Reconciliation automation for complex e-commerce flows

Your systems are connected. Month-end still needs investigation.

MatchPass helps e-commerce finance teams automate recurring payout and clearing-account exceptions across Shopify, payment gateways, wholesale, bank and ledger, after standard connectors have done their job.

Routine matches clear automatically. Your team reviews a short, explained exception queue instead of rebuilding the same reconciliation every month.

Review one recurring workflow Watch the engine run

Free 20 to 30 minute fit call. No system access required.

Month-end reconciliation Reviewing 7 lines
Bank ledger
DEP-0918 · Mar 144,166.54

One bundled deposit. Hundreds of orders behind it, net of fees and refunds. This is why nothing ever ties out by hand.

Shopify orders
#10421,284.50✓
#104396.20✓
#10442,410.00✓
#1045318.75✓
#1046149.00✓
Refund #1038−57.09✓
Wholesale inv. W-2210−34.82✓!
0 of 7 tied out Reviewing

Self-qualify first

Is this a MatchPass problem?

Two columns, honestly drawn. Most visitors belong on the right, and that is a useful answer.

Good fit

  • A2X, Bookkeep, Synder or a native connector is already configured, but clearing accounts still require monthly investigation.
  • Revenue moves through multiple gateways, channels, entities or currencies.
  • Shopify, PayPal, Stripe, Amazon, wholesale, bank or ledger data must be tied together.
  • The same exception patterns return every close.
  • A meaningful amount of finance time goes into investigating transactions that follow repeatable rules.
  • A finance owner is available to approve accounting treatment and exception rules.

Probably not a fit

  • Straightforward Shopify Payments reconciliation.
  • A connector already handles the workflow cleanly.
  • A few isolated exceptions per month.
  • One-off cleanup with no recurring process to automate.
  • General bookkeeping, tax or catch-up accounting.
  • A process where every decision genuinely requires accounting judgment.

If standard software already produces a clean close, keep using it. MatchPass is for recurring work left behind.

From our reference build: a synthetic month of multi-gateway payout data with ten real-world problems deliberately injected · every figure below is our own and reproducible on demand

96.40%
auto-match rate across seven cascading passes on the reference build
4,812
payout, order, refund, and bank lines ingested across three gateways
10 of 10
deliberately injected real-world problems caught, classified, and explained
173
exceptions routed with a reason code, an owner, and a suggested fix

See it work

From four messy files to a clean exception queue

This is our reference build running a synthetic month for a fictional brand, Northshore Supply Co., across Shopify Payments, PayPal, and Stripe. Start with the raw feeds exactly as they arrive, run the engine, then open the dashboard a client sees every morning.

Dataset: northshore_supply_co · March · synthetic demo data · 4,812 lines
shopify_payouts_mar.csvShopify Payments
Payout DatePayout IDOrderGrossFeeNet
2026-03-14po_9H2KD4#10421,284.50-37.551,246.95
2026-03-14po_9H2KD4#104396.20-3.0993.11
2026-03-14po_9H2KD4#10442,410.00-70.192,339.81
2026-03-14po_9H2KD4refund #1038-57.091.66-55.43
paypal_activity_download.csvPayPal
DateTypeTransaction IDGrossFeeNet
03/14/2026Express Checkout7AK33901BC318.75-9.54309.21
03/14/2026Express Checkout2XN80417QQ149.00-4.62144.38
03/15/2026Chargeback5RW10982LM-212.40-20.00-232.40
03/16/2026General Withdrawal1PL55620HH-453.590.00-453.59
stripe_balance_txns.csvStripe
createdidamountfeenetpayout
2026-03-14T09:12:44Zch_3Nx8Ta2eZ57090198655104po_1OqPzT
2026-03-14T11:03:07Zch_3Nx9Cb7rK1042503323100927po_1OqPzT
2026-03-14T16:40:19Zch_3NxAqD5mW88800288585915po_1OqPzT
2026-03-15T02:11:52Zre_3Nx8Ta2eZ-570900-57090po_1OqR8b
bank_statement_mar.csvBank feed
DateDescriptionAmount
14-MAR-26ACH CREDIT SHOPIFY no reference4,166.54
14-MAR-26STRIPE TRANSFER ST-K8Q22,419.46
16-MAR-26PAYPAL INST XFER453.59
17-MAR-26ACH CREDIT SHOPIFY3,881.02

Four sources. Three date formats. Stripe reports amounts in cents. The bank posts one bundled number with no order reference. This is the version your bookkeeper sees, and matching it by hand is where the hours go. Press run and watch the engine do it instead.

