Use case #0003

Conversion tracking: how Cross-Sell AI measures revenue attributed to each campaign

A cross-sell campaign that is measured only by open rates and click-through rates has not been measured. Open rates tell the institution how many borrowers read the subject line. Click-through rates tell it how many borrowers were curious enough to look. Neither tells it how many borrowers took a new loan and how much interest income that loan will produce over its tenor. The Cross-Sell Campaign Agent AI measures every campaign against one metric: disbursed loan value and its associated NIM contribution, attributed to the specific trigger that prompted the outreach. Everything else — opens, clicks, applications — is diagnostic data that explains the conversion rate, not the measure of success.

A cross-sell campaign that is measured only by open rates and click-through rates has not been measured. Open rates tell the institution how many borrowers read the subject line. Click-through rates tell it how many borrowers were curious enough to look. Neither tells it how many borrowers took a new loan and how much interest income that loan will produce over its tenor. The Cross-Sell Campaign Agent AI measures every campaign against one metric: disbursed loan value and its associated NIM contribution, attributed to the specific trigger that prompted the outreach. Everything else — opens, clicks, applications — is diagnostic data that explains the conversion rate, not the measure of success.

The revenue attribution problem in cross-sell campaigns

Cross-sell attribution in lending has a structural challenge that does not exist in e-commerce: the conversion takes 14 to 21 days and passes through multiple teams and systems between the campaign touchpoint and the disbursement. A borrower who receives a WhatsApp on November 13 applies on November 15, goes through credit on November 18, is sanctioned on November 20, and disburses on November 22. That 9-day journey — from campaign touchpoint to funded loan — requires tracking a lead ID through the CRM, the LOS application, the credit system, and the disbursement record, matching each step back to the original campaign. Without that linkage, the marketing function knows the campaign went out but not what it produced.

The Cross-Sell Campaign Agent AI maintains a campaign ID → lead ID → LOS loan number mapping from the moment the outreach is sent. When a borrower receives a campaign and responds (via WhatsApp reply, email click, or inbound call), a campaign-tagged lead is created in the CRM with the campaign ID embedded. That ID travels through every system the application touches. When the loan is disbursed, the disbursement record carries the campaign ID — and the attribution is complete, traceable, and auditable.

"A campaign that produced 8 funded loans worth ₹1.4 crore in disbursements cost ₹7,400 to run. That is a return that makes the board meeting uncomfortable — uncomfortable with how little they have been spending on cross-sell relative to acquisition."

The November cross-sell campaign revenue dashboard

Cross-Sell Revenue Dashboard — November 2025 · All Campaigns
2,284 campaigns sent · 384 conversions (16.8%) · ₹8.4 Cr disbursed · ₹68.4L NIM contribution (10-year avg tenor)
Campaigns sent2,284
Conversions (funded)384 (16.8%)
Total disbursements attributed₹8.42 Cr
Campaign cost (send + RM time)₹3,12,000
Campaign type (trigger) Sent Conv% Disbursed NIM contr. ROMI
T01 — EMI calculator used 284 24.6% ₹1.24 Cr ₹10.1L 32×
T06 — MSME new GST state 84 28.6% ₹0.74 Cr ₹6.0L 38×
T03 — 12-month anniversary (zero DPD) 421 19.2% ₹1.88 Cr ₹15.3L 29×
T05 — MSME GST revenue +20% 218 22.0% ₹1.24 Cr ₹10.1L 34×
T02 — Voluntary prepayment made 184 17.4% ₹0.84 Cr ₹6.8L 27×
T08 — CIBIL score improved 50+ pts 142 13.4% ₹0.44 Cr ₹3.6L 18×
T04 — Outstanding below 60% 384 11.7% ₹1.08 Cr ₹8.8L 14×
All other triggers (T07/11/12/14/15/16/17) 567 8.6% ₹0.96 Cr ₹7.7L 11×
Total November cross-sell NIM contribution
₹68.4 lakh
vs campaign cost ₹3.12L → 21.9× ROMI · All triggers combined
Highest-ROMI trigger (T06 — new GST state)
38× ROMI · 28.6% conversion · ₹6L NIM from 84 campaigns
Cost: ₹8,200 · NIM: ₹6,00,000
● NIM contribution: 10-year average tenor assumed · 2.60% NIM on disbursed amount · All campaigns attributed via campaign ID → LOS loan number tracking · 48-hour attribution lag after disbursement

The revenue attribution flow: campaign ID to NIM ledger

01
Campaign sent · Campaign ID assigned

Every outreach carries a unique campaign ID linked to the trigger, the borrower, and the product offered

When the Cross-Sell Campaign Agent AI sends an outreach, it creates a campaign record: Campaign ID (CSELL-NOV-T06-0884), borrower ID, trigger type (T06 — new GST state), product offered (MSME WC top-up), offer amount (₹17.4L), send timestamp, and channel (WhatsApp). This record is the attribution anchor — everything that happens downstream is linked back to this campaign ID.

