Overview

The business fit is an important dimension of B2B targeting, enabling Sales and Marketing to focus spending and effort on prospects with budget and need. Used with first-party and third-party engagement information, Account Fit scoring allows your account-based marketing (ABM) program to appropriately scale across priority tiers and channel spending.

In Rev.Up, use the Account Fit (Custom Event) modeling to build an account fit rating for your ABM audiences. You can build your training data by exporting a business segment you want to use for model training, adding your best fit accounts from the past, adding the success criteria (event) column, and loading the data into Rev.Up for training a machine learning model.

Rev.Up builds your Account Fit model and uses it to add scores to the accounts in your business segment that you selected for scoring. Now you are ready to run ABM campaigns like High Fit/High Engagement, High Fit/High Intent, High Fit/Brand Nurture, and more. 

 

Step by step: build an Account Fit model

Prerequisites

Create a segment in Rev.Up that contains the accounts you want to augment with Account Fit scores. After the model is trained, it will score the accounts in this segment.

Create a segment just for prospects. It should contain accounts who have never purchased, so it does not include any current or past customers. It is recommended that this segment contains at least 10,000 accounts. 

 

Step 1: Create training data

Create a CSV file with data that will be used during model training. The training CSV file should have the following fields:

• Account ID (required) - the same ID used in Rev.Up Account Import
• Event (required) - a success event flag that can have one of two values: 
  • Value 1 marks a successful past target
  • Value 0 marks an unsuccessful or untried prospect
• Website (optional when using the Company Name field)
• Company Name (optional when using the Website field)
• City (optional)
• State (optional)
• Zip (optional)
• Country (optional)
• Phone Number (optional)
• DUNS (optional)

To create training data from your prospect segment, first use the Segment Export functionality to export this segment to a file.

Add a column called "Event" to the exported file, populated with zeros (Event value = 0) for your prospects. 

Next, append a list of accounts that exemplify the success events that you want to model. The number of such accounts should be at least 200, more is usually better, 500 or more is recommended. Populate the Event column for these accounts with ones (Event value = 1). For example, you could append a list of "Closed-Won" deals, or accounts that have opportunities for them, etc.

As much as possible, choose success examples that would have been in your prospect segment before they became successes. For example, if your prospects are businesses with 500+ employees, choose successes using the same criteria, that is, accounts with 500+ employees. 

To build a good model, your training file should have at least 10,000 records in it. If you don’t have that many companies in your prospect segment, consider supplementing your training data with companies found in your lead database.

 

Step 2: Chose the Account Fit model type

Navigate to Models, click Create Model, then choose the Custom Event model type.

 

Step 3: Select a segment for scoring

Select the segment of accounts you want to augment with Account Fit scores, then click Next.

Step 4: Upload training file and select training options

Upload the training file that you created in Step 1. Check the Account Model Training Options. In most cases, you can use the default settings. Click Next.

Sometimes changing the Account Model Training Options can help you build a better model:

• One record per account: when checked, Rev.Up will make the best effort to deduplicate your accounts after matching them with the D&B Data Cloud. Only uncheck it when your first party account data is highly site-specific, or if you prefer to use your own deduplication method before uploading the training data. 
• Include personal email domains: when checked, Rev.Up will accept all email domains as they are specified in the training file. Try unchecking this when you are supplementing your training data with lead records that may have personal email domains in place of company information.
• Use curated attributes: when checked, Rev.Up will attempt to use built-in transformations to curate better predictor attributes and enhance the model. Try unchecking it when you see curated predictors in the model that do not make business sense for your use case (e.g., curated territories when your sales teams are split by territory).

 

Step 5: Identify columns to be used for matching

The Field Mapping screen displays the fields in the uploaded file that Rev.Up was able to match to standard fields in D&B Data Cloud. Verify the current mapping and edit it if necessary.

Map any additional unmapped fields if possible. You must map all the required fields. Mapping as many fields as possible will allow for a better match.

Step 6: Start the modeling job

Enter a meaningful model name and any notes about your model. Click Next. 

Step 7: Wait for the modeling job to complete

Review your model job. Confirm all of your settings, and monitor job progress.

 

You can also track the current jobs and the job history by clicking the Admin menu, clicking the Jobs item and selecting the Model Jobs tab.

Modeling jobs can take several hours to complete. You will receive an email when your job has completed.

 

Step 8: Review model

Navigate to the Models page to find your newly trained model. You can rename your model by clicking on the menu icon in the tile’s top right corner.

Click on the tile of your new model to go to the Dashboard page for that model. Here you can review the new model and activate scoring. 

Locate the new model iteration in the Creation History table and click View Model. This will open the Attribute List tab. In addition, Attribute Analysis, Performance and Model Summary tabs will be available for your model. These pages provide lots of information you can use to review the quality of your model. It is important to review the model before activating it to ensure that it will perform well for your use case. See the article “How to review an Account Fit model” to learn more about the charts and data in these pages.

 

Step 9: Activate scoring

In the Creation History table of the model Dashboard page, hover over the newly created model iteration and click the Activate button that appears for that iteration. Alternatively, you can click on the Ratings tab.

This displays the Rating Setup tab with where you can see the the distribution of lift for every model score value for your target accounts. The current bounds, lifts and the number of accounts for each rating are also displayed. 

You can drag the boundary separator lines in the graph to the left or right to tune the score ranges that are assigned to each rating.

Ratings A, B, C and D are present by default. Click anywhere in the graph to add additional separator lines. This will create additional ratings E and F. Drag a separator line down below the graph to remove it and reduce the number of ratings. 

Click Activate Configuration to activate the scoring with this model iteration. The next time you run the Data Processing & Analysis job, the model scores and ratings will be calculated and saved for your accounts. After that you can use the model scores and ratings to create new segments, rule-based models, or plays.

 

Using and maintaining an Account Fit model

If you no longer need the model to calculate new scores, use the Deactivate Scoring button on the Dashboard tab. This shortens the duration of the Data Processing & Analysis job. 

All scoring configurations that were previously active for this model can be viewed in the History tab. 

Use the Notes tab to add, edit or delete any notes about your model. 

If you want to rebuild a model using any changed settings or attributes, view a model iteration and click Remodel on the Attribute List tab. This will train a new iteration of the model which you can later review and activate.