Contract integrity & program intelligence for loyalty programs

Your breakage is modelled. Your invoice is unchecked. Both are assumptions.

You buy the points. They calculate the bill. Nobody checks it against the contract.

We recalculate every transaction against the contract itself — to show you what you were owed, and what your terms are actually worth.

Worked example — synthetic contract and transaction population, built to mirror a real co-brand agreement.

The problem

Nobody compares the document to the data.

Commercial signs the contract. A vendor builds the rules. Finance pays the invoice. No one holds both ends.

Most agreements already give the operator a right to audit. It is rarely used, because exercising it costs money and reads as hostile toward a partner the program depends on.

The annual audit checks a few hundred. We check all of them.

A misimplemented rule is not an error — it is a consistent difference applied to every transaction it touches. Sampling cannot find it. It shows up as divergence.

~300Audit sample

2,417,338Population tested

Worked example — synthetic contract and transaction population, built to mirror a real co-brand agreement. PointSphere has not yet published findings from a live engagement.

Clause

4.2 — Earn capped at 14 pts per USD 100Cap not enforced in production

Contract derives

System issued

Variance

6.9% over, three periods running

Clause

7.1 — Multiplier applies to base earn onlyApplied to bonus earn in three campaigns

Contract derives

System issued

Variance

22.3% over, compounding monthly

Next — how a rederivation actually runs

Where the complexity lives

The difficulty is not the arithmetic. It is the count.

Every rate, cap, tier and exclusion below is contractually defined — and implemented by hand, once, years ago.

Bank co-brand

One agreement, deep clause structure. Caps, tier multipliers, promotional overlays, funding rates.

1 partner × 60+ clauses

Airline and hotel

One currency sold to many partners at many prices, with liability carried on your balance sheet.

1 currency × 20+ rate cards

Coalition

Every partner both issues and redeems. Obligations run in both directions and net against each other.

12 partners × 132 positions

Program platforms

Operators running the same infrastructure across many bank programs, each on its own contract.

1 platform × 20+ programs

Built for bank co-brand, coalition, airline, hotel and retail programs — and the platforms that operate them. Wherever a contract sets the rate and a system does the issuing, the two can diverge.

Next — how it runs

How it runs

Four steps, on documents the program already produces.

01You send two inputsOne live contract under NDA, and the transaction files for the period it governs. Standard monthly extracts are sufficient.
02The contract becomes rulesEvery rate, cap, tier multiplier, exclusion and funding split is extracted and structured. You review and confirm the rule set before anything runs.
03The full population is rederivedEvery transaction is recalculated against those rules, and what the contract derives is placed beside what the system issued.
04You receive a divergence reportEach variance traced to the clause that defines it, quantified in points and in currency, with its trend and its onset date.

Findings are reported in whichever direction they run. Some of what a rederivation surfaces will be in the partner’s favour rather than the operator’s, and it is reported identically. An engine that only ever finds overbilling is not a measurement, it is a position.

Next — what a variance looks like over time

Divergence

A variance is an error. A trend is a term you no longer have.

One rule, misimplemented, produces a widening one-directional gap — not noise. Tier I measures and projects it.

Drift has a start date. Usually a release, a repricing, or an amendment nobody re-tested. Finding the onset is what separates this from a reconciliation.

What the contract requires

What the system issued

Tier I · Derived

Drift

What the contract requires against what the system issued, period over period. No model and no assumptions — it falls out of the document and the transaction file.

Tier II · Modelled

Breakage

Points that will never be redeemed, estimated from your own history rather than carried at a flat assumed rate. Needs data, not just the contract.

Why it doesn’t end when the errors are fixed

Contracts get re-cutEvery repricing is a new rate card implemented in someone else’s billing system.

Campaigns launch monthlyEach one carries its own funding rules, caps and eligibility conditions.

Mappings changeCategories get reclassified. Nobody re-tests the accrual logic against the schedule.

Amendments land mid-termOften with retroactive effect, and often implemented from the wrong date.

The errors are a stock. The change is a flow. We are there the month the implementation changes.

