Online Service description for the Digital Marketing Mix Modelling service


This service description provides a general description of Sellforte’s Online Service (Digital Marketing Mix Modelling service). Sellforte reserves the right to change and modify features of the service over time based on users’ feedback and needs without separate notice.

The Online Service is a software as a service platform for analysing the effectiveness of marketing that is provided to users electronically and accessible via Internet using a commonly used browser.


Common business questions that the Online Service can help users analyse are following:

  • How are marketing activities impacting outcome metrics, such as subscriptions, purchases, sales, margins?
  • How do different media and marketing campaigns differ in their effectiveness of driving outcome metrics?

To enable users analyse these questions, the Online Service can be used to generate a statistical estimate, which splits the outcome metric (e.g. subscriptions, sales) to several components:

  • Base: the outcome that would be achieved without marketing activities
  • Increment due to each marketing activity: the increase to the outcome metric resulting from conducting marketing

Based on this breakdown, the Online Service may be used to estimate financial return for each marketing investment (ROI), as long as user provides corresponding financial values for the outcome metrics (such as financial value of a subscription) and for the marketing metrics (such as cost of an impression).


To be able to provide statistical estimates of marketing’s impact to outcome metrics, the user needs to provide data to the Online Service in accordance with the instructions from Sellforte, but including at least following data:

  • Outcome metric data (e.g., subscriptions, purchases, sales, margins)
  • Marketing data: Activity (e.g., impressions, clicks) and investment (e.g., cost of impressions) data for each marketing activity that might affect the outcome metric
  • Data for external/control variables: Any non-marketing data that enables a more accurate analysis, e.g., competitor activity data, distribution network coverage

All data needs to be (i) accurate, (ii) in time-series format, (iii) machine-readable, and (iv) have a suitable time granularity (typically daily-level) and (v) to cover the whole modelling time period.

The scope and dimensions of the analysis results in the service depends on the data that the user has provided. For example:

  • If the user would like to compare campaigns with each other, the marketing data needs to have sufficient information about which media belongs to which campaign at any given time.
  • If the user would like to compare geographies with each other, geography-dimension needs to exist in all data, including in the outcome metric and each marketing metric.

Having a certain dimension in the data does not always mean that the analysis results can be provided with exactly the same dimensions. This can happen for example in the case of certain marketing investments being so small compared to other activities that the underlying statistical models determine their impact statistically insignificant and thus the estimate would be unreliable.


The analytics engine of the Online Service incorporates statistical methods that the users can leverage to generate analysis results. As such, the quality of the analysis results depends on several factors, including but not limited to (i) quality and accuracy of the data provided by the user, (ii) completeness of the dataset provided by the user (are all the factors influencing the outcome metrics available?), (iii) nature of business and industry, (iv) major external events.


The user-interface of the online service has following features (subject to change over time):

  • Connecting data to the service:
    • Uploading data via files
    • Connecting data via third-party service(s) (feature not yet available for users as self-serve, but is available as a Sellforte-assisted feature)
  • Reviewing analysis results
    • Various charts for reviewing data inputted to the service, and ability to filter the information
    • Various charts for reviewing decomposition of the outcome metric with filters/selections, and ability to filter the information
    • Various chart for reviewing the ROIs, and ability to filter the information
  • Finding optimal budget allocation
    • Feature for finding an estimate for optimal budget allocation by setting a target for the outcome metric and giving certain optimization constraints
    • Feature for finding an estimate for optimal budget allocation by setting total marketing investment level and giving certain constraints