Marketing Mix Modeling, Multi-touch attribution • 8 min reading time

Is daily MMM a good substitute for MTA?

Phil Spencer - published on January 24, 2024

Understand how daily MMM, positioned as an alternative to MTA, impacts your analytics strategy. Discover the strengths and limitations of each approach and learn to leverage them effectively.

The increased focus on consumer privacy in recent years has caused well-documented disruption in the marketing ecosystem, causing some marketers to turn away from solutions that analyse person-level consumer journeys (such as Multi-Touch Attribution – MTA) and towards aggregated analyses such as Marketing Mix Modeling (MMM). The challenge this presents is how to maintain the granularity of results that marketers have come to expect, using data and techniques that are designed for something slightly different. One approach that has grown in popularity is daily MMM – using marketing mix modeling approaches on more granular data, with some vendors positioning this as an alternative to multi-touch attribution, and potentially over-promising the level of granularity that can be achieved. We still see important differences between MTA and daily MMM and in this blog post we outline some of the key considerations.  

A general comparison (MMM vs MTA): 

1) MTA only applies when the KPI comes from a website running a suitable analytics suite (i.e. a website controlled by the brand in question). MMM can apply to any KPI that is tracked consistently over time. 

2) MMM can include non-trackable media such as Out of Home, Linear TV, Event Marketing, etc. MTA can only include trackable digital media (including OTT channels). 

3) MTA will provide more granularity around what specifically is measured, at times measuring each specific piece of content on each specific platform including the order and time gap of exposure (i.e., click path or journey), MMM will typically be at the media channel or campaign level, with some newer approaches claiming to get as low as the ad group. 

4) MTA does not generally measure ‘true’ incrementality, it allocates credit between multiple touchpoints; MMM measures incrementality but does not contain information about the number of touchpoints per conversion/sale. To find out more about incrementality, you can read our blog post How incremental is your attribution here. 

The biggest difference is that MMM approaches (whether weekly, daily or any other periodicity) will always be top-down, whereas MTA approaches work bottom-up from user-level data which means there is a lot more signal (1 data point per website visitor vs. 1 data point per day). User-level data allows for real time granular reporting that is naturally actionable for tactical optimization. 

Yes, daily data for MMM will increase the degrees of freedom and hence, increase the chances of being able to measure more granular media activity (vs. weekly data) but this increase is not automatic and not without cost. A large portion of day-to-day variance in the KPI will be explained by “day of week” seasonality terms. Some models that report enhanced detail from daily MMM data are likely doing so via priors and business rules that assign volume up and down hierarchies, which in some cases can be misleading (as to what was actually “modeled”). Weekly MMM must also walk a fine line between using enough historical data for stability, while still reporting on current results and new campaigns. 

While failing to assess true incrementality, MTA will still evaluate data over a shorter time window, so that changes in media effectiveness will be seen faster and marketers can act on this more quickly. For example, an MTA solution like Roivenue calculates at the most granular level, down to individual touchpoints (i.e., impressions or clicks) – so the user has flexibility in how results are rolled up postmodel, e.g. into campaigns or ad groups, or even custom dimensions that can be created on-the-fly. 

MMM modeling requires some level of aggregation pre-modeling (at least to daily, and probably aggregating multiple ad groups and campaigns). Daily MMM will typically use 3 to 24 months of daily data, so changes in actual efficiency (i.e. a new curve) will be slower to reveal themselves. ROI will almost always change day by day, but that will be due to moving along the same curve, better/worse leverage of AdStock, or changes in cost per unit of media – none of which truly represent a change in consumer response to media (they just represent better/worse media planning). 

On the other hand, MTA is limited to attributing value across digital media touchpoints, whereas daily MMM can account for the full scope of factors and will pull in other explanatory factors like offline media, pricing, and other kinds of non-marketing related external variables. That said, there are some known dangers of daily MMM (vs. full MMM) which are under-estimation of the true ROI of media as well as less accurate quantification of the longer-term base drivers. 

Typical use cases for MMM are: 

  • Measuring effectiveness of both online and offline marketing channels and activities, serves as the authoritative truth for this purpose
  • Marketing budget planning and budget allocation
  • Budget optimization and spend re-allocation on bi-weekly, monthly or quarterly basis
  • Holistic strategic view on what drives the results for your company – which media, what other factors (pricing, market trends, competitors)
  • Senior management and C-level reporting
  • Business scenario modeling
  • Forecasting of future results

Typical use cases for attribution modeling (MTA) are: 

  • Measuring effectiveness of digital channels (most often biased towards “click-heavy” channels)
  • Daily and weekly optimization of online (and specifically performance) marketing campaigns
  • Daily and weekly detailed reporting
  • Getting ROI insights on individual ad-set or creative level

Typical use cases for daily MMM: 

  • Daily modeling might be suitable to fill the middle ground – to get a quicker and more granular view than full MMM but need a more holistic measurement than MTA can offer (ie. also including offline media/non-digital campaigns). This might be exactly what some clients want, completely irrelevant for others.
  • Daily MMM can also be good when more campaign detail is needed but MTA is not possible e.g. because of the KPI that needs to be measured or restrictions on the usage of user level data.


So, in short, MTA will still be more genuinely “real time” and it will be more granular. It can blend with the AdTech ecosystem for on-the-go optimization and all the data sources can be refreshed automatically, meaning the model is updated daily to provide near-real-time results on the effectiveness of different media buys. Daily MMM on the other hand will be more strategic and likely better for real time “business planning”. It is delivered less frequently compared to MTA and is more costly in time and effort. 

So, if the goal is for a digital media manager to make small (but collectively important) changes to existing campaigns then MTA is still significantly better. If the goal is for marketing to have a more frequent pulse on what is driving the overall business, then daily MMM would be better. 

The key lies not only in understanding the differences between daily MMM and MTA but also in harnessing their unique strengths to craft an analytics strategy that resonates with the specific needs of your business. With these insights, we at ScanmarQED empower marketing professionals to make informed decisions that drive success in an era where data-driven precision is paramount. 

Picture of Phil Spencer

Phil Spencer

Managing Director UK