The same derivation, asked forward

Finding the gap is half of it. The other half is knowing what your terms are worth.

Once the contract is machine-readable and the population has been rederived once, the same engine answers questions that have nothing to do with error. This is where the work stops being remedial.

Contract modelling What would a different clause have paid? Run last three years of transactions against a proposed rate card before you sign it. Walk into a renewal knowing what each term actually delivered, not what it was projected to.

Campaign pre-flight Is this promotion funded correctly? Test a campaign against the agreement before launch — caps, funding splits, eligibility, stacking rules — rather than discovering the breach in arrears.

Margin leak Which clause is costing you? Decompose the economics by clause rather than by month. The exclusions, the caps, the multiplier scopes — priced individually, over the full population.

Rate benchmarking How do your terms compare? What the category actually funds, derived from agreements rather than surveyed. Requires scale, and cannot be assembled by anyone sitting on one side of a deal.

Conformance is what gets bought first. This is what makes it worth keeping. See the tiers

Next — what the exposure is worth

What it is worth

A miscoded rule is a recurring cost that runs until someone finds it.

Where the issuer funds the points, every point issued beyond the contract terms is a direct overpayment on the invoice. The exposure is a function of portfolio size and the size of the error, and it accrues every month until it is corrected.

Annual co-brand rewards spend0.25% error1% error4% error

$25m$63k$250k$1.0m

$100m$250k$1.0m$4.0m

$400m$1.0m$4.0m$16.0m

Arithmetic, not a claim. The relevant question is which column a program is in, and no one currently measures it.

The other side of the ledger

Breakage is not in the contract. It is management’s estimate of the points that will never be redeemed — revised every period, with revisions running through current-period revenue. It is produced by a model rather than derived from the redemption population it describes, which means an error is not a one-off misstatement. It corrects into earnings.

Point liability on the balance sheet1pt3pt5pt

$80m$1.0m$3.0m$5.0m

$400m$5.0m$15.0m$25.0m

$1.6bn$20.0m$60.0m$100.0m

Movement in the liability for a 1, 3 or 5 percentage point error in the assumed redemption rate, at an assumed 80% redemption. Arithmetic again, not a claim — but the assumption is the programme’s own, and the population that would settle it is already sitting in the ledger.

Corrected billingQuantified evidence of excess issuance, clause by clause, supporting correction going forward and recovery of past overpayment.

A stronger audit positionFull-population testing with a documented method gives auditors more than a sample. Breakage reserves and point liability rest on evidence.

Renegotiation from factsA term unenforced for years means the program has been operating on different economics than were signed. The report shows which terms are actually in effect before a renewal opens.

Errors caught before launchOnce the rules are structured, new campaigns and amendments can be tested against them before they reach production.

What we build

Three tiers. Each earns the next.

Nothing in an outer ring can be sold without the ring inside it. That is the constraint, not the pitch.

IRecalculate the contract against every transaction

IIThe same derivation, pointed at what your terms are worth

IIINet and settle between programs on a shared derivation

Conformance and drift

Your contract becomes rules. Every transaction is recalculated against them.

Full population — no sampling
Every variance traced to its clause
Monthly files. No integration

AvailableContract documents and monthly files

Program intelligence

Questions that need your history, not just your contract.

Breakage and liability, modelled
Margin leak and partner benchmarking
Campaign checks before launch

Built on your historyScoped after the first engagement

III

Clearing

Rederive several programs against one standard and settlement stops requiring reconciliation.

Netting across counterparties
Disputes against a shared derivation

TrajectoryNot built, not sold today

The first engagement

One contract. Four to six weeks.

Send one live contract and the transaction history for the period it governs. You get the rederivation and the divergence read.

Designed to run monthly thereafter — against each invoice, inside the window before you pay it.

You provide: One contract, under NDA
And: Transaction files
Duration: Four to six weeks
Integration: None
Live access: None
Your engineering time: None

Request an engagement For finance, audit and controllership. Or write directly to hello@pointsphere.ai.