Creating a model is a crucial component in the segmentation of your audiences within Rev.Up. If we think about it in terms of the core pillars of Rev.Up, modeling is a key part of the segment pillar.
A quick review of the three pillars
Connect: Import your first party data into Rev.Up, match and merge multiple connections together to create a unified view and append attributes from the D&B data cloud to your own data.
Segment: Segment your data into audiences that will be used in campaigns.
Activate: Send your audiences to an end platform such as Marketo, Eloqua, SalesForce, LinkedIn and more.
How are models used in segmentation?
The purpose of any model in Rev.Up is to score a set of accounts for their propensity to convert or to take a specific action. A common example is using an Account Fit model to score your prospect accounts for their likelihood to become a customer. To achieve this, you build a segment of your prospects, create a model that is suitable to score your prospects and then activate your model.
When the model is activated, your prospect accounts will have a rating (from A to F). The higher the model rating, the more likely it is for the prospect to become a customer. These model ratings will be available for you to use as an attribute when building any segment. Hence you can include or exclude account records from a segment based on their model rating, so you now have an ability to target your prospects differently based on the model prediction of how likely they are to convert to a customer.
For example, here is a segment that includes prospects that are rated as A or B for their likelihood to convert to a Gizmo customer.
Why should you use model ratings in your segments?
Using a model rating in a segment allows you to narrow down the accounts that you reach out to using a digital channel. For example:
- Significantly narrowing down the accounts may be important when you are running a campaign in a high-cost channel, e.g. having your sales representatives make calls or paying for impressions or clicks on an ads platform.
- For some channels such as email, narrowing down the accounts using a model rating is less impactful on the overall campaign cost, hence a broader range of model ratings can be used.
What types of models can be built in Rev.Up?
There a several different types of models that can be built in Rev.Up. Here is a short description of each type. For a detailed description on how to build these models, please review the links at the bottom of this article.
Machine learning models
These models use machine learning and are trained to predict success events.
- Account Fit models score your prospect or customer accounts to determine how likely they are to convert. The word convert may mean different events: the model can be used to predict how likely a prospect is to become a customer, or to determine how likely a customer is to purchase another product.
- Cross-sell models are the most advanced model type. They use product purchase data to determine when an existing customer account is likely to purchase the same product or another one of your products. These models require importing your product purchase data. Cross-sell models are used for prioritizing within your existing customer accounts. Cross-sell models can also use transaction data and will rank your customers based on their likely amount of spending. There are two types of cross-sell models:
- Cross-sell First Purchase model scores customer accounts to predict their likelihood to purchase a product that they have never bought from you before.
- Cross-sell Repeat Purchase model scores customer accounts to predict their likelihood to purchase the product from you again within a defined timeframe.
Rule-based models
A rule-based model allows you to create your own custom model that assigns ratings to accounts in a segment using the exact rules that you define. This model does not use machine learning.
You are able to use the query builder to define the criteria (rules) for how the accounts in the selected segment should be further partitioned into accounts with the rating A, rating B, etc. It is similar to how you further partition accounts in a segment into subsegments. Such ratings defined by your rules in a rules-based model can then be used to build other segments.
Where can I find more resources on how to build models?
Rule-based models:
Account Fit models:
- How to use and build an Account Fit model
- How to review an Account Fit model
- FAQ: Account Fit models
Cross-sell models:
Comments
0 comments
Please sign in to leave a comment.