▤ Insights · credit & collections
Credit management, collections & disputes.
A practitioner's playbook covering strategy, process design, tooling and governance — from low-volume SMB environments to high-complexity global enterprise operations.
Credit risk
Collections strategy
Dispute management
GBS / SSC design
AI & automation
SAP S/4HANA
Credit management, collections and dispute handling sit at the revenue-protection heart of order-to-cash. Poor execution directly erodes cash flow, inflates DSO, creates bad-debt provisions and damages customer relationships. Best-in-class organisations treat this domain not as a back-office enforcement function but as a proactive, data-driven, customer-centric discipline — balancing risk control with commercial growth.
Cash impact
1–3%
Revenue at risk from unmanaged credit & disputes (typical B2B)
DSO reduction
8–15d
Achievable DSO improvement through structured collections
Dispute cost
€50–500
Average cost per dispute to resolve (manual)
Auto-rate target
70%+
Disputes auto-routed or resolved without analyst touch
Bad debt
0.05–0.3%
Best-in-class bad debt as % of revenue (B2B)
Credit risk
Assess, limit and monitor exposure before and during the customer lifecycle. Prevents revenue becoming uncollectable.
Scope: onboarding → limit setting → ongoing monitoring → review triggers
Collections
Structured outreach and escalation to recover overdue receivables — from proactive pre-due reminders to legal action.
Scope: ageing analysis → dunning → negotiation → legal
Dispute management
Receive, classify, root-cause, resolve and learn from invoice and payment disputes — reducing both volume and cycle time.
Scope: capture → triage → resolution → root-cause → prevention
Effective credit risk management starts before the first invoice is issued and continues throughout the customer relationship. The goal is calibrated exposure — not zero risk, but risk commensurate with the commercial value of the relationship.
STEP 01
Customer onboarding & KYC
Trade references, credit-bureau pull, legal-entity check, ownership verification
STEP 02
Credit scoring
Internal score + external rating → risk-tier classification (A/B/C/D)
STEP 03
Limit setting
Credit limit = f(score, exposure appetite, trading history, guarantees)
STEP 04
Order-release control
Automated hold/release against limit, exception workflow for overrides
STEP 05
Ongoing monitoring
Event triggers: late payment, news alerts, bureau updates, limit utilisation
STEP 06
Annual / event review
Re-score, