once per quarter or so.
As for Conversion Modelling, Once the model is up and running (3-4 weeks of historical data) you can get the results in close to real-time and start acting upon that in around 1 day.
MMM’s primary goal is the high level budget allocation, so it won’t be able to get you lower than the macro channel level (like “Digital Marketing”, “TV Ads”, “Football match sponsorship”, “Billboards”, etc.). Technically it is possible to run MMM on digital activities only, but still in this case you wouldn’t be able to get more granular than channel level (like “paid-social”, “paid-search”, “direct”, “organic”, etc.). Conversion Modelling on the other hand allows for lots of flexibility and supports a very detailed analysis down to campaign, ad content and ad term.
Since MMM is analysing data on a channel level, and not necessarily tied to cookies, it’s fair to say that MMM mitigates for ITP and cross-device/cross-browser. As for Conversion Modelling, it is a real master in battling ITP and cookie limitations even on a user level since it evaluates the user’s probability to convert straight away when the visit happens, without waiting for the final conversion to happen within the same cookie.
Marketing Mix Modelling and Conversion Modelling are operating on opposing levels and are solving different challenges.
If you’re looking over all marketing channels and need a fundamental approach to assess all your marketing activities, including digital, TV and offline, Marketing Mix Modelling is a good way to go.
We believe that it can be run in tandem with Conversion Modelling, since these two solutions can be complementary to one another. However, when it comes to digital marketing, Conversion Modelling is much more agile than MMM. Over the past few years we’ve seen how fast the industry and the customer behaviour can evolve which often makes today’s marketers adjust their strategies “on the go”. Conversion Modelling’s flexibility and predictive nature allow not only to evaluate freshly launched campaigns on all levels from channel to ad term but also generate more valuable signals that help to get the most out of smart bidding and ad delivery optimisation.
If you’re keen to see how it works, request a trial!
You might also be interested in
How does iOS 17 affect marketing attribution?Learn more
[VIDEO] Can we really trust ad platforms' reporting?Learn more
[VIDEO] Last Non-Direct Click vs. SegmentStream’s attributionLearn more
Never miss an article
Get the latest articles, event invitations and product updates delivered straight to your inbox.
Thank you! You’ve been signed up for our newsletter.
Get started with SegmentStream
Learn about Conversion Modelling and why it is a true next-generation solution to outdated marketing attribution and conversion tracking tools.