Create a Model
Robust Model Creation
Mix Modeler raises the bar for marketing measurement accuracy by leveraging incremental analysis and a log-log statistical modeling approach. Incrementality, the gold standard for marketing measurement, establishes a baseline —answering the question of what business outcomes would still occur if all marketing activities were paused. Mix Modeler uncovers this baseline and quantifies marketing’s incremental impact to reveal the true value of every campaign and channel, providing marketers with precise insights to better optimize performance and maximize ROI.
Unlike simpler additive models, Mix Modeler’s advanced multiplicative approach assumes the real-life dynamic interplay between channels, offering a more accurate view of marketing’s true impact as well as a plan’s forecasted outcomes. Built on a foundation of advanced data science and enhanced by flexible model configuration, Mix Modeler empowers marketers with robust measurement insights, enabling confident decisions that fuel business growth.
The flexible model configuration functionality in Mix Modeler allows you to quickly configure, train, and score AI/ML models specific to your business objectives. You can modify configurations in a scalable and repeatable way withing a single UI, producing insights within hours, not months or quarters.
A model in Mix Modeler is an AI/ML model employed to measure and predict a specified outcome based on a marketer’s investments. Mix Modeler allows you to flexibly create models based on different sets of variables, dimensions, and outcomes. For example, you could have a model focused on revenue, another on units sold, and a third on leads. In addition to conversions, we also see customers create separate models for regions and product lines.
Create a Model Exercise
- Navigate to Models in the Mix Modeler UI.
- Click on Open model canvas.
- You will need to provide a name for your model.
- Name = Online Sales MMM + MTA Model + {initials}
- Click Next.
- Now you will be able to configure your conversion event (target feature) and the marketing actions (features) that have been defined in the process of Harmonizing the data.
- For the Conversion goal select Online Sales.
- Next, let's add all our Marketing touchpoints availble to select from the drop down.
- For External Factors we will be adding S&P 500, CPI, Competitor Price, and Thanksgiving. You will need to add a Factor Name and select a Dataset for each factor. You will keep Auto Selected for the Impact on conversion for all factors. Make sure to click on Add external factor after adding each factor until you get to Thanksgiving.
- Factor Name = SPX Dataset Name = SP 500
- Factor Name = CPI Dataset Name = CPI Factor
- Factor Name = Competitor Price Dataset Name = Competitor Price
- Factor Name = Thanksgiving Dataset Dataset Name = Thanksgiving Factor
- For Internal Factors we will be adding Promo and Flash Sale. You will need to add a Factor Name and select a Dataset for each factor. You will keep Auto Selected for the Impact on conversion for all factors. Make sure to click on Add internal factor after adding each factor.
- Factor Name = Promo Dataset Name = Promo Factor
- Factor Name = Flash Sale Dataset Name = Flash Sale Factor
- Leave everything else as default and click Next.
- To finalize the model process, make sure that Allow spend share is unselected and that MTA enabled is toggled on.
- There is flexibility to fine tune the model based the portion of a marketing touchpoint and the confidence in the portion of the data. Leave everything as default and click on Next.
- Lastly you have the option to schedule training and scoring. Leave everything as default and click on Finish.
Congratulations! You have completed creating your first model.