Creating a new decision model from scratch

Understanding decision models in Lendsqr

A decision model in Lendsqr is a structured framework that defines how loan applications are evaluated within the system. It enables lenders to translate their credit policies, eligibility rules, and risk appetite into a set of automated conditions that determine whether a loan application is approved or rejected.

Rather than relying on manual reviews, decision models allow lenders to automate the evaluation process. This ensures that every application is treated consistently and according to predefined rules. It also reduces the time required to process applications and minimizes the likelihood of human error.

For example, a lender may want to create a loan product that is only accessible to individuals who are married and own their own homes. Instead of manually verifying each application, the lender can configure a decision model that checks for these conditions automatically. If an applicant meets the criteria, the system proceeds with approval logic. If the applicant does not meet the criteria, the system rejects the application or routes it for further review depending on the configuration.

This approach allows lenders to scale their operations efficiently while maintaining strict control over risk exposure.

How decision models support risk assessment

Decision models serve as the backbone of risk assessment in Lendsqr. Every loan product created within the platform relies on a decision model to define how applications are evaluated.

Each model is composed of multiple modules and variables that work together to form a complete decisioning system. These modules represent different aspects of borrower evaluation, such as eligibility checks, financial assessments, and data verification processes.

By configuring these modules, lenders can define the exact conditions under which a borrower qualifies for a loan. This includes setting thresholds, defining acceptable criteria, and determining how different data points influence the final decision.

Because decision models are tied to specific loan products, lenders can create multiple models for different use cases. For instance, a short-term consumer loan may have a very different decision model compared to a business loan. This flexibility allows lenders to tailor their offerings to different customer segments while maintaining control over risk.

Key considerations before creating a decision model

Before creating a decision model from scratch, it is important to understand how the system initializes and how different components interact during configuration.

When a new decision model is created, Lendsqr automatically provides a default system model. This default model includes all the modules that currently exist within the platform, along with their default values. It serves as a foundation upon which lenders can build and customize their own decision logic.

Some of these modules operate entirely within the Lendsqr system, while others depend on external service providers. Modules that rely on external providers require additional setup, including API key configuration and alignment with the provider’s pricing model. Because of this, lenders are advised to contact Lendsqr support staff before attempting to configure such modules.

Another important consideration is the structure of the configuration process. Decision models are built in segments, and each segment represents a portion of the overall logic. It is essential to save each segment before moving to the next one. This ensures that all configurations are preserved and reduces the risk of losing progress during setup.

Creating a model from scratch

Creating a decision model from scratch involves a sequence of clearly defined actions within the Lendsqr admin console. Each step must be followed carefully to ensure that the model is properly created and ready for configuration.

Step 1: Log in to your lender admin console

Begin by accessing your Lendsqr admin console through your web browser. Enter your login credentials and ensure that you have the appropriate permissions to create and manage decision models.

This step establishes access to the system and ensures that you can interact with the product management features required for model creation.

Step 2: Navigate to the decision model tab

Once you are logged in, locate the left-hand sidebar of the admin console. Within this sidebar, find the Product Management grouping.

Under this section, click on the “Decision Model” tab. This page displays all existing decision models and serves as the central location for creating and managing them.

Step 3: Click on “Add a new model”

On the decision model page, locate the “Add a new model” button and click on it.

This action initiates the process of creating a new decision model and opens the configuration interface where you will define the model’s details.

Step 4: Enter the decision settings name and description

In the configuration interface, you will be required to provide basic information about the model.

Enter a clear and descriptive Decision Settings Name. This name should make it easy to identify the model later, especially if you manage multiple loan products.

Next, provide a Decision Settings Description. This description should explain the model’s purpose, the type of borrowers it is intended for, and any key rules that define its logic.

Once you have entered both fields, save your progress before proceeding. This ensures that the model is successfully created and that you can continue with further configuration.

After saving the initial details, you can begin configuring the modules that make up the decision model. Watch the video below to learn how to configure your decision models.

 1. Configuring your decision model (Internal integrations)

 2. Configuring your decision model (External integrations) 

For more information on payment gateways available to lenders, Read here

Read more: What’s a decision model?

Configuring decision models with integrations

Decision models can be enhanced through the use of integrations, which allow the system to incorporate additional data into the decision-making process.

