The DARP Framework

The framework that helps solveall your finance issues

  1. Discover.
  2. Assess.
  3. Recover.
  4. Prevent.

Discover, Assess & Recover identifies the leak and gets your money back. Prevent is your strategic solution to plug the leak for good, employing the same rules deployed during recovery.

See the four stages
  • Read-only to start
  • Full population, not a sample
  • A number in about two minutes
  • Funded by what it recovers
  • No second build during Prevent

How it works

Suspicion to proof to cash to control

Four stages. You can stop after any of them, and most of the value arrives before you’ve committed to anything.

  1. Stage 01

    Before any contract

    Discover

    We ingest the transaction history your systems already hold and test the full population for every anomaly, duplicate, mismatch and unclaimed entitlement.

    1. 1No contract

      Scope and access

      Agree the process area and period. You send an export, or grant read access. NDA and security review happen here.

    2. 2~2 minutes

      Discover runs

      The full population is tested against every rule for that area, in about two minutes of run time, not weeks of fieldwork.

    3. 3Same day

      You have the number

      The total recoverable figure, split across cash, tax, misstatement and control exposure. Yours to keep, whatever you decide next.

    What we need

    Read access to the relevant history, an export is enough. For payables that means AP transactions and the vendor master. We are reading, not writing.

    What you get

    The total recoverable number in about two minutes of run time, yours before there is anything to sign. The waiting is access and NDA, not analysis.

  2. Decision point

    Contract

    You decide here. You have a quantified number and you have paid nothing. Everything above happened on your data, with no commitment. Everything below starts once a contract is in place.

  3. Stage 02

    After contract

    Assess

    Now the number opens up. Every finding is drilled to the last level: the transaction, the clause breached, the evidence behind it. Assess ends with your sign-off on what gets pursued.

    What we need

    Contracts, price agreements and program terms where they govern the amount, plus your team’s judgement on what is collectible and what is worth the relationship.

    What you get

    Drill-down to the last level on every finding, a prioritised list agreed jointly, and a signed-off recovery scope before a single claim goes out.

  4. Stage 03

    Execution

    Recover

    We do the work to bring back what you signed off. Each finding becomes an instrument to follow up with the counterparty, and we help you chase it rather than just handing you a list.

    What we need

    Your existing route to the counterparty, and someone on your side to countersign what goes out.

    What you get

    Debit notes, revised returns, correcting journals and dispute packs, each carrying its evidence. Cash and credit tracked through to settlement.

  5. Stage 04

    The engine turns around

    Prevent

    Everything until now ran over historical data. Prevent points the identical engine forwards: the same rules now test each transaction as it arrives. You built them once; they simply change direction.

    What we need

    A live connection rather than an export, and agreement on who owns each exception type.

    What you get

    The same tests running at the point of entry, exceptions routed with an owner and an SLA, and a leakage rate you can watch decline month over month.

Steps Assess and Recover return the money once. Prevent is the step that stops it going out again, every period, without another engagement.

Indicative for a single process area. Broader scope takes longer, and we will say so before we start rather than after.

Why the DARP Framework

You’ve been offered this multiple times before. And been disappointed every time.

Every CFO we meet has already heard a version of this pitch. Here’s what was wrong with each one, and what we do instead.

What you’ve already been offered

The software vendor

“Buy the platform. The business case comes with it.”

The value is a projection built on someone else’s benchmark, and you pay before it’s tested against a single one of your own transactions.

The recovery audit firm

“We’ll find your duplicate payments for a share of what we recover.”

Money comes back once. Nothing changes upstream, so the same leakage restarts the following quarter and you pay a contingency fee on it again.

The controls programme

“A transformation project to fix the control environment.”

Long, expensive and paid up front, with the benefit still unproven at the exact point you have to commit to it.

What DARP does instead

D

Evidence, not a benchmark

Findings come from your own ledger, not an industry average. The number is yours because the data is yours.

A

A number before a commitment

The value is quantified and agreed before anyone signs for a platform. You decide with the figure in front of you.

R

Funded by money already written off

The engagement pays for itself out of cash the business had given up on. No new budget line to defend.

P

Observability, not another audit

What found the money running backwards runs forwards at transaction entry, permanently instrumented against your goals and the standards. No second build.

The structural point

The diagnostic and the product are one engine pointed in two directions.

