Home · Solutions · Finance & accounting

Solution · Finance & accounting

Seven partners, one register: every commission line traced to a contract

F&I commissions matched to the deal that earned them

Robots read every partner statement, match each line to the deal file by customer, VIN and contract number, compute the adviser's share and queue what was never paid.

DepartmentalMicrosoft TeamsHuman in the loopAI where it earns its place
720financing contracts and insurance policies a month are arranged by this illustrative dealer group, and seven partners settle them on seven schedules.

Executive summary

Challenge

Your F&I income is checked when an adviser asks about a bonus, not when a partner underpays.

What changes

What we build joins two things the group already has and has never joined: the deal file in each brand's dealer management system.

Business value

Every commission line is tied to a customer, a VIN and a contract, so F&I income can be explained to an adviser.

Systems involved

the contract register and evidence library on SharePoint; the group accounting system; the partner query files

Business problem

F&I settlement

A dealer group earns a commission every time it arranges finance or insurance alongside a car. The margin on the vehicle is thin and largely set by someone else; the F&I commission is neither, because it is earned in the showroom and calculated somewhere else. Each bank, lessor and insurer applies its own rate table, its own settlement cycle and its own document: a PDF statement, a portal export, a payment carrying a reference nobody in finance recognises. Seven partners produce seven vocabularies for one event, a customer signing a contract in front of an adviser.

The reconciliation lands on one person with a workbook, and it is done to the total rather than to the line. What arrives is booked; what does not arrive is invisible, because no register pairs a signed contract with the commission it should have produced. Clawbacks appear as negative lines inside a later statement and are accepted without checking whether the customer really cancelled, or whether the adviser's share has already been paid out. Advisers, site directors and the CFO each hold a different F&I number, and an eighth partner adds a column and a day to the close.

How it works today

What we find in most multi-brand groups, whichever system each brand runs.

  1. PersonThe F&I adviser closes the contract in the partner's portal and writes the contract number onto the deal file by hand, when there is time
  2. WaitingThe commission settles on the partner's own cycle, a fortnight to a quarter later, and no calendar tracks those seven cycles
  3. PersonStatements arrive as PDFs or portal exports and are copied into one workbook, partner by partner, in the line names each partner uses
  4. SystemThe monthly total per partner is posted as F&I income; individual lines are never matched back to a contract
  5. Risk of errorA contract that produced no commission produces no document either, so it surfaces only when an adviser queries a bonus
  6. PersonClawbacks and short payments are accepted as they arrive, because reconstructing the original line costs more than the difference is worth
PersonWaitingSystemRisk of error

Why the current process costs more than it appears

The bill that never reaches the budget.

  • Unpaid commission behaves like a discount nobody granted: a contract the partner never settled leaves no trace in the group's own accounts, so the loss is silent, and permanent once the query window closes.
  • Booking to the total hides errors in both directions. A statement €400 light and one €400 heavy net out in the ledger, so a partner rating one product a point low all year is never caught by anyone.
  • Adviser trust is spent on arithmetic. A share that cannot be traced to a contract becomes a conversation in the sales director's office, and the group pays twice: the disputed amount, then the hours spent proving it.

Cost of inaction

Twelve months of statements matched to a total≈ €41,184
The same workbook carried through three settlement years≈ €123,600
At 900 contracts a month once the two planned sites open≈ €51,500

Nothing here fails loudly enough to reach an agenda. The statements arrive, the totals are booked, the month closes, and the difference between what the group earned and what its partners paid is a number it has never held. That is what keeps the workbook alive, and why the middle row is the realistic one rather than the dramatic one.

The commission itself is deliberately absent from the table. What one under-rated product costs over a year is a figure the group can produce from one partner agreement and one statement, and we would rather it came from there than from us.

Illustrative scenario

A plausible organisation with realistic proportions. The figures are there to be recalculated on your data; they are not a client result.

Organisation

A dealer group in Poland: four brands across six sites in three cities, about 520 employees, each brand on its own dealer management system, a five-person finance team and Microsoft 365 E3.

Volume

720 financing and insurance contracts a month with seven partners, from roughly 430 new and used vehicles delivered: bank and captive-lessor agreements, motor insurance, GAP and extended-warranty products, several on the same car.

Current process

Statements arrive as PDFs and portal exports on seven cycles, are copied into one workbook by a finance specialist and posted as a monthly total per partner; the adviser share is computed on a second sheet.

Bottleneck

About eleven minutes per contract across the month, spent copying, hunting a customer in a statement and answering advisers, with no list anywhere of contracts that should have settled and have not.

Solution

Robots match every statement line to the deal file by customer, VIN, contract number and product; matched income posts with the adviser share computed from rules finance owns, while unmatched lines, missing commissions and clawbacks become tasks in Microsoft Teams with the deal file attached.

Potential outcome

F&I income is recognised as it is earned rather than as it is remembered, queries reach partners inside their windows, and the CFO's number and the adviser's number come from one record. The figures are ours, built on the assumptions below rather than on a client's measurement.

Proposed solution

What we build joins two things the group already has and has never joined: the deal file in each brand's dealer management system, and the statement each partner sends. Every night, robots read new and changed deal files into one contract register on SharePoint, keyed by VIN, customer, adviser, site, product and contract number. On the partner side, where an export is published the robot takes the file; where only a PDF reaches the finance mailbox, UiPath Document Understanding reads that layout into the same fields, and an uncertain page becomes a validation task rather than a posting.

Matching is deterministic and finance owns the rules. A line matches on contract number, on VIN and customer, or on whatever combination the rule table allows, and each rule carries the tolerance it may apply. Matched lines post to the group accounting system with the adviser share computed from a rate table held as versioned rows in Microsoft Lists, so a rate agreed in March applies from March and stays attached to the postings made under it. Everything else becomes a task in Microsoft Teams with the statement page and the deal file side by side: a line matching no contract, a clawback, an amount outside tolerance, and the most valuable case of all, a contract older than its partner's cycle with no commission against it.

Native capabilities used

UiPath Document Understanding with Document Validation actions in UiPath Action Center; UiPath Orchestrator queues, time triggers, credential store and run log; UiPath Integration Service connectors for Microsoft Outlook 365 and Microsoft OneDrive & SharePoint; Action Center actionable notifications in Microsoft Teams; Microsoft Lists version history; Power BI as a tab in Microsoft Teams

What we build

The contract register and its nightly reads, statement intake per partner, matching rules and tolerances, the adviser rate table, posting and clawback reversal, the exception queue in Teams, the partner query drafts and the Power BI F&I view

Custom integration

Each brand's dealer management system through a scheduled export, a read-only database view or UI automation; each partner portal through its own export where one exists and UI automation where it does not; the group accounting system for postings and reversals

How the automated process works

  1. AutomationNightly, robots read new and changed deal files from each brand's system into one contract register: VIN, customer, adviser, site, product and contract number
  2. AutomationStatements and payment advices are collected from the finance mailbox and the partner portals; exports are taken as files, and Document Understanding reads the layouts of partners that send only PDFs
  3. SystemEvery line is matched to a contract on the rules finance owns, within the tolerance each rule allows, and matched lines post with the adviser share from the versioned rate table
  4. PersonA page the model is unsure about becomes a validation task, where a person sees the extracted fields next to the page itself
  5. PersonUnmatched lines, clawbacks, amounts outside tolerance and contracts that should have settled and have not reach finance as Action Center tasks in Microsoft Teams, each with its deal file
  6. AutomationAccepted queries leave as a claim drafted from that same evidence, and the Power BI view refreshes with income earned, matched, paid, open and clawed back
AutomationSystemPerson

Human-in-the-loop model

Automation handles

  • Reading deal files and partner statements into one register with one set of fields
  • Matching on the rules finance wrote, posting matched income and computing each adviser's share
  • Ageing every contract that should have settled and has not, and drafting the query to the partner

People decide

  • The matching rules, tolerances and rate table, which stay rows in a list with owners and validity dates
  • Whether a clawback is accepted or challenged, and whether an adviser's share is reversed with it
  • Every unmatched line and every difference: accepted with a reason code, or queried before the window closes

Before and after

BeforeAfter
Commission checked against the dealon the monthly total, per partnerline by line, on contract number or VIN
A commission never paidfound when an adviser asks about a bonusdated in a queue once the partner's cycle has passed
Clawback handlingaccepted as a negative line in a later statementmatched to the original line and the adviser's share

Systems and integrations

Everything below runs on licences and systems you already hold, or would need anyway.

Inputs

  • the deal file in each brand's dealer management system
  • partner statements and payment advices in the finance mailbox
  • partner portal exports
  • bank and ledger postings

Automation layer

  • UiPath Orchestrator
  • UiPath Robots
  • UiPath Document Understanding
  • UiPath Integration Service
  • UiPath Action Center

Target systems

  • the contract register and evidence library on SharePoint
  • the group accounting system
  • the partner query files
  • the Power BI semantic model

Human touchpoints: validation, unmatched-line and clawback tasks in Action Center in Microsoft Teams; the F&I channel in Teams; the Power BI view as a Teams tab

the deal file in each brand's dealer management systemUiPath OrchestratorUiPath Robotsthe contract registervalidation

Technologies used

UiPath Document Understanding (IXP)

reads the statement layout of each partner that sends PDFs; Document Validation actions for uncertain pages

A
UiPath Robots + Orchestrator

nightly reads, matching runs, postings and reversals; time triggers, queues, credential store, run log

A
UiPath Integration Service (Microsoft Outlook 365, Microsoft OneDrive & SharePoint connectors)

collects statements from the finance mailbox; keeps the register and the evidence library

A
UiPath Action Center in Microsoft Teams

validation, unmatched-line and clawback tasks with due dates, completed in Teams

A
Microsoft Lists

matching rules, tolerances and the adviser rate table as versioned rows owned by finance

A
Power BI

F&I income by partner, product, site and adviser: earned, matched, paid, open, clawed back

A
Averified product capability (vendor documentation)

Illustrative economic model

A model, not a promise.

Illustrative model
720 financing and insurance contracts a month × 11 minutes of matching and follow-up= 132 h / month
132 h × €26 fully loaded hourly cost= €3,432 / month
× 12 months≈ €41,184 / year
Annual finance capacity released (illustrative)≈ €41,184

Divide a month of reconciliation work by the contracts in it and the eleven minutes below is what remains: copying statements into the workbook, hunting one customer among a hundred lines, recomputing an adviser share, answering the question that follows. €26 is a fully loaded hourly cost for a finance post in a Polish dealer group. Neither the commission itself nor anything recovered by chasing it appears here, and nothing was measured at a client.

Run the numbers on your data

hours released per month
of annual capacity released

An illustrative estimate from your own inputs. It models released capacity; it is not a promise of savings.

Business benefits

  • Every commission line is tied to a customer, a VIN and a contract, so F&I income can be explained to an adviser, an auditor or a partner from one record
  • A contract that should have paid and has not becomes a dated item in a queue, chased with its evidence while the query window is open
  • Clawbacks are matched to the original line and to the adviser's share before they are accepted, instead of being netted into a monthly total
  • The adviser share is computed once, from a rate table finance owns and versions, and lands on the figure the adviser can trace for themselves

The management view

  • F&I stops being a monthly total and becomes a managed revenue line: earned, matched, paid, open and clawed back, by partner, product, site and adviser
  • Partner behaviour becomes visible in the group's own numbers, which is what a rate renegotiation needs and what a partner review has never had
  • The reconciliation stops depending on one workbook and one person's memory of seven partners' habits, so an eighth partner is rows, not another close day

Board-level KPIs

F&I income per vehicle deliveredshare of commission lines matched first timecommission open beyond the partner's settlement cycleclawback rate by productdays from statement to matched posting

Security and governance

Trust in automation is built on the audit trail, not on a promise.

  • Robot users are named accounts on each brand's system, each partner portal and the accounting system, reading where reading is enough and posting only where the robot posts; secrets live in the Orchestrator credential store or Azure Key Vault, never inside a workflow
  • A commission line carries a customer name, a VIN and a contract number, so the register and the evidence library are open to finance and the site director only, retention is set through Microsoft Purview, and the Orchestrator log records every read and posting
  • Extraction is the only place a model touches this process and it proposes fields rather than deciding them; matching is arithmetic on rows a person can read, whoever edits a rule cannot approve a posting computed from it, and robots run in the EU region of UiPath Automation Cloud, writing only into the group's own Microsoft 365 tenant

Why now

01

CECRA, the European dealers' body, describes recent returns for dealerships as "negligible" and fair remuneration as "a central success factor" while manufacturers move parts of their networks to agency distribution; where the margin on the car is set elsewhere, the commission a group earns for itself is the line that repays counting properly

02

Poland registered 597.4 thousand new passenger cars in 2025, 8.3% more than in 2024: 411,000 to company and institutional buyers, 186,000 to private ones (PZPM and KPMG, published 3 February 2026; the ratio of roughly seven in ten is our computation, not their figure). A growing market adds contracts to the same workbook and the same person

03

Nothing in the build is exotic: a document engine that learns a partner's layout from corrections your team already makes, connectors for the mailbox and SharePoint, tasks completed in Teams, a Power BI tab. Against those, the modelled €3,432 a month of desk time is the smallest argument on the page

Relevant executive roles

Group CFO

F&I income becomes a receivable built from contracts, matched to the euro and defensible in front of a partner or an auditor

Group Managing Director

An eighth partner or a fourth brand means new rows in a rule table rather than another reconciliation day in the close

Sales Director

An adviser's commission share can be traced to the contract that earned it, which settles the argument before it reaches your office

Common questions and objections

Every partner sends something different, and two send only PDFs.

That is the normal starting point. Where an export exists the robot takes the file; where it does not, a model is trained on that layout and uncertain pages become validation tasks. A layout change then shows as a rise in validation tasks, not as a wrong posting.

The contract number does not always reach the deal file.

Then matching starts from VIN and customer, and the register reports how often each key was missing, by site and by adviser. Most groups find the field gets filled within a quarter, because the queue that used to be invisible now carries somebody's name.

Won't this simply move the disputes onto us?

The disputes exist already, settled today by whoever remembers the case. What changes is that a query reaches the partner with the contract, the statement page and the arithmetic attached, inside the window.

When this is not the right solution

  • A single site with one or two finance partners and a few dozen contracts a month, where a disciplined monthly check costs less than a build
  • Commission terms agreed case by case and never written down as a rate a rule can apply; putting them on paper is the first project, not this one

A question for the next management meeting

If one partner had been underpaying us on one product for a year, which report in this group would show it, and how many statements would have passed before it did?

Implementation approach

The first week looks the same at every client: we look at the data.

We deliver

  • One closed quarter matched by hand once, so the rules come from evidence rather than from what people believe the partners do
  • The contract register and the nightly reads from each brand's dealer management system
  • Statement intake per partner: the export where one exists, a trained document model where it does not
  • Matching rules, tolerances, clawback treatment and the adviser rate table as rows finance owns, with validity dates
  • Posting and reversal, the exception queue and query drafts in Microsoft Teams, and the Power BI F&I view

We need from you

  • Three months of statements from every partner, with the payments and postings that went with them
  • The commission agreements or rate schedules in force, and the adviser rate table as applied today
  • Read access to each brand's system, a robot user per partner portal, and an owner in finance for the rules

Stages

Discovery

One closed quarter matched by hand, then match rules, tolerances, clawback treatment and the rate table agreed as rows with named owners

Build

Register, nightly reads, statement intake, matching, posting, exception queue, query drafts, Power BI

Back-test

The same quarter re-matched by the robots and compared line by line with what was booked

Go-live

One partner and one brand first, every posting reviewed, then the rest with hypercare

Departmental. Effort follows the number of partners and statement formats, how reliably the contract number reaches the deal file today, and what each brand's system will give up.