GGA RESEARCH & METHOD
A Practical Debtor Scoring Method
A working scoring model for trade channel receivables, built to run on a debtor ageing report and nothing else — no purchase history, no CRM extract, no system. Four classes, one judgement about your billing rhythm, and a customer-level list nobody in the room can argue with.
The argument
A debtor ageing tells you how old a balance is. Scoring tells you what to do about it — and the difference is one afternoon of work on data you already have.
In short
Why this exists. Most companies segment debtors by how much is owed or how old it is. Neither tells you what to do. Scoring by customer behaviour does — and it needs no data you do not already hold.
- It runs on a debtor ageing alone. Customer code, total balance, ageing buckets. No purchase history is needed. Where billing or purchase history is unavailable, the presence of recent open invoices is used as a proxy for continuing commercial activity. It does not establish that a customer with no recent outstanding has made no recent purchases — they may have bought and already paid.
- Four classes, tested in strict order. D legal, then A where 20% or less of balance sits beyond the buying window, B above 20% and below 80%, and C at 80% or more. The first rule a customer satisfies is their class.
- One judgement is required from you — whether your business is seasonal or not. That declaration sets the buying window: 90 days non-seasonal or 180 days seasonal.
- It applies to trade channel customers only. Dealers and distributors buy repeatedly, which is what makes a gap in purchasing meaningful. Institutional and government buyers purchase on tender cycles and must be read separately.
- Read each example by its own labels. The retained figures describe published examples. They do not validate the current calculator or establish performance on another book.
- The output is a customer-level list. Which is the point: it cannot be negotiated by whoever owns the relationship.
Before you start
One scope rule. This model is for trade channel customers — dealers and distributors who buy repeatedly. Institutional buyers, government departments, armed forces and e-commerce marketplaces buy on tender cycles, contract renewals or platform terms, and a quarter of silence tells you nothing about them. Filter them out before scoring.
Keep a record of the customers included and excluded before scoring. Attribute a change in a class share to filtering only when the same dated ledger is compared under the same rules.
And strip out what is not a receivable: retention, security deposits, amounts not yet due, and tax deducted at source. Scoring an unadjusted ledger produces a worse-looking book than the company actually has.
Step one — four fields
| Field | Why |
|---|---|
| Customer code and name | The unit of analysis. Score the customer, not the invoice. |
| Total outstanding balance | The denominator. Every threshold is tested against the customer's own balance, never an absolute rupee figure. |
| Ageing buckets | Buckets must split balances within and beyond the selected buying window: 90 days non-seasonal or 180 days seasonal. |
| Legal status flag, if you have one | Separates accounts already in legal process. Apply D only where an affirmative legal flag is supplied. |
Step two — declare your buying window
This is the one judgement the model asks of you. It sets the buying-window boundary at 90 days for non-seasonal businesses or 180 days for seasonal businesses.
| Billing rhythm | Buying window | Typical businesses |
|---|---|---|
| Non-seasonal | invoiced within 90 days | footwear, FMCG, lubricants, electricals, building materials, sports goods |
| Seasonal | invoiced within 180 days | agri inputs, woollens, cooling, fans, festival goods |
The far end of the window is the one immediately after normal terms have run out — money past due but not yet aged, on a customer still trading. It is the cheapest money on the ledger to collect and the most time-sensitive.
Step three — score four classes, in order
The first rule a customer satisfies is their class. Stop there.
| Order | Class | Test | What it means |
|---|---|---|---|
| 1 | D — legal | Flagged as in legal process and still in the debtors ledger | Separate for legal follow-up; no write-off, provisioning or DSO decision. |
| 2 | A — buying and paying | 20% or less of balance beyond the buying window | Pays substantially to terms. A routine internal cycle, nothing more. |
| 3 | B — buying, delaying | Above 20% and below 80% beyond the window | Still transacting while carrying aged balance. Supply leverage exists now and will not later. |
| 4 | C — not buying, not paying | 80% or more beyond the window | The relationship is a recovery problem regardless of any recent order. No leverage remains. |
Why C and declining-C are not separated. Without an affirmative legal flag, 80% or more beyond the buying window scores C, irrespective of recent purchasing. D is reserved for an affirmative legal flag; it is not a further ageing band.
Why B is the intermediate ageing-share class. After the legal override, B applies when more than 20% but less than 80% of a balance is beyond the declared buying window. It identifies a balance to review. Recent outstanding invoices do not prove continuing purchases, and the class alone does not establish a cause or a recovery outcome.
Does it actually discriminate?