Discover, Assess and Recover (DAR) acquires your history, processes and analyses it, and hands you the recovery actions to take. Recover first. Then, if you are happy, point the same actions forwards at transaction entry. Nothing is rebuilt, so no cost for a second build.

Backwards · DAR

Reads your history

Every past transaction is tested against the rules. The output is recovery actions.

Same rules

Built once

Nothing is rebuilt or re-implemented.

Forwards · Prevent

Guards the entry point

The identical rules test each transaction as it arrives.

That is the claim. Here is what it looked like for someone else.

Read a recovery case studySoon

What Discover finds

Four kinds of finding, valued four different ways

These are outcome types, not topic areas. The first two put cash back. The third is audit position with no cash attached, and we say so plainly, because that is what makes the first two credible. The fourth explains why the other three keep recurring.

01

Cash recoverable

Money back to the bank

Puts cash back
  • Duplicate and near-duplicate payments across entities
  • Paid above contract or PO price
  • Tier and volume discounts never applied
  • Early-payment discounts available, not taken
  • Credit notes and debit balances never applied
  • Advances, deposits and retentions not recovered
  • Quantity paid exceeds quantity received

How it is valued

The delta between paid and payable, netted against open vendor balances and prior claims.

02

Tax recoverable

Credit or refund

Puts cash back
  • Input credit unclaimed, or taken on blocked spend
  • Purchase register vs vendor-filed return mismatch
  • Reverse charge missed; wrong place of supply
  • Withholding at the wrong rate or section
  • Thresholds missed on aggregated vendor spend
  • Lower-deduction certificates expired or unapplied
  • Import duty paid, credit never taken

How it is valued

Credit claimable against credit at risk, with the statutory claim window checked per item.

03

Misstated

Wrong numbers, no cash

No cash attached
  • Goods received, not invoiced, not accrued
  • Accrual raised and invoice booked: double count
  • Accruals never reversed the following period
  • Prepaids expensed in full, not amortised
  • Provisions never released after settlement
  • Capex expensed; wrong entity or cost centre
  • FX rate or rate-date errors on open balances

How it is valued

Period-by-period P&L and balance sheet impact. No cash moves; this is the audit-readiness line.

04

Control weakness

Why it keeps happening

Explains recurrence
  • Duplicate vendors; bank shared with an employee
  • Bank details changed just before a large payment
  • Invoices repeatedly just under approval limits
  • Self-approval or approval beyond delegated authority
  • Non-PO spend, retrospective POs, wide tolerances
  • Vendor master edited by the payment approver
  • MSME timelines and sanctions screening breached

How it is valued

By recurrence rate and exposure across the population, not by the size of any one instance.

Recognise one of these? Tell us which, and we will show you where we would look.

The objection

“Our auditors already do this”

They don’t, and it isn’t a criticism of them. An audit exists to form an opinion, not to find money, and those two jobs need different methods.

The audit approach

A sample, against a materiality threshold

Items below materiality are out of scope by design. What comes back is a set of instances (this invoice, that approval) sufficient to support an opinion on the financial statements.

It cannot tell you the rate at which something happens, and it cannot see the thousands of small items that are individually immaterial and collectively worth recovering.

The DARP approach

The full population, then every transaction as it lands

Discover tests every transaction in the history against every rule. Nothing is out of scope for being small, because the value is in the aggregate, and the aggregate is where recovery lives.

Reading everything reports rates, not instances. A rate tells you whether you have an incident or a broken process, which is exactly what makes the case for prevention.

Then Prevent keeps it running. The same tests fire on each transaction in near real time, as it arrives, so the population is never re-sampled after the fact. An audit looks backwards once a year; DARP looks backwards once and then never stops looking forwards.

How full-population testing worksSoonGlossary: sampling, materiality, N-waySoon

What it costs you to find out

Almost nothing, and that is the point

The barrier to knowing what you are losing should not be another project. Here is the entire ask.

Read access, nothing more

An export of the relevant history is enough to start. We do not write to your systems, we do not need a sandbox, and nobody on your team changes how they work.

Handled to enterprise standard

ISO 27001 certified and SOC 2 attested, with role-based access and an immutable audit trail on every record. Cloud agnostic: AWS, Azure, GCP or your own private cloud.

No counterparty cooperation

Discover runs entirely on data you already own. Your vendors, distributors and customers are not involved until you decide to pursue something.

That is the entire ask.

One export~2 min run timeyou know

Where DARP applies

Anywhere there is a reconciliation

DARP is not a channel product. It works wherever two or more records are supposed to agree and nobody has the time to prove it, and the more sources a reconciliation spans, the more it finds. Two-way is useful. N-way is where the value compounds.

Vendor reconciliation (AP)

Invoice against PO against goods receipt against contract price against payment. Duplicates, overpayments, discounts never taken and balances never applied surface where those five stop agreeing.

Customer reconciliation (AR)

Invoice against receipt against bank against gateway settlement, many-to-many. Withheld tax separated from genuine shortfall, so what is truly collectible is stated invoice by invoice.

Taxation (both sides)

Purchase register against vendor-filed returns; output tax against invoices raised; withholding against certificates held. Each item checked against the statutory window still open to it.

Sharpest case

Channel rebates & claims

The sharpest case, because the counterparty controls the evidence. Claims re-priced against the agreement in force on the resale date and tested against POS, roster and shipment history.

Inter-company & the close

Sub-ledger to GL, inter-company balances, accruals and provisions, prepaid schedules, FX revaluation. Where the reconciliation is the close, the finding is a cleaner sign-off.

Payouts, commissions & incentives

Agreement against activity against what was actually paid. Money owed to a third party, calculated outside the ERP, is reconciled the same way as everything else.

Two-way

N-way · where the money hides

The harder the reconciliation, the more DARP is worth. A two-way match between two clean systems is something a competent team already does. A reconciliation spanning five sources, in four formats, against terms that live in a contract nobody has opened since signing, that is where the money hides, and it is precisely the work that never gets done by hand.

Questions we get asked

Before you send us anything

Still something we have not answered?

What exactly do you need to run Discover?

Read access to the transaction history for one process area, and the master data that governs it. For payables that is AP transactions plus the vendor master; for channel rebates it is claims, POS files, contracts and shipment history. An export is enough, a live connection is only needed at the Prevent stage.

How long does it take to get a number?

The Discover run itself takes about two minutes. Once the data is in, the full population is tested against every rule for that process area and the total recoverable figure comes back split across cash, tax, misstatement and control exposure.

What takes longer is everything around it, agreeing scope, NDA, your security review, and pulling the extract. That is usually days rather than weeks, and none of it is analysis time. Nobody spends a quarter in fieldwork before you see a figure.

Do our vendors, distributors or customers need to be involved?

Not for Discover or Assess. Both run entirely on data you already hold, which is why the first conversation costs nothing but data access. Counterparties only enter the picture at Recover, and only for the findings you decide to pursue.

How is this different from our recovery audit firm?

Three ways. They sample; we read the full population. They take a contingency share and leave, so the same finding recurs next year; we turn each root cause into a control that runs at transaction entry. And their output is their file, ours is a system of record you keep, with the evidence attached to every finding.

What if you don’t find much?

Then we tell you that, and you owe us nothing. Discover runs before any contract exists.

You still end up with something useful: independent evidence, tested across your full transaction population rather than a sample, that this process is clean. That is a straight answer to a question your board or your auditors will ask eventually, and it cost you one data export.

We would far rather report a small number than talk you into a big one we cannot stand behind. The number has to survive your team, your auditor and the counterparty you send a claim to.

Does control weakness actually return cash?

Mostly it does not, and we say so up front. Cash only comes back where a loss actually crystallised. The value in that column is avoided loss and audit position, and being straight about it is what makes the cash and tax numbers credible.

Do we have to take Prevent?

No. You can stop after Recover with the cash and the findings. Prevent exists because the alternative is running the same diagnostic again next year, and since the rules already exist, turning them forwards is a deployment rather than a fresh build.

Where does our data sit?

It depends on the stage, and the distinction matters.

Discover and Assess run in the DataTwin cloud. You send an extract, we analyse it there, and you get the findings back, so there is nothing to install and no infrastructure decision to make before you know whether the number is worth acting on.

Prevent is deployed wherever you want it. It is cloud agnostic and runs on AWS, Azure, GCP or your own private cloud, so the system that sits permanently alongside your ERP lives inside your own estate and under your own controls.

Throughout: ISO 27001 certified, SOC 2 attested, role-based access, data residency options and an immutable audit trail on every record.