Internal integrations rely on data that already exists within the Lendsqr system. These are typically easier to configure and do not require external setup. They are useful for evaluating existing borrower information and applying system-based rules.

External integrations, on the other hand, involve third-party providers. These integrations enable lenders to access additional data sources, thereby improving the accuracy of risk assessments. However, they require proper setup, including API key configuration and coordination with Lendsqr support.

Practical tips for configuring internal vs external integrations

Configuring integrations within a decision model requires a clear understanding of how internal and external data sources behave within the Lendsqr environment. While both types of integrations enhance decision accuracy, they differ significantly in setup, dependency, and operational impact. The following practical tips will help ensure that your configurations are effective, reliable, and aligned with your lending objectives.

Internal integrations

Internal integrations rely on data that already exists within the Lendsqr system. Because they do not depend on third-party providers, they are generally easier to configure and faster to implement. However, their effectiveness depends on how well your internal data is structured and maintained.

One important consideration when working with internal integrations is data consistency. Ensure that the data fields you are relying on are consistently populated across all user records. For example, if your decision model checks for a borrower attribute such as marital status or employment category, inconsistencies or missing values can lead to inaccurate decisions. Before configuring rules, it is advisable to audit your existing data to confirm that it is complete and reliable.

Another practical tip is to align your internal rules with actual user behavior rather than assumptions. Internal data often reflects historical interactions, so it is useful to base your conditions on patterns that have proven to correlate with repayment performance. This helps ensure that your decision model is grounded in real outcomes rather than theoretical criteria.

It is also important to keep your internal logic simple and interpretable. Since internal integrations are fully within your control, there may be a temptation to layer multiple conditions across different modules. While this can increase precision, it can also make the model difficult to debug and maintain. Start with a few key variables that have the strongest impact, and expand gradually as needed.

Testing plays a critical role when configuring internal integrations. Because you have direct access to the data, you should run sample evaluations to see how different borrower profiles are treated by the model. This allows you to identify unintended exclusions or approvals before the model is applied in a live environment.

Finally, ensure that internal integrations are regularly reviewed and updated. As your user base grows and borrower behavior evolves, the assumptions built into your model may become outdated. Periodic reviews help keep your decision logic relevant and effective.

External integrations

External integrations introduce additional data sources into your decision model by connecting to third-party providers. These can significantly improve the accuracy of your risk assessment, but they require more careful planning and coordination.

The first practical tip when working with external integrations is to engage Lendsqr support early in the process. External modules often require API key configuration and alignment with provider-specific requirements. Attempting to configure these without proper setup can lead to errors or incomplete data retrieval. Working with support ensures that the integration is correctly implemented from the start.

It is also important to understand the cost implications of external integrations. Many third-party providers charge per request or per data pull. Before enabling these modules, review the pricing model and ensure that it aligns with your business objectives. You may need to limit when and how often certain checks are triggered to manage costs effectively.

Another key consideration is response reliability. External services depend on network connectivity and provider uptime. When configuring your decision model, think about how the system should behave if an external service fails or returns incomplete data. You may need to define fallback rules or alternative decision paths to prevent disruptions in the loan approval process.

Latency is another factor to keep in mind. External data requests can introduce delays in decision making, especially if multiple providers are involved. To manage this, prioritize the most critical checks and avoid unnecessary duplication of external calls. This helps maintain a balance between decision accuracy and processing speed.

Data interpretation is equally important when working with external integrations. Different providers may structure their data in unique ways, and it is essential to understand what each data point represents before incorporating it into your decision logic. Misinterpreting external data can lead to incorrect approvals or rejections.

Testing external integrations requires a slightly different approach compared to internal ones. In addition to validating decision outcomes, you should also verify that API calls are functioning correctly and that the returned data matches expectations. This includes checking for edge cases such as partial responses or unexpected values.

Lastly, ensure that you monitor the performance of external integrations over time. Track metrics such as success rates, response times, and impact on decision outcomes. This will help you identify any issues early and make necessary adjustments to maintain system reliability.

Quick checklist for validating a new decision model before going live

Before deploying a new decision model, it is important to validate that it behaves exactly as intended under real-world conditions. This checklist is designed to help you quickly confirm that all critical elements have been reviewed and tested. It can be used as a final gate before linking the model to a live loan product.

Model setup and structure

Confirm that the decision model has been properly created and saved. Ensure that the Decision Settings Name and Description are clear, accurate, and aligned with the intended loan product.

