Matching Models are the methods used to detect duplicate records. DataGroomr provides two options for duplicate detection, Classic matching model (exact/similar) and machine learning based matching model. 


Accessing Matching Models


To access matching models, select Settings under Dedupe on the Navigation Menu.



From the Settings page, click Matching and then select your desired object. 


Creating a Matching Model


There are two types of Matching Models in DataGroomr. Machine Learning Models or Classic Matching Models

Press ADD MODEL button and select desired model type:


Updated screenshot



Classic Matching Model

  • Rule-based approach: You manually define which fields to compare (e.g., First Name, Email, Company) and assign weights to each.

  • Deterministic logic: Matches are found based on exact or fuzzy logic using predefined thresholds.

  • Transparent scoring: You can clearly see which fields influenced a match and by how much.

  • Best for: Simple or predictable data scenarios where you want full control over match logic.


Machine Learning Matching Model

  • AI-assisted approach: Uses a machine learning algorithm trained on labeled examples of duplicates and non-duplicates.

  • Pattern recognition: Learns complex relationships between fields that may not be obvious or linear.

  • Adaptive: Gets better with more training data (you label examples, the model adjusts).

  • Confidence scores: Matches are scored based on statistical confidence, not rigid rules.

  • Best for: Large or messy datasets where manual rules may miss patterns or create false positives.


Selecting either model option will open a dialog that includes the following elements, giving you the option to manually select fields or let AI infer most suitable fields for your object based on the best practices and the statistics of the data population. The AI recommendation also takes your existing dictionaries (synonyms and ignore words) and field grouping into account, so the suggested model comes back as a complete setup rather than a list of fields only.



Updated screenshot


If you chose 'Select Fields Manually', you will see below elements:

  • Name - enter a unique name for your model or generate one via AI button
  • Fields - select all fields that should be included as part of this model.
  • Field Sets - contains a list of Salesforce's configured matching rules and sets of fields that DataGroomr recommends based on commonly matched fields for that object for you.

Press Save button to create a model.


To configure the fields of the new model, see Edit a Matching Model.


Machine Learning Matching Model


Machine Learning Models are based on algorithms powered by machine learning.  Before ML model can be used in a dataset it will need to be Trained. Click Train button to start training. 


There are two options available for training machine learning models


  • Train by AI Assistant (recommended) autonomous training and fine-tuning the machine learning model by AI. If you choose this option, once AI has profiled and trained ML model, you may apply directly to specified dataset.
  • Train Manually is a process where you tell DataGroomr which records are duplicates and which are not, in order to teach the algorithm to identify patterns in data based on the fields you specify in the Set Up process.


Learn more: Training a machine learning model


Editing an Existing Model


To edit an existing model, select it and then press the Open button. For field settings and which models can be edited, see Edit a Matching Model.



Cloning a Model


Any model maybe cloned to generate a new copy.  To do this, select the rule and then press the CLONE button. 



Change Model Type When Cloning

You can now choose the model type during the cloning process.
When cloning, select whether the new model should be a Machine Learning (ML) model or a Classic model.

  • All existing field mappings, weights, and confidence thresholds are preserved.

  • The new model will adopt the logic of the selected type.

This feature allows you to test the same dataset across both model types without recreating your configuration from scratch.




Deleting Model


A model can be deleted by selecting rule and then pressing the Trash button.



Assigning to Datasets


Classic matching models and trained machine learning models can be assigned to datasets or can be designated as default model for your organization.


From the Matching Models feature, select a model press the Assign button.  Then choose the datasets to apply and press Assign button.



Good to Know: Matching models can also be assigned directly int Dataset Configuration window.


To set a model as a default, select the rule and then press the SET DEFAULT button. The green 'Default' label will be displayed next to that model.