The following table retains two anonymised examples as published. It reports three value shares under its original labels; these are not a recomputation under the current four-class calculator.
| Class | Agri inputs — seasonal, 765 dealers, ₹57.02 cr | Sports goods — non-seasonal, 1,470 accounts, ₹62.03 cr |
|---|---|---|
| A — buying and paying | 11.7% | 40.2% |
| B — buying, delaying | 10.5% | 8.8% |
| C — not buying, not paying | 77.8% | 51.0% |
The table reports class A at 11.7% for agri inputs and 40.2% for sports goods. Read those shares under the example labels. They are not substitutes for outputs from the current calculator.
The agri example reports 84.2% of value beyond 180 days, 72.2% beyond a year and 32.7% beyond two, with 2.9% current. Its table allocates 77.8% to C. These ageing figures describe open balances; they do not by themselves establish that dealers have stopped trading or that balances cannot be recovered.
The sports goods table totals ₹62.03 crore. C is its largest class at 51.0%, followed by A at 40.2% and B at 8.8%. At that total, B represents approximately ₹5.46 crore. These shares do not predict recovered cash or future movement between classes.
An ageing report records unpaid balances, not a complete purchase history. Confirm recent orders or invoices before describing a customer as active or inactive. A customer with no recent open balance may have bought and already paid. Decisions about supply, field verification or settlement require that wider account evidence.
Step four — read the output
| Signal | What it usually means |
|---|---|
| B is more than 40% of value | Review ageing, terms and disputes. A high B share does not identify its cause. |
| C is larger than B | More value is in the higher ageing-share class. Check the reasons account by account. |
| Most of C still buying | Confirm recent orders before considering supply leverage. Ageing alone does not confirm activity. |
| D exceeds 15% of customers | Current class D denotes a legal flag. Confirm status and the appropriate next step. |
| E still inside the debtors file | E is a historical legal label; current calculator D serves that role. Review accounting separately. |
| A over 50% of value but DSO still high | Review concentration and timing together. A low aged share does not explain whole-book DSO. |
Then run three diagnostics on the scored list
Ageing bucket within class. A B class weighted toward the far end of the window is C being manufactured.
Re-apply receipts against the invoice they were intended for. Take the twenty largest persistent balances and match each receipt to the invoice its remittance advice referenced, rather than to the oldest open item. Where the profile shifts materially, first-in-first-out application has been concealing a dispute rather than resolving one.
Estimated carrying cost against class EBIT. Average outstanding balance × annual funding rate × days in the stated period ÷ 365. Compare with EBIT for the same period. A 10% annual rate is illustrative; use the applicable funding rate.
What the model will not do
It will not tell you whether a customer is strategically important. The score describes payment behaviour and nothing else. A significant customer can score C; that is information, not a verdict, and the commercial decision remains yours.
It will not work on institutional or government buyers. The activity test carries no information where purchasing runs on tender cycles.
The thresholds are working conventions, not empirically derived boundaries. The 20% and 80% lines and the 90-day and 180-day buying windows are calibrated to typical billing rhythms. Where your cycle is unusual — annual capital equipment, twice-yearly festival billing — set the bands to your actual cycle rather than the category default.
And scoring is a point-in-time cut. A book scored at quarter end after a collection push reads better than the same book scored mid-quarter. Score consistently or the comparison across periods means nothing.
Where it fits
This is the Intelligence stage — establishing what is actually in the book. It tells you which customers need what, and where the recoverable money sits. It does not make an invoice payable, and it does not collect anything.
Assurance is the work that follows: locating agreements, matching purchase orders, evidencing deliveries, completing onboarding, closing reconciliations — making each balance collectable before anybody is asked to pay it. Realization is the follow-up that turns each payable balance into cash in your bank. Most companies begin at the third stage, which is why so much effort produces so little.
Method and limitations
Why 20% and 80%. The lower cut-point matches what practitioners already use — one client ledger carries its own definitions in the file: “A: more than 6 months is less than 20%” and “C: more than 1 year is 20–50%”. The model formalises a convention rather than importing one. The upper cut-point is set where a balance stops being a delay and becomes a recovery: at 80% beyond the window, one recent order does not change what the account is.
Each cut-point governs one boundary. Changing the lower cut-point can move accounts between A and B; changing the upper cut-point can move accounts between B and C. D remains the first test and applies where the legal flag is affirmative.
The buying window is set by your billing rhythm, not chosen. Ninety days where customers order monthly or quarterly, 180 where they order by season. It stays fixed between periods, because changing it changes the classification without anything changing in the book.
