Marketing Mix Modelling, Attribution, Multi Touch Attribution, Unified Measurement Approach, advantages and disadvantages of Marketing Mix Modelling and Attribution, effective marketing measurement, marketing measurement challenges, AdoptoMedia

There is a variety of tools for measuring marketing effectiveness and even more solution providers on the market. Using the right ones is crucial for your business success. In this article you will find out how to choose the most suitable vendor and correctly apply marketing mix modelling and attribution to achieve outstanding results.

We are trying to keep you updated on all marketing measurement trends, that’s why we prepared an article on the use of measurement tools after the phase-out of cookies and IDFA. Get ready to privacy changes and apply solutions that won’t be affected by the new conditions. See how MMM, MTA, brand studies and other approaches are going to function without third-party cookies.

Now there are no doubts about whether to evaluate marketing performance because statistics and analytics is more reliable than experience and gut feeling. According to Gartner’s 2018 survey, where 503 respondents from large companies were interviewed, marketers who use strategic measurement tools report significant performance improvement. It happens because high-quality measurement allows improving planning capabilities and increase enterprise value. The debates now are going on about which tool is the most suitable to evaluate marketing effectiveness. Keep in mind that before applying any measurement approaches define what success should look like and choose the right KPIs that align with your overall strategy.

Strategic marketing measurement is on the rise for three reasons:

  • Significant performance improvement;
  • Increasing cost of media; for example CPM for primetime TV has increased by 43% over the last 5 years. To justify growing spends or find new, more efficient channels high-performing marketers rely on tools which help explain their choices and convince the leadership to invest in a certain activity.
  • Higher marketing analytics maturity. According to a Gartner research, 85% of CMOs plan to improve data and analytics maturity in the nearest future.

Marketing Mix Modelling has been around for a few decades, helping companies to evaluate the effectiveness of advertising, but with the rise of digital era, attribution became increasingly helpful to measure campaign performance. The two tools have different characteristics, so it seemed reasonable to join them into a single, comprehensive solution. This approach was actively advocated mainly by those who sell data panels, because accurate attribution results require highly granular data. Some vendors included attribution models in their offers and attracted many new clients with this unified measurement approach (UMA). You might be surprised, but this move turned out to be wrong, and now marketers are looking for a more effective way to tackle the measurement issue. Even though UMA offers such benefits as work with a single vendor and well aligned reports, it complicates planning and implementation processes. According to Interactive Advertising Bureau’s 2019 research, in unified approaches, major contribution of 80% comes from MMM, while MTA accounts only for 20% of the output. Moreover, the results of MMM and MTA often differ dramatically, and when you run a combined model you might get something that is even further from reality. It doesn’t mean one of them is useless though. Experience and figures show they shouldn’t be competing or be joined together; the right approach is using them separately where they are more suitable, so they could complement each other. Further you will find the necessary information to better understand both tools, their strengths and weaknesses to better apply them and choose the perfect provider.

Marketing Mix Modelling

It’s a time tested, reliable tool, well understood by marketers, that has been around for a few decades. The fact it’s not new doesn’t mean it’s ineffective. Advanced MMM techniques are able to produce granular and actionable insights, so it’s wrongly considered to be slow and backward looking. Such opinions exist because there is no industry standard, so a wide range of MMM solutions vary in actionability, speed and granularity. As figures above show, it’s more valuable than attribution. The tool is perfect for strategic measurements, as it gives an overall picture and shows how individual channels and activities perform within a campaign. At the same time, model output becomes more granular with customization and availability of high-quality, detailed data. The more granular a model is, the more expensive and difficult it is to build. MMM can work with numerous variables and KPIs simultaneously. It accounts for all major sales drivers like advertising, promotion, seasonality, competitor’s activities, etc., and covers offline media (TV, radio, OOH, sales points), which is impossible for MTA. This tool works with aggregated datasets, so privacy regulations don’t limit the analysis quality. Advertisers use MMM not only to see how marketing activities perform, but also to improve campaign planning processes.

But there are some limitations to this approach:

  • Scale: there is no agreement on threshold amount of investments, but in AdoptoMedia’s case, if a company spends less than $1 mln annually, it’s not enough for MMM to generate meaningful output.
  • Budget: since measurement tools are not cheap, your marketing budget has to be substantial, so that the cost of implementation is relatively small compared to it.
  • Media Mix: the key element of MMM is analyzing variation in data, so it will only work if media mix changes over time, because otherwise there will be nothing to analyze. At the same time, the variation can’t be too dramatic, because in this case there will be no benchmark for the model to function.
  • Data: to build an accurate MMM model you need three years of historical data on marketing activities and KPIs.
  • Transparency: not all solution providers are open about the way their models work, which makes it difficult to interpret the obtained results and plan future campaigns.

All these limitations mainly concern advertisers who want to use it, so not everybody meets the requirements to be able to use MMM and benefit from it. There are no limitations that make us question the effectiveness of the approach on the whole.

Multi-touch attribution

Attribution is a relatively new tool, so fewer advertisers are familiar with it. It is most effective for measuring digital campaign performance and fine-tuning allocations within a channel. Given that in the first six months of 2019 digital advertising spend in the US alone made up $57.9 billion, according to IAB PwC Internet Advertising Revenue Report, this tool should be very useful now. MTA gives highly granular output and tactical insights on improving ROI. It is not so limiting in terms of marketing activities scale, and can be applied on a single campaign because it measures fewer touchpoints. The restriction concerning the amount of variations in media mix doesn’t apply to MTA either, but it also has some weaknesses:

  • Data privacy: MTA’s strength, ability to function at a highly granular level, backfired when legal restrictions on private data exchange (GDPR and CCPA) came into force. These regulations severely limit MTA’s functions. For example, some browsers are starting to use temporary cookie files. It’s also difficult for attribution to function within walled gardens (Facebook, Google, Twitter). To address this issue IAB is currently working on a solution that will satisfy both advertisers and consumers, but when and if it arrives is unknown.
  • Data collection: It takes about four weeks of data to build a working optimization model, but by the time you have all the necessary information, campaigns usually end, because display advertising campaigns on average run for about a month. To be able to apply the data you gather, such ads should be executed continuously throughout a year, which is too expensive.
  • Identity: Even though MTA tracks every customer individually, it doesn’t keep this data for more than 2-3 weeks which is not long enough in many cases. So if more than 3 weeks pass between the first touchpoint and the purchase, the algorithm will see these actions as if performed by two different customers. Another problem is stitching together touchpoints from different devices into a single customer journey.
  • Offline metrics: It was originally intended for digital world, so it usually doesn’t cover offline touchpoints. It is possible to add offline KPIs, but the process is quite complex and increases error rate.
  • Lack of standards: Since it’s a relatively new approach, there are no industry standards against which to check its quality.
  • Transparency: MTA is not transparent enough and not so well understood by marketers.
  • KPIs: It can work with limited number of KPIs only.

As you can see, there are a lot of problems related to MTA. Most marketers agree that using it alone is not a reasonable approach, because companies usually implement a complex media mix that includes offline and digital channels.

You can see that the two tools are completely different, but they have some things in common they are both expensive to run, require high-quality data to get accurate results and skilled experts who can interpret the output and make decisions based on them. Some companies have different teams working with MMM and MTA, that study the results separately and don’t cooperate. In the graph below you can see what approach is more widely used among IAB survey respondents.

Marketing Mix Modelling, MMM, Attribution, Multi Touch Attribution, MTA, marketing performance measurement

Now that you know enough about the two approaches, it’s time to choose a solution provider. There are four aspects you should pay attention to when doing that.

  1. Vertical experience: it’s perfectly logical that a vendor knowing more about your business will be able to better analyze processes and outcomes, so you should look for somebody who has experience in your sphere. It doesn’t mean, you shouldn’t work with a provider if it’s new to your area, but it will take more time to implement the solution. AdoptoMedia, for example, has solid experience with banks and telecommunication companies. If you work in either sphere, we can quickly build a model for you and customize it to your specific needs.
  2. Transparency of modelling approach: vendors should explain how they build their model and how it works for you to better understand the results and limitations. In AdoptoMedia we are completely open about our algorithms, with us there is no black box.
  3. Granularity, Accuracy, and Access to Data: You want to know about the vendors data collection process, data partnerships, data granularity and model verifications. AdoptoMedia provides actionable insights at the level of company branches. Results are highly accurate because models pass over 20 statistical tests before they are used.
  4. Actionability of insights: it’s not enough to just understand what happened, today analytics has to provide insights on how to optimize marketing strategy and planning. AdoptoMedia provides a solution that results in up to 30% ROMI increase through focus on more effective channels and elimination of ineffective ones.

AdoptoMedia offers a highly effective solution, combining advanced MMM and AI, which allows building models more quickly and adapt them to client’s needs. Our models can forecast ROMI and track it in real time, simultaneously work with many variables and KPIs, which allows for a comprehensive analysis of you marketing activities. We are not only transparent about our modeling algorithms, but can also help you set up a transparent full-cycle in-house agency with our Media Plan Manager and Media Billing modules. With us you can track all marketing activities end-to-end, from planning to execution and easily check agreement compliance. Our CheckMedia Solution is a flexible software that can be smoothly integrated in existing IT infrastructure and address tasks on both strategic and tactical level. The platform is easy to operate, so after we carry out the implementation process, you can manage it yourself. It doesn’t require constant manual maintenance, with automatic ingestion of new data, the results are updated as well. If the model stops passing statistical tests or market situation changed drastically, analytics will adjust the model and it will continue functioning properly. With our platform you can optimize marketing spends, make smart, data-driven decisions and achieve best-in-class marketing accountability.