Lines ingested
4,812
Auto-matched
0
Exceptions
—
Run time
—
Idle · press run to start 0.00%
01Exact match on order reference and amount——
02Amount match inside a two-day settlement window——
03Rollup: many orders to one bundled payout——
04Rollup net of gateway fees——
05Rollup net of refunds and chargebacks——
06Payouts split across two bank deposits——
07Residual tolerance band on rounding variance——
Northshore Supply Co · Exception queue March close · refreshed after every scheduled run
Open
141
In review
24
Resolved
8
Value flagged
$6,241.87
ReasonExceptionAmountStatusOwner
What happened
The March 1–15 and March 10–20 exports overlap, so the same Shopify payout landed in the ledger twice. Revenue is overstated by $4,166.54 until one copy goes.
How the engine caught it
Two ledger entries share one payout ID and one bank deposit. Only one bank line exists, so one entry can never match.
Suggested action
Delete the duplicate import batch and lock the export date ranges so they cannot overlap again. Two minutes, not two hours.
What happened
The plan rate is 2.9% + 30¢. This charge was assessed at 3.4%, which is $5.19 over on one order, and the same pattern shows on 27 more lines this month.
How the engine caught it
Every fee is recomputed against the plan schedule on ingestion. Anything outside tolerance gets its own reason code instead of disappearing into "bank fees."
Suggested action
Check whether these orders used a card type on a different rate, and raise the batch with the gateway if not. The engine has the line-level evidence attached.
What happened
PayPal pulled $212.40 plus a $20 dispute fee out of the March 16 withdrawal. The ledger still shows the original sale and nothing else, so cash and books disagree.
How the engine caught it
The withdrawal only rolls up if every activity line inside it exists in the ledger. The chargeback line had no ledger side, so the whole payout was held with a named cause.
Suggested action
Post the chargeback and fee to the dispute account. The entry is pre-drafted with the PayPal case ID attached, waiting for approval.
What happened
A bank credit arrived with a blank description. By hand this means scrolling three systems hoping a number jumps out.
How the engine caught it
Pass 2 found exactly one order for $96.20 inside the settlement window. Because the match key was amount and date rather than a reference, it routes for a one-click confirm instead of auto-clearing.
Suggested action
Confirm the suggested pairing. Strict rules auto-clear, looser rules ask first. That discipline is why the auto-matched pile stays trustworthy.
What happened
Orders from March 30–31 settled April 2. Close March without handling it and revenue and cash land in different periods, the classic cause of a clearing balance nobody can explain.
How the engine caught it
Every rollup is checked against the period boundary. Cross-period payouts get flagged instead of silently matching into the wrong month.
Suggested action
Accrue the in-transit amount at close, and the engine clears it automatically when the April deposit lands.
What happened
Eight $1.00 and $0.50 charges from checkout testing were still flowing into the feed and polluting every manual reconciliation.
How the engine caught it
Known test card fingerprints and amount patterns are classified on ingestion and excluded from matching entirely.
Suggested action
None. Rule added, resolved automatically, logged for the audit trail. This is what "resolved by engine" looks like every morning.
167 more lines in the live queue, grouped under 10 reason codes Every action logged · full audit trail per line

Click any row to see what your team would actually do with it. Each exception arrives explained, owned, and pre-drafted, not as a mystery in row 3,204 of a spreadsheet.

This run uses synthetic data we generated ourselves, with ten real-world problems deliberately injected into it. It is a demonstration of our method, not a client result, and your numbers will differ: the audit in step one is where we find out by how much. The exceptions are the point. Any tool can match the easy 60%. The value is a short, explained list of what genuinely needs a human, with the evidence attached and the fix pre-drafted, instead of a spreadsheet with 4,812 rows and no idea where to start. Your dashboard runs on your infrastructure, refreshes on a schedule without anyone remembering to run it, and keeps a log your auditor can walk through line by line.

Recurring problem patterns

Where standard connectors commonly leave work behind

Third-party gateway clearing

The connector posts the sales correctly. PayPal, Stripe or other gateway fees, timing and settlements still leave recurring clearing differences.

DTC plus wholesale

Orders, invoices, payments and deposits live in different systems and must be reconciled together.

Multi-currency settlements

Transaction currency, settlement currency, base currency and bank deposits do not align cleanly.

Refunds and chargebacks

Cross-period refunds, disputes and payout deductions create exceptions that require repeated investigation.

Key-person spreadsheet workflow

One person knows how the workbook works. MatchPass turns repeatable matching logic into scheduled runs, documented rules and an owned exception queue.

The story behind the name

Why MatchPass

MatchPass takes its name from how we match: multiple passes, strictest rules first. Clear the exact one-to-one matches before any broader rule runs, and real breaks surface instead of drowning in noise.

The method comes from an industry-sponsored engineering capstone at Toronto Metropolitan University, where our founder, Karan Agrawal, co-led the redesign of a daily, enterprise-scale securities reconciliation. With his project partner he designed and configured a seven-pass cascading engine with rollup rule sets, derived-field classification, and exception workflows, built on a DMAIC frame and validated in testing against the operations team's trusted manual view.