Verify that all required modules are present. Since the model is based on the default system structure, confirm that no critical module has been unintentionally removed or left unconfigured.

Check that each segment within the model has been saved. Unsaved segments can result in missing logic during execution.

Rule accuracy and alignment

Review all configured rules to ensure they reflect your actual lending criteria. Each condition should map directly to a clearly defined eligibility requirement or risk rule.

Confirm that there are no conflicting conditions. For example, rules that simultaneously allow and restrict the same borrower attribute can lead to unpredictable outcomes.

Validate that thresholds and conditions are correctly set. Ensure that values such as income limits, borrower attributes, or qualification criteria match your intended policy.

Internal data validation

Ensure that all internal data fields used in the model are consistently populated across borrower records. Missing or inconsistent data can lead to incorrect decisions.

Check that the variables referenced in the model correspond to the correct data fields within the system. Misaligned fields can cause the model to evaluate the wrong information.

Run sample checks using existing borrower data to confirm that the model produces expected outcomes based on known profiles.

External integration readiness

Confirm that all required external integrations have been properly configured. This includes API keys, provider setup, and any required approvals.

Verify that each external module is returning valid and complete data. Test API responses to ensure they align with expectations.

Review the pricing model for each external provider and confirm that usage levels are acceptable for your expected loan volume.

Define fallback behavior for cases where external services fail or return incomplete data. Ensure the model can still proceed without breaking the decision flow.

Performance and response checks

Test the model’s response time to ensure that decisioning is completed within an acceptable timeframe. This is especially important if external integrations are involved.

Check that the model does not trigger unnecessary or duplicate checks, particularly for paid external services.

Ensure that the overall decision flow is efficient and does not introduce avoidable delays in loan processing.

Scenario testing

Run multiple test scenarios that reflect different borrower profiles. Include both expected approvals and expected rejections.

Test edge cases, such as borderline applicants who barely meet or fail certain criteria. This helps confirm that thresholds behave as intended.

Validate that the model consistently produces the same outcome for the same input data. Consistency is critical for reliable decision making.

Review and approval

Conduct a final review of the entire model configuration. Ensure that all modules, rules, and integrations align with your risk strategy.

If applicable, have another team member review the model. A second perspective can help identify issues that may have been overlooked.

Confirm that the model is properly linked to the correct loan product and not mistakenly assigned elsewhere.

Go-live readiness confirmation

Ensure that monitoring processes are in place to track the model’s performance after deployment. This includes tracking approval rates, rejection rates, and any anomalies.

Verify that internal documentation has been updated to reflect the new model. This should include its purpose, rules, and associated loan product.

Confirm that support teams are aware of the new model and understand how it functions in case issues arise after launch.

Once all items in this checklist have been completed and validated, the decision model can be confidently deployed for live loan processing.

Best practices for creating decision models

Creating a decision model is not just about following steps. It also requires thoughtful planning and careful execution to ensure that the model performs effectively in real-world scenarios.

One important practice is to clearly define your eligibility criteria before you begin configuration. Having a well-documented understanding of who qualifies for a loan helps you translate those requirements into accurate system rules.

It is also advisable to start with a simple model and expand over time. Trying to include too many conditions at once can make the model difficult to manage and more prone to errors. A simpler model is easier to test, maintain, and improve.

Testing is another critical aspect of building effective decision models. Before applying a model to live loan applications, it is important to validate its logic using sample data. This helps identify any gaps, inconsistencies, or unintended outcomes.

When working with external integrations, always collaborate with Lendsqr support. External modules often require technical setup and pricing considerations, and proper guidance ensures that the integration is configured correctly.

Finally, maintain clear internal documentation for each decision model. This should include its purpose, key rules, and the loan products it supports. Proper documentation makes it easier for teams to understand, manage, and update models over time.

Conclusion

Decision models are a fundamental component of the Lendsqr platform, enabling lenders to automate and standardize their loan approval processes. By defining clear rules and leveraging both internal and external data, lenders can ensure consistent and efficient decision-making.

Creating a model from scratch involves a structured process that includes setting up the model, configuring its modules, and carefully reviewing its logic. Each step plays an important role in ensuring that the model functions as intended.

By following best practices and paying close attention to detail during configuration, lenders can build decision models that not only improve operational efficiency but also strengthen their overall risk management strategy.