설명서Mix ModelerMix Modeler 안내서

Build models

Last update: Tue Mar 24 2026 00:00:00 GMT+0000 (Coordinated Universal Time)
  • 주제:

작성 대상:

  • 사용자
  • 관리자

To build your custom AI-powered models, the interface provides a step-by-step guided model configuration flow.

Models​Open model canvas

설정

Setup

  1. NameDemo modelDescriptionDemo model to explore AI features of Mix Modeler

  2. Next​Cancel

구성 configure

Configure

  1. Conversion goal

    1. Conversion​🔗 Harmonized datasets​예: Online Conversion.

    2. Create a conversion

  2. Marketing touchpoints​🔗 Harmonized datasets

    1. Touchpoint include

      • Clear all

    2. Create a touchpoint

    note note
    NOTE
    You cannot set up the model with touchpoints that have overlapping data and there must be at least one touchpoint with spend.

  3. Eligible data population

    • For each container, define one or more events.

      1. For each event:

        1. __

        2. equals​ not equals ​ less than ​ greater than ​ starts with ​ doesn’t start with ​ ends with ​ doesn’t end with ​ contains ​ doesn’t contain ​ is in ​is not in

        3. __

      2. Add event

      4. Any of​ All of ​ Include … Or … ​Include … And …

    • Add eligible population

    • Remove container

    • ​​​

  4. Factor dataset

    • Add Factor

      1. Factor dataset​🔗 Factor type​****​Internal​ ​ ​External**

      2. Impact on conversion​ Auto ​ Positive ​ Negative ​Auto

  5. 1 52Give contribution credit to touchpoints occurring within​ weeks prior to the conversion ​Define lookback window

  6. Next
    Back
    Cancel

고급

Advanced

  1. Spend share

    • Allow spend share

      • A channel doesn't have enough observations (for example, low frequency of spend, impressions or clicks).
      • You are modeling spiky but regular, and potentially high-spend media (like TV for some brands), where data may be sparse.
      note note
      NOTE
      For one-off investments (for example a Super Bowl ad), consider to incorporate that data as a factor rather than to rely on spend share.

  2. MTA enabled

  3. Prior knowledge

    1. Rule type​Absolute values

    2. Name​Contribution proportion

    3. Level of confidence

    4. Clear all​ Contribution proportion ​Level of confidence

Set options

🔗 🔗 🔗Set options

일정

Schedule

To scheduled model scoring and training:

  1. Enable scheduled model scoring and training

  2. Scoring frequency

    • Daily05:22 pm
    • Weekly05:22 pm
    • Monthly05:22 pm

  3. Training frequency​ Monthly ​ Quarterly ​ Yearly ​None

Training window

Define training window

  • Have Mix Modeler select a helpful training window

  • Manually input a training window. Include events the following years prior to a conversion

Granular insights reporting fields

Granular insights reporting fields

You define these harmonized fields so you can drill down in the reporting of your model using granular reporting columns instead of having to create separate models.

For example, you build a model that is focused on revenue, but you are also interested in the campaigns, media types, regions, and traffic sources performance. Without the granular incrementality reporting functionality, you would have to build four separate models. With the granular incrementality reporting functionality, you can break down your revenue model on campaigns, media types, regions, and traffic sources.

  1. __​Includes
  2. ******
  3. Clear all

🔗 🔗conversionPassthrough​touchpointPassthrough

Finish

  • Finish

    • Create instance?​Ok Awaiting training

      Cancel

    • If more configuration is needed, a red outline and text explains what additional configuration is required.

  • Back

  • Cancel

recommendation-more-help
d5f9b631-c793-4214-8dc7-f78d1750e4f4