Improving Loan Application Quality Without Raising Credit Risk
1. Executive Summary
The company’s rising verification spend, combined with the fact that many SME applicants are ultimately rejected, indicates a funnel design problem rather than only a credit assessment problem. Too many low-probability applicants are being allowed to progress into costly verification before the business has enough evidence that they fit policy, risk appetite, and target economics.
The practical implication is clear: verification is being used too often as a discovery mechanism for poor-fit applicants, when it should mainly be used to confirm already promising ones.
Our recommendation is to redesign the front of the funnel around risk-adjusted triage:
- use low-cost signals earlier;
- segment by source campaign and credit score;
- set explicit “verification-worthy” thresholds;
- improve applicant expectation-setting before submission;
- create a closed-loop feedback system from rejection reasons and defaults back into acquisition and pre-screening.
This approach supports the stated goal: improve application quality and approval rate without increasing credit risk in an uncontrolled way. The target is not maximum approval volume. The target is a higher share of applicants entering verification who are both approvable and consistent with the firm’s portfolio growth and risk appetite under OJK constraints.
2. Corrected Problem Diagnosis
The core issue should be defined more precisely as:
The company likely has an inefficient pre-screening and triage process that allows too many low-likelihood SME applicants from weaker channels or weaker fit segments to enter expensive verification, reducing operating efficiency and depressing approval conversion.
This diagnosis is stronger than saying “verification costs are rising,” because it identifies where the system is breaking:
- Acquisition quality is uneven:
Some campaigns may be optimized for lead volume or low CAC, not for approved-and-performing loans.
- Eligibility and fit may be unclear upstream:
Applicants may be applying before understanding documentation requirements, common rejection drivers, or whether the product fits their business profile.
- Pre-screen thresholds may be too generic:
The same screening logic may be applied across heterogeneous channels and applicant types, despite observable differences in approval and default outcomes.
- Verification may be triggered too early:
Low-cost information such as campaign source, available credit score, and historical rejection patterns may not be used aggressively enough before manual or higher-cost checks begin.
- Learning loops may be weak:
Rejection reasons and default outcomes may not be feeding back quickly enough into growth, channel management, and front-end product design.
So the objective is not to “approve more people” in a blanket sense. It is to increase the quality of applicants entering costly stages, which should improve approval rate among verified applicants while keeping default risk aligned with policy and growth targets.
3. Evidence Base and What It Does / Does Not Prove
What the internal data can support
The company has the right core data to diagnose the funnel:
- source campaign;
- credit score;
- approval and rejection reason;
- default rate;
- CAC.
With these variables, management should be able to answer high-value questions such as:
- Which acquisition sources produce applicants with the best approval/default economics?
- At what credit score bands does verification become inefficient?
- Which rejection reasons could have been identified before costly verification?
- Which channels have low CAC but poor downstream approval or default performance?
- Which combinations of source + score + rejection reason indicate avoidable verification spend?
This is enough to support a practical triage redesign, even if more granular operational data would improve precision.
What the cited evidence contributes
Most listed references are not direct evidence on SME credit funnel optimization. They do, however, loosely reinforce several implementation themes:
- Hyper-personalization and engagement design (Centi, 2018):
Suggests that tailored user journeys can influence completion behavior and engagement. Relevant for personalized pre-application education and document readiness prompts, but not proof of improved lending outcomes.
- Strategic alignment of enterprise systems (Taşkın, 2022):
Supports the need to align systems and functions. Relevant because acquisition, credit, and operations should share funnel objectives and data definitions.
- Organizational resistance to digital transformation (Mənin, 2020):
Relevant as a caution that operational redesign may fail without adoption by teams whose incentives currently favor volume or speed over qualification quality.
- SME perceptions of risk (Nurzhanova, 2025):
Suggests SME behavior is shaped by perceptions and uncertainty. Potentially relevant to applicant communication and trust, though not specific to Indonesian SME lending.
What the evidence does not prove
The current evidence does not prove:
- the exact score cutoff that should be used;
- which campaigns must be shut down;
- whether verification cost inflation comes more from process inefficiency or applicant mix;
- whether approval rate can increase materially without some trade-off in application volume;
- the causal impact of messaging changes on approval quality.