→ Campaign ID: CSELL-NOV-T06-0884 · Borrower: Kaveri Constructions · Created Nov 13 09:15
02
Borrower responds · Campaign-tagged lead created

Borrower's "YES" reply creates a CRM lead with the campaign ID embedded as the source

Kaveri Constructions replies "YES" to the WhatsApp on November 14. The response creates a CRM lead record with source field populated: CSELL-NOV-T06-0884. The RM is assigned and calls within 2 hours. The RM's call notes are logged against the same lead. The campaign ID travels with the lead through every subsequent system. If the borrower calls the support desk instead of replying directly, the agent who picks up can look up the campaign record and tag the inbound inquiry to the correct campaign.

→ CRM lead created: source CSELL-NOV-T06-0884 · RM assigned · Campaign ID embedded in all downstream records
03
Application submitted · LOS loan number created

Application in LOS carries the campaign ID as the origination source — linking the credit record to the campaign

The RM submits the application in the LOS on November 15. The LOS loan number (LA-2025-XXXX) is created. The origination source field in the LOS is populated with CSELL-NOV-T06-0884 — the campaign ID. Credit assessment runs. Sanction issued November 18. The loan record at every stage — application, credit, sanction, disbursement — carries the campaign ID. No manual attribution is required; the linkage is structural from the first response.

→ LOS origination source: CSELL-NOV-T06-0884 · Sanction: Nov 18 · Campaign ID in sanction record
Disbursement confirmed · Revenue attributed · NIM ledger updated

Disbursement record closes the attribution loop — NIM contribution computed and credited to campaign T06

Loan disbursed November 22: ₹17.4 lakh. The disbursement record reads back to campaign CSELL-NOV-T06-0884. The Marketing Analytics AI receives the disbursement notification and computes the NIM attribution: ₹17.4L × 2.60% NIM × 10-year expected tenor = ₹4.52L lifetime NIM contribution, of which approximately ₹45,240 is the year-1 NIM contribution. Campaign cost allocated to T06 for this account: ₹98 (send cost + proportional RM time). ROMI for this specific campaign instance: 4,614×. The aggregate T06 ROMI of 38× reflects the campaigns that did not convert diluting the ROI of those that did.

→ Disbursement: ₹17.4L · NIM: ₹4.52L lifetime · Campaign cost: ₹98 · Attribution: complete · T06 ROMI updated
21.9×November cross-sell ROMI — ₹68.4L NIM contribution vs ₹3.12L campaign cost · Every trigger profitable on its own
38×T06 (new GST state) ROMI — highest trigger by ROI · 84 campaigns · 24 funded loans · ₹6L NIM · ₹8,200 cost
16.8%Overall conversion rate — 384 funded loans from 2,284 triggered campaigns · vs 2–4% for calendar-based equivalent
48hAttribution lag — disbursement confirmed → NIM ledger updated within 48 hours · No manual calculation

₹3.12 lakh in campaign costs produced ₹68.4 lakh in NIM contribution — a ratio that makes acquisition marketing look like an expensive habit

The institution's paid acquisition marketing — Google, Meta, DSA commissions — produces disbursements at a CAC of ₹15,000 to ₹35,000 per funded loan and a ROMI of 2× to 8×. The cross-sell campaign produces disbursements at an average ROMI of 21.9× — significantly better, from an audience that already trusts the institution, using data the institution already holds, with credit assessments that are already done. The reason cross-sell underperforms acquisition in most institutions is not that it is less productive — it is that it is less resourced. Without a system that monitors 22 triggers, runs eligibility checks, personalises outreach, tracks revenue, and operates without a campaign manager, cross-sell remains a periodic newsletter rather than a systematic revenue programme. The Cross-Sell Campaign Agent AI is the system that converts cross-sell from an aspiration into a measurable, attributed, board-reportable revenue line.

← Back to Cross-Sell Campaign Agent AI