The business fit is an important dimension of B2B targeting, enabling Sales and Marketing to focus spend and effort on prospects with budget and need. Used with first and third party engagement information, Account Fit scoring allows your ABM program to scale across priority tiers and channel spend accordingly.
In Atlas, use Custom Event modeling to build an Account Fit Rating for your ABM audiences. You can build your training set by exporting a Business Segment you want to score, adding your best Fit Accounts from the past, and loading into Atlas for machine learning.
Atlas builds your Fit model and adds scores to your Business Segment. Now you’re 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
Create a Segment in Atlas for the Accounts you want to augment with Fit scores.
Create a Segment just for prospects (who have never purchased) that doesn’t include any current or past customers. Current and past customers are at a different place in their journey with you, and need different campaign treatment.
Step 1: Create your training set
Create a CSV file of your training data. The training CSV file should have the following fields:
- Account!D (required) - same ID used in Atlas Account Import
- Event (required) - Flag with value 1 (marks successful past target) or value 0 (unsuccessful or untried prospect)
- Website (optional when using Company Name)
- Company Name (optional when using Website)
- City (optional)
- State (optional)
- Zip (optional)
- Country (optional)
- Phone Number (optional)
- DUNS (optional)
To create a training set that matches your prospect Segment, use Segment Export as a starting point. Add a column called Event, populated with zeros (value Event = 0) for your prospects. Now append a list of Accounts (at least 200, more is usually better) that exemplify the success you want to model, with Event populated with ones (value Event = 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 with the same logic - Accounts with 500+ employees.
To build a good model, your training file should have at least about 10,000 records in it. If you don’t have that many companies in your prospect segment, consider supplementing your training set with companies found in your lead database.
Step 2: Chose the model type
Navigate to Models, click Create Model, then choose Custom Event Model.
Step 3: Select your segment
Select your Segment of the accounts you want to augment with Fit scores, then click Next.
Step 4: Select modeling attributes
Atlas will build your model using the Lattice Data Cloud with your first-party attributes from My Data. You can turn either of these attributes off for the model by clicking its tile.
Step 5: Select the right model training options
Upload your training file from Step 2. Check your advanced modeling settings. In most cases, you can use the default settings. Click next to kick off the model training job.
Sometimes changing the advanced modeling settings can help you build a better model.
- One record per account: when checked, Atlas will make a best effort to dedupe your Accounts after matching with the Lattice Data Cloud. Only uncheck when your first party account data is highly site-specific, or you prefer to use your own dedupe logic prior to loading.
- Include personal email domains: when checked, Atlas will accept all domains as loaded in the training file. Try unchecking this when you are supplementing your training with lead records that may have personal domains in place of Company information.
- Use curated attributes: when checked, Atlas will attempt to use built-in transformations to curate better predictors and enhance your model. Try unchecking when you see curated predictors in your model that don’t make business sense for your use case (e.g., curated territories when sales teams are split by territory).
Step 6: Identify the columns in the training file to be used for matching
Map your required fields.
Step 7: Kickoff modeling job
Review your model job. Confirm all of your settings, and monitor job progress. You can also track the jobs and history by clicking on the gear at top right. Modeling jobs can take several hours to complete.
Step 8: Wait for the modeling job to complete
You will receive an email when your job has completed.
Step 9: Review Model
Log into Lattice to review your model and Activate Scoring. Navigate to the Models page to find your new Model tile. You can name your model by clicking on the menu in the tile’s top right corner.
Click on the Model tile to land on the details page for your model. The left navigation bar in this view links you to the Attributes, Performance, and Summary pages for your model. These pages provide you with lots of information you can use to review the quality of your model. See also “Get Started with Model Review” to learn more about the charts and data in these pages.
Step 10: Activate Scoring
Click Activate Scoring on the model dashboard to see what your model predicts for your target Accounts. You can see the Rating buckets, and change the bounds to tune your Ratings.
Grab x-axis tabs on the bucket separators to drag, add and remove the separators.
Lattice shows your model performance as Lift and Account Volume in each Rating bucket.
Click Publish Configuration to create a Scoring Action in your Process & Analyze queue. The next time you run Process & Analyze, your Ratings and Scores will be updated in My Data. You can use them to make new Segments, Rules-Based Ratings, or Plays.