A 15 percentage point improvement in application completion rate — from 28% to 43% on a 2,400-application-month pipeline — is not an operational achievement. It is a revenue event. Those 360 additional monthly completions, at an average disbursement of ₹35 lakhs, represent ₹126 crores of incremental monthly lending volume. At a net interest margin of 2.60%, that is ₹3.3 crores of additional annual NIM from applications that were already in the pipeline, already qualified, and already lost to an obstacle the Drop-Off Agent AI would have resolved.
The baseline: what a 28% completion rate costs the institution every month
A loan application that reaches the form-fill stage represents significant prior investment: the cost of the marketing or DSA channel that generated the lead, the SDR AI qualification time, the RM or AE touchpoint, and the credit bureau pull. By the time a borrower reaches the application form, the institution has typically spent ₹800 to ₹2,400 per lead depending on the acquisition channel. A 28% completion rate means that for every 100 applications started, 72 of the acquisition-cost investment produces nothing — not even the information about why the borrower dropped.
Worse, dropped applications do not simply disappear — they move. A borrower who starts a home loan application on the institution's platform and cannot complete it will, in most cases, try a competitor within 48 hours. The institution has paid to acquire a borrower who then converts at a competitor. The drop-off cost is not just the wasted acquisition spend — it is the value of the loan that a competitor disburses instead.
The ROI model: three scenarios
Conservative scenario — 10pp improvement: 28% → 38%
240 additional completions/monthBase scenario — 15pp improvement: 28% → 43%
360 additional completions/month · Primary projectionOptimistic scenario — 20pp improvement: 28% → 48%
480 additional completions/month · At full assist library deploymentThe full revenue impact: base scenario modelled across the year
Where the 15pp improvement comes from: the source breakdown
A 15 percentage point improvement in application completion is not achieved uniformly — it comes disproportionately from the highest-abandonment sections, where the specific assist has the highest recovery rate. Income declaration (the 23% drop at the funnel's steepest point) recovers 38% of its abandonments with the SE assist. Document upload (19% drop) recovers 58% with the WhatsApp alternative and password fix combination. Aadhaar OTP (8% drop) recovers 53%. Consent hesitation (6% drop) recovers 39%.
Applying these recovery rates to the actual drop volumes produces the 15pp overall improvement: 23% × 38% recovery = 8.7pp from income section alone. 19% × 58% recovery = 11pp from document upload. But these recoveries are not additive from the same pool — each percentage point of improvement represents borrowers who would otherwise have dropped at that section and not returned. The 15pp aggregate reflects the weighted recovery across all four high-abandonment sections with realistic assumptions about overlap and section sequencing.
The 15pp figure is achievable in the first 90 days of deployment for an institution that currently sends no abandonment re-engagement communications. For institutions that already send some form of generic re-engagement, the incremental improvement from targeted specific assist is typically 8–12pp above the existing baseline.
The secondary value: what recovered applications reveal about the funnel
Each abandonment event that the Drop-Off AI detects and logs — whether the recovery succeeds or not — is a data point about the application form. If 23% of applications abandon at the income declaration section, and the majority of those are self-employed borrowers, that is a clear signal that the income declaration section needs a product design fix: a dedicated SE income path with appropriate field labels, not just a re-engagement message. The Drop-Off AI's abandonment data, aggregated over 30 days, produces a prioritised list of product design improvements that would reduce the abandonment rate structurally — rather than relying on re-engagement messages to patch a broken experience.
This secondary value compounds the ROI: an institution that uses drop-off data to improve its application form over 12 months will see the baseline completion rate improve from 28% to 35% through design improvements alone, before the Drop-Off AI's re-engagement is applied on top. The 15pp re-engagement improvement is then applied to a higher baseline — producing 43pp of structural improvement plus 15pp of re-engagement improvement for a combined 50% completion rate.
The most efficient growth is from the pipeline already inside the funnel
An NBFC that wants to grow disbursements by ₹126 crores per month can either spend on marketing to increase the pipeline by 54% — or it can deploy the Drop-Off Agent AI and recover 15 percentage points of the pipeline it is already losing. The marketing route costs proportionally more per rupee of incremental disbursement than the recovery route, because the recovery route is working with borrowers who have already been acquired, already qualified, and already demonstrated intent by starting the application. They need an explanation and an alternative, not another impression. The Drop-Off Agent AI is the most capital-efficient growth lever available to an institution with an existing digital application pipeline — because it earns revenue from investment that has already been made.