So the right immediate posture is test-and-learn, not a one-time hard reset.
4. Integrated Strategic Recommendation
We recommend a three-layer funnel redesign.
A. Build a “verification-worthy applicant” standard
Define explicit criteria for who should advance beyond low-cost screening. This should combine:
- source campaign quality;
- credit score band;
- likely approval probability;
- likely default risk;
- expected unit economics after CAC and verification cost.
This standard should become the common operating definition across growth, credit, and operations.
B. Introduce staged triage before full verification
Move from a broad funnel to a staged one:
- Eligibility and fit gate:
- Risk and economics gate:
- Verification gate:
Basic policy fit, segment suitability, and document readiness.
Simple rules using source, score, and historical rejection/default patterns.
Only applicants with sufficient expected approval probability and acceptable risk-adjusted economics proceed to costly checks.
This sequencing reduces spend on obviously poor-fit applications while preserving capacity for promising segments.
C. Reallocate acquisition toward quality, not raw volume
Shift channel optimization away from lead count and headline CAC alone. Evaluate campaigns on:
- approval rate after pre-screen;
- approval rate after verification;
- default performance;
- expected contribution after acquisition and verification costs.
Some low-CAC channels may be value-destructive if they fill the funnel with weak applicants.
D. Improve applicant guidance before submission
Use clearer eligibility messaging and document guidance to reduce premature or unsuitable applications. This is especially important in SME contexts, where applicants may not fully understand underwriting expectations.
The goal is not to discourage good applicants. It is to help borderline or unprepared applicants self-qualify, prepare better, or delay application until ready.
E. Establish a closed-loop management system
At minimum, track monthly:
- verification rate by source and score band;
- approval rate by source and score band;
- top rejection reasons entering verification;
- avoidable rejection rate;
- default indicators for newly approved cohorts;
- unit economics by funnel segment.
This turns the funnel into a managed portfolio rather than a series of disconnected decisions.
5. Marketing, Stakeholder, Operations, and Finance Implications
Marketing implications
- Redefine success metrics from lead volume to approved-and-performing applications.
- Rewrite campaign and landing-page messaging to clarify:
- who the product is for;
- minimum likely-fit criteria;
- required documents;
- common reasons applications fail.
- Personalize front-end guidance by source or applicant segment where feasible.
Stakeholder implications
- Ensure transparency and fairness in pre-screening logic to remain consistent with regulatory expectations and customer trust.
- Avoid overly opaque declines; where appropriate, provide simple reasons or readiness guidance.
- Align sales, growth, and credit teams around one funnel-quality objective rather than competing local metrics.
Operations implications
- Add low-cost screening checkpoints before manual review or external verification spend.
- Route applicants differently based on source and score quality.
- Use rejection-reason analysis to remove preventable process waste.
- Protect operational capacity for higher-quality applicants so service levels improve where conversion potential is strongest.
Finance and risk implications
- Measure funnel economics on a risk-adjusted basis, not just cost per lead or gross approval rate.
- Tighten or relax thresholds based on observed approval/default trade-offs, not intuition.
- Expect near-term application volume to potentially decline while verification efficiency and approval conversion improve.
- Ensure any threshold changes remain within risk appetite and OJK-compliant credit policy governance.
6. 30-60-90 Day Action Plan
First 30 days: diagnose and define the control points:
- Build a simple funnel fact base:
- application volume, verification rate, approval rate, rejection reasons, default indicators, CAC by source campaign and credit score band.
- Identify avoidable verification spend:
- isolate rejection reasons that could likely have been screened earlier.
- Define the first version of “verification-worthy”:
- create pragmatic rules using existing data only.
- Align governance:
- agree shared KPIs across growth, credit, and operations.
Days 31-60: pilot triage and messaging changes:
- Launch segmented pre-screen rules:
- differentiate by source quality and credit score bands rather than one broad standard.
- Adjust front-end communication:
- clarify eligibility, likely-fit criteria, and document readiness requirements.
- Create routing logic:
- fast-track stronger segments;
- hold back or decline weaker segments before costly verification.
- Start weekly monitoring:
- verification-to-approval conversion;
- rejection reasons after verification;
- channel-level unit economics.
Days 61-90: optimize and prepare scale-up:
- Compare pilot vs. baseline:
- verification cost per approved loan;
- approval rate among verified applicants;
- early risk indicators by approved cohort.
- Refine thresholds:
- tighten where avoidable verification remains high;
- relax selectively where approval quality is strong and growth economics support it.
- Reallocate channel spend:
- reduce spend on weak sources;
- increase support for sources with stronger approved-and-performing outcomes.
- Formalize operating cadence:
- monthly funnel review combining growth, operations, and risk.
7. Risks, Assumptions, and Validation Questions
Key risks
- Over-tightening the funnel:
- Channel misread:
- Fairness and transparency concerns:
- Internal resistance:
Approval rate may improve mechanically while total booked volume falls too sharply.
Historical performance may reflect temporary conditions rather than durable source quality.
Aggressive triage without clear governance may create customer trust or compliance issues.
Growth teams may resist changes if they are still measured on lead volume.
Core assumptions
- Source campaign and credit score meaningfully predict downstream approval and/or default economics.
- A material share of current rejection reasons is detectable before costly verification.
- There is sufficient consistency in historical data to support practical segmentation.
Validation questions
- Which rejection reasons are truly pre-screenable versus only knowable after verification?
- Which source-score combinations produce the worst verification leakage?
- How much approval-rate improvement is achievable without unacceptable volume loss?
- Are there segments with modest approval rates but excellent risk-adjusted economics that should be preserved?
8. Decision Checklist
Before approving the redesign, management should confirm:
- Do we agree the target metric is not raw application volume, but higher-quality verified applicants?
- Have we defined “verification-worthy” using both risk and unit economics?
- Are channel teams measured on approved-and-performing outcomes, not just CAC or leads?
- Have we identified the top rejection reasons that should be intercepted earlier?
- Have we designed pre-screen rules that are explainable and governance-ready?
- Do we have a pilot plan with clear success and stop criteria?
- Are risk, operations, and growth leaders jointly accountable for the results?
- Have we set guardrails to ensure compliance with OJK requirements and internal risk appetite?
9. References Used
- Centi, A. (2018). *Participant Engagement with a Hyper-Personalized Activity Tracking Smartphone App*. iProceedings. https://doi.org/10.2196/11876
- Nurzhanova, A. (2025). *SME perceptions of global risks: Survey-based evidence from Kazakhstan*. Problems and Perspectives in Management. http://dx.doi.org/10.21511/ppm.23(4).2025.10
- Pranata, S. (2025). *Increasing awareness and implementation of green marketing for SMEs in supporting sustainable development in Cirebon City*. Aspirasi Masyarakat. https://doi.org/10.71154/f1ntkf73
- Toma, F. (2022). *Assessment of Geography Teaching-Learning Process Through Game, in Pre-University Education*. Studia Universitatis Babeș-Bolyai Geographia. https://doi.org/10.24193/subbgeogr.2022.07
- Taşkın, N. (2022). *An Empirical Study on Strategic Alignment of Enterprise Systems*. Acta Infologica. https://doi.org/10.26650/acin.1079619
- Ivashchenko, V. (2023). *Distance Lexicographic Discourse: Media Platforms of Communicative Interaction*. Integrated Communications. https://doi.org/10.28925/2524-2644.2023.161
- Harada, Y. (2021). *Quantitative evaluation of visual guidance effects for 360-degree directions*. Virtual Reality. https://doi.org/10.1007/s10055-021-00574-7
- Pang, Y. (2025). *Exploring Chinese Sustainable Path Based on International Sustainable Disclosure Guidelines*. Accounting, Marketing and Organization. https://doi.org/10.71204/3357xh49
- Mənin, Х. (2020). *Change of organizational culture of the enterprise as overcoming of digital transformation resistance*. Vestnik Universiteta. https://doi.org/10.26425/1816-4277-2019-12-66-70
- Andreichenko, I. S. (2023). *A View of an Adult Person with ADHD from the Standpoint of Transactional Analysis*. Transactional Analysis in Russia. https://doi.org/10.56478/taruj20233332-35