Every number we publish is our own. The reference build above runs that same architecture on a synthetic month of multi-gateway payout data, and it mirrors what we install for clients: matching engine, exception dashboard, scheduled runs, and parallel testing against your current process before cutover.

Karan Agrawal, founder of MatchPass Labs
Karan Agrawal
Founder · Industrial engineer · Six Sigma Black Belt

The multipass method, in one glance

  • 01Exact one-to-one matches clear first. No judgment calls needed.
  • 02Each following pass loosens one rule at a time, so precision degrades in a controlled order.
  • 03Rollup rules handle the hard part: one bundled payout matched against hundreds of orders.
  • 04Whatever is left is a genuine exception, routed to a dashboard with a reason code and an owner.
  • 05Every run is logged. Your auditor gets a trail, not a shrug.

Where MatchPass sits

Your connector may not be the problem.

A2X, Bookkeep, Synder and native integrations are often the right tools for clean, repeatable e-commerce posting. MatchPass is not a replacement for software that already works.

We step in when several systems must be reconciled together, clearing balances repeatedly survive the connector, or the remaining exceptions follow rules that standard configuration cannot express.

Keep the existing stack. Add a custom reconciliation layer only where the recurring manual work justifies it.

The gap between QuickBooks and BlackLine

E-commerce finance teams often land between two categories of software. Connectors and built-in matching are designed for clean, single-channel posting. Enterprise reconciliation platforms start around $12,000 a year and assume a controls team to run them. MatchPass is built for the teams in between, complex enough that ordinary connectors leave residue, not large enough to justify an enterprise platform.

A lower-friction starting point

Not ready for a call? See where we would look first.

We will build a quick outside-in Payout Reconciliation Snapshot using only public information about your sales setup. It shows the parts of your payment flow we would investigate first, where manual matching is most likely to creep in, and what a cleaner month-end view could look like.

No bank access No accounting login No made-up ROI numbers

Just a useful first look at whether there is enough complexity here to investigate properly.

What we can see

DTC, wholesale, marketplaces, subscriptions, retail locations, currencies, and other publicly visible payment-flow signals.

Where we would test first

Two or three specific reconciliation points that may be creating repetitive close work, each written as something to test rather than something we claim is true.

What good could look like

One month-end view showing what cleared automatically, the exceptions still needing review, and the exact dollar value flagged.

Get my free snapshot See an example snapshot

Directional and based on public information only. The paid Exception Assessment is where we validate the process against your actual data.

Start small. Prove the economics.

Understand the exception before paying for a full build.

Two small steps come first. The implementation tiers below them are alternatives chosen after the assessment, not stages you pay for in sequence.

Step 1

Case Fit Call

Free

Twenty to thirty minutes. Bring one clearing account, payout flow or recurring spreadsheet process. We will determine whether the problem appears repeatable, automatable and economically meaningful.

Book a Case Fit Call

Step 2

Exception Assessment

$950 fixed

One entity, one period, up to approximately two primary data sources.

  • Quantify recurring exceptions and current manual effort
  • Identify the matching rules and the decisions that require judgment
  • Estimate the likely automation scope and the business case
  • Recommend whether to stop, implement one workflow, or scope a more complex system
  • Deliver a fixed implementation quote
The full $950 is credited toward an implementation begun within 60 days
Start with an assessment

Step 3

Choose the appropriate implementation after the assessment

One of the following, selected against what the assessment actually found. These are alternatives, not cumulative.

Single-Workflow Implementation

from $3,500

One tightly scoped production workflow with defined inputs, matching rules and exception output.

  • One entity
  • One recurring workflow
  • Approximately two principal data sources
  • Matching and exception logic
  • Scheduled run or defined operating process
  • Documentation, handoff and 30 days of post-launch support

Complex or Multi-Source Implementation

$5,000 to $8,000

For workflows involving several sources, gateways, rule families, currencies or materially more complex matching logic.

Multi-Flow or Multi-Entity System

$8,000 to $15,000+

For several connected workflows, entities or operational teams requiring shared controls and reporting.

Optional Care Plan

For monitoring and limited maintenance of one implemented workflow as file formats, gateway fees and existing rules change.

from $350/month

Complex, multi-flow and higher-service coverage is quoted separately. New workflows, entities, integrations or substantial logic changes are not included in the base care plan.

Additional entities, data sources, workflows, historical cleanup or materially new logic are separately scoped. All prices in USD, plus applicable taxes. Quotes are locked for 30 days. Implementations begin on a 50% deposit. How we work with your data.

No pitch, just diagnosis

Have a recurring reconciliation problem already?

Bring the clearing account that will not tie out, the payout that never matches, or the process your team keeps rebuilding every month. We will spend 20 to 30 minutes determining whether it is worth assessing.

10 min the problem you brought 10 min whether the pattern repeats 5 min honest fit check
Book a Case Fit Call

Calendar not loading? Book directly on cal.com or email karan@matchpasslabs.com