3 weeks→20 min
Month-end settlement reconciliation cycle — marketplace settlement reconciliation.
00 / FINANCE
You move money and you answer to regulators. Every rupee has to reconcile.
Settlement, escrow, failed-transaction reversals, refunds and collections. Different systems, different identifiers, none of them agreeing. We automate the matching, and we build the evidence trail that survives an audit.
FIELD EXTRACTIONFINANCE
01 / THE PROBLEM
Where it breaks in your world.
04 PROBLEMS, SCROLL TO READ
You are not short of reports. You are short of a system that reads all of them and tells you what does not agree.
02 / BUILT FOR
Three audiences. One reconciliation discipline.
Facing DPDP obligations? See getdpdpcompliant.com ↗Payment and collections businesses
Gateway settlement against ledger, bank reconciliation, refunds, reversals and failed-transaction tracking.
Lenders and lending platforms
Collection returns and re-presentation, repayment allocation, loan ledger reconciliation, partner and co-lending splits.
Payment aggregators
Escrow position against settlement obligations, merchant payouts, refunds and audit evidence.
03 / WHAT WE AUTOMATE
Matched one to one against every problem above.
01
Gateway payout to transaction ledger
The system reads the payout file, decomposes it into its underlying transactions, and matches each one back to your ledger with fees, taxes and adjustments accounted for separately rather than absorbed into the variance. What does not match becomes an exception with the supporting records attached.
02
Failed transactions against their deadlines
Every failed transaction gets a clock attached at the point of failure, set by its payment system and its scenario. The system surfaces what is approaching its deadline while there is still time to act, instead of reporting the breach afterwards.
03
Refund lifecycle across systems
Initiation, processing and credit tied into a single tracked object. Refunds that stalled between two states and refunds that went out twice both surface automatically rather than arriving as a customer complaint.
04
Collection returns to loan ledger
Presentation, return and re-presentation mapped back to the account and the bucket as the files land, so the collections position and the books show the same number on the same day.
04 / HOW IT WORKS
From raw files to a close-ready position.
WHAT GOES IN
Gateway and settlement reports, bank statements, your transaction ledger, return files, tax reports. Read on a schedule, with format changes absorbed in the ingestion layer so a new column from a provider does not take your close down.
WHAT WE MATCH ON
Rarely one clean identifier. The matching is built around what is genuinely present in your data, usually a combination of transaction and settlement references, bank references, amounts, dates, statuses and business rules, including one-to-many and many-to-one relationships where a single payout covers many transactions.
WHAT COMES OUT
Matched records, unmatched records on both sides, duplicates, timing differences and amount variances, separated by cause rather than piled into one break list.
WHAT FINANCE GETS
An exception queue with supporting records already attached to each break, a reconciled position, and a close-ready report. Your team resolves breaks instead of investigating them.
GATEWAY
LEDGER
BANK
↓
MATCHING ENGINE
↓
MATCHED
FEE VARIANCE
TIMING DIFFERENCE
Illustrative. Gateway, ledger and bank feeds converge on the matching engine; output splits into matched, fee variance and timing difference.
05 / PROOF
What we have removed here.
67%less
Reduction in finance team effort on that reconciliation, from the same engagement.
3.5 hr→2 min
Gift-card validation run — fintech client.
Client names withheld under confidentiality. Figures are from delivered work and we will walk you through the method on a call.
06 / BUILT FOR YOUR STACK
One team, on both sides of the reconciliation.
We build against the systems you already run, on both sides of the reconciliation. Payment providers such as Razorpay, Cashfree and PayU on one side, your ledger, core system and accounting on the other.
APIs and database access come first. Screen automation is a fallback for when no reliable interface exists, not a default, because a bot that scrapes a portal breaks the week that portal changes and takes your close down with it.
Reconciliation, validation and reporting come from one team, so there is nothing to stitch together afterwards.
07 / HOW WE WORK
The process, in four steps.
01
We read your data before we scope
Send one month of the actual files. A settlement report, a bank statement, an export from your own system. Most of what matters is visible in the data and never reaches a requirements document.
02
We tell you what not to build
Some of what your team reconciles by hand should stay manual because the volume does not justify a system. We will say so.
03
One flow to production before we start the second
A first flow live and trusted beats six flows half-finished, and it tells you early whether this was worth doing.
04
It runs in your infrastructure
The system runs on your side, with code and documentation handed over as set out in the engagement. We price the implementation as a project rather than charging a fee on every transaction processed.
08 / FAQ
Questions we get asked first.
No. We build alongside what you run and read from it. Nothing gets migrated.
Yes. A single source can often be handled by hand. It becomes a systems problem when gateway, bank, ledger and operational data all have to agree, whether that is one provider or four.
We build the reconciliation and the evidence trail underneath it. We are engineers, not your auditor or your compliance counsel, so the certification stays with them. What we provide is the daily position and the supporting records they sign against.
The ingestion layer is built expecting formats to change. We handle changes under a support arrangement, or your team does if you would rather hold it in house.
Built. We are not selling a rules engine you have to fit your business into. The matching logic is written against your data and your flows, which is why we ask for a month of files before quoting.
For a well-defined single flow, implementation is typically measured in weeks rather than quarters. We confirm the range after seeing your data.
No. It is a project, priced on the system and the scope rather than on transaction volume.
09 / START HERE
Show us the check that takes all afternoon.
A working session, not a sales call. We will find the manual work that is slowing your operations and size what it costs.
Smoketrees Digital LLP
Custom software systems for growing businesses.
HSR LAYOUT, BENGALURU
© 2026 SMOKETREES DIGITAL LLP. ALL RIGHTS RESERVED.