How an Income Estimator API Adds Context to Thin-File Lending Decisions
Income is an important part of lending decisions, but it is not always easy to assess early in a digital application. Some applicants may not have a traditional payslip on hand. Others earn through freelance work, online platforms, or several sources. Asking every applicant for more documents can slow onboarding and create unnecessary friction.
An income estimator API gives lenders another way to assess an application before deciding whether more verification is needed.
Smile API’s Income Estimator uses behavioral signals to estimate the likelihood that a user belongs to a higher-income segment. With the applicant’s consent, a lender submits an email address and country code. The API returns a synchronous response that can be added to an existing onboarding, credit, or risk workflow.
The result is not an exact salary calculation. It is an additional signal designed to help lenders segment applicants, prioritize reviews, and decide when verified income documents are still necessary.
Why traditional income checks can create friction
Traditional income verification remains important, especially when a lender needs definitive evidence. But requiring documents too early or from every applicant can introduce several challenges:
- Applicants may need to find, download, or photograph a payslip or bank statement.
- Submitted documents may require manual review.
- Non-salaried or gig-economy workers may not have standard proof of income.
- Extra steps can increase the time required to complete an application.
- Thin-file applicants may have limited conventional data even when they have the capacity to repay.
The goal of an income estimation signal is not to remove verification. It is to help a lender use verification more selectively. When the estimator returns a weaker or less certain result, the lender may request more evidence. When the signal is stronger, the lender can use it alongside its other data and policies to determine the appropriate next step.
What is Smile API’s Income Estimator?
Smile API’s Income Estimator is an email-based income estimation API for lenders and fintech companies. It predicts the likelihood that a user belongs to a higher-income group and maps that prediction to a predefined income tier.
The API returns four main outputs:
1. Probability
Probability is the model’s primary predictive output. It represents the likelihood that the user belongs to the higher-income segment.
For decisioning, this value carries more information than the simplified tier. Two applicants may fall into the same tier but receive different probabilities, allowing a lender to distinguish between a stronger and a weaker model signal.
2. Income tier
The tier translates the model output into a simpler category, such as Low, Middle, or High. It can support customer segmentation, user-interface displays, portfolio reporting, or straightforward business rules.
A tier should not be interpreted as the applicant’s verified or exact monthly income.
3. Income range
The income range shows the predefined reference boundaries associated with the returned tier. It provides context for the category, but it is not a personalized salary estimate. Applicants assigned to the same tier receive the same reference range.
4. Confidence
Confidence describes the model’s certainty relative to its internal decision boundary. It does not mean the same thing as probability.
For example, the model may be highly confident that a user is unlikely to belong to the higher-income segment. In that case, the probability may be low while confidence is high. Risk and data teams should interpret the two fields separately.
How the Income Estimator API fits into a lending flow
Income Estimator can be introduced at the application or early assessment stage:
- The applicant provides consent for the relevant data processing.
- The lender sends the applicant’s email address and country code to the API.
- The API returns the probability, tier, reference range, and confidence level synchronously.
- The lender combines the result with its existing credit, fraud, identity, and policy checks.
- The workflow determines whether to continue, prioritize manual review, or request verified income documents.
Because the response is synchronous, the signal can support real-time journey segmentation without creating a separate applicant-facing step.
Possible uses include:
- identifying applications that need additional income verification;
- prioritizing files for manual review;
- supporting different onboarding paths based on the strength of the signal;
- enriching thin-file applications with an additional data point;
- comparing model results with portfolio outcomes during a backtest; and
- supporting reporting and portfolio segmentation.
The final policy and thresholds should be calibrated using the lender’s own historical outcomes, risk appetite, customer profile, and regulatory requirements.
For broader context, the World Bank discusses how alternative data can support lending decisions.
Income estimation and income verification are not the same
The distinction matters.
Income estimation uses a model to produce a probability and an income segment. It is useful when a lender needs an early signal and wants to decide whether additional checks are warranted.
Income verification relies on evidence such as payroll data, employment records, payslips, tax documents, or bank transactions. It is appropriate when the lender must confirm reported income or meet a specific policy requirement.
These approaches can work together. Income Estimator can help determine where verification adds the most value, while verified data can provide stronger evidence where the lender’s policy requires it.
Responsible ways to use an income estimation signal
Model outputs are most useful when their role is clearly defined. Smile API recommends that lenders treat Income Estimator as one input within a broader decision framework.
Good implementation practices include:
- Do not treat the result as exact monthly income. The returned range represents tier boundaries, not a personalized salary prediction.
- Do not use it as the sole approval criterion. Combine it with the lender’s other credit, identity, fraud, affordability, and policy checks.
- Use probability as the main model signal. Tier is better suited to simplified segmentation and reporting.
- Review confidence separately. A high-confidence result may support either a high or low probability prediction.
- Set thresholds using portfolio evidence. Test how different probability bands perform against actual outcomes before applying them in production.
- Request consent and provide appropriate notice. The lender should ensure that its collection and use of applicant data has an appropriate legal basis and follows applicable privacy requirements.
- Maintain a review path. Applicants should not be unfairly excluded by a single inferred signal, particularly when other evidence is available.
For Philippine deployments, the National Privacy Commission provides the text of the Data Privacy Act of 2012 as a reference for data-processing principles and data-subject rights.
How lenders can evaluate Income Estimator
Income Estimator is currently available as a Smile API product preview for selected customers. Preview products can be assessed through an evaluation or backtest before broader deployment.
A practical evaluation can answer questions such as:
- How often does the API return a usable result for the lender’s target applicants?
- How does probability relate to verified income or relevant portfolio outcomes?
- Which thresholds create useful segments for the lender’s risk strategy?
- Where does the signal reduce unnecessary document requests?
- Which results should trigger additional verification or manual review?
- Does the signal add value beyond the lender’s existing data?
The purpose of testing is not to copy generic thresholds. It is to understand how the signal behaves for the lender’s own market, customer base, and decision process.
Add another signal without adding another applicant step
For lenders serving applicants with limited conventional data, the decision is rarely between “enough data” and “no data.” More often, the challenge is deciding what to verify next without adding the same friction to every application.
Smile API’s Income Estimator gives risk teams an additional, synchronous signal from an email address. Used responsibly, it can help segment applications, focus manual review, and identify when verified income evidence is still needed.
Income Estimator is not a replacement for sound underwriting or verified income documentation. It is a way to make the next step in the lending journey more informed.
Interested in evaluating Income Estimator with your portfolio? Contact the Smile API team to discuss preview access and testing.