The retained published table describes a seasonal agri inputs book of 765 dealers and ₹57.02 crore, and a non-seasonal sports goods book of 1,470 accounts and ₹62.03 crore. It preserves the earlier example labels and figures; it is not a current-rule calibration or a new run of the underlying ledger.
These anonymous examples illustrate how to read reported class shares. Their relationship to the dealer-only examples in the linked study is not established here. Do not combine the populations or compare their class labels as if they were one model run.
The band boundaries are operating thresholds, not externally validated predictions. Assess sensitivity using the same dated customer list and balances. A change in classification after changing a threshold does not by itself establish better recovery performance.
The earlier client-classification examples in the GGA Debtor Segmentation Model Study total ₹83.42 crore across 1,575 dealers. Their historical labels differ from the current calculator. Use the four-class definition stated here for current scoring.
No client, dealer or individual is identified. GGA does not publish the payment behaviour of any named company.
Definitions
| Term | Definition |
|---|---|
| Buying window | The recent period against which each balance is measured — three months for non-seasonal businesses, six for seasonal. Declared before scoring, and stated alongside any output because it is the most consequential choice in the model. |
| Share beyond the window | Balance dated outside the buying window divided by that customer’s total balance. The single measure that sets the class. |
| Old band | Balance beyond the selected buying window — 90 days non-seasonal or 180 days seasonal — as a percentage of that customer's total balance. |
| Supply leverage | The ability to influence payment because the customer still needs supply. Available only while they are still buying, and checked separately from the ageing score. |
| FIFO application | Applying an incoming receipt against the oldest open invoice first. Correct for accounting, misleading for diagnosis — it settles disputed invoices with money intended for current purchases. |
| Intelligence stage | Establishing what is actually in the book, before any attempt to make invoices payable or to collect. The stage this tool belongs to. |
Frequently asked questions
How do I segment my dealer or distributor base?
A debtor ageing report is enough — no purchase history needed. Where billing or purchase history is unavailable, the presence of recent open invoices is used as a proxy for continuing commercial activity. It does not establish that a customer with no recent outstanding has made no recent purchases — they may have bought and already paid. Filter to trade channel customers only and strip out retention, security deposits, not-yet-due amounts and tax deducted at source. Declare your billing rhythm: split each customer's total balance into amounts within and beyond 90 days for non-seasonal businesses or 180 days for seasonal businesses. Then score four classes in strict order — D legal, then A where 20% or less of balance sits beyond the buying window, B above 20% and below 80%, and C at 80% or more.
Do I need purchase history to score a debtor book?
No. The calculator uses balances within and beyond the selected buying window, plus the legal flag if supplied. The window is 90 days for non-seasonal businesses or 180 days for seasonal businesses. An empty recent bucket does not establish when a customer last bought or was invoiced; recent purchases may already have been paid.
Why does seasonality change how a debtor book should be scored?
Seasonality changes the buying-window boundary: 90 days for non-seasonal businesses and 180 days for seasonal businesses. Split the balances at the selected boundary before scoring. The calculator classifies the share beyond that window; it does not infer actual purchase history from an empty current bucket.
What is the warning class in debtor scoring?
In the current calculator, B means more than 20% but less than 80% of the balance is beyond the declared buying window, after the legal override. It is a prompt to review the account. The ageing share alone does not prove current purchasing, explain a dispute or predict recovery.
Can this scoring model be used on government or institutional customers?
No. The model is intended for dealers and distributors with repeated purchases. Institutional, government, armed-forces and marketplace balances can follow tender, contract or platform cycles. Keep the eligibility rule and excluded accounts on record before comparing class shares.
What should I look at once the book is scored?
Three diagnostics. Ageing within class — a B class weighted toward the far end of the window is turning into C. Receipt re-application — match each receipt to the invoice its remittance advice referenced rather than to the oldest open item, and see whether the profile shifts. And estimated carrying cost against EBIT for the same period; the 10% annual rate is illustrative.
How do I know the scoring is telling me something real?
Check the rules against known account facts and subsequent outcomes on a defined population. The retained table reports A at 11.7% for agri inputs and 40.2% for sports goods, with C the largest class in each. Different reported shares alone do not validate the current calculator or predict recovery.
Evidence boundary
This publication distinguishes reported examples, modelled illustrations and operating observations. The examples do not validate a calculator or predict the outcome of an individual receivable, counterparty or company.
Public law, policy terms and procedures can change. Current primary material and the facts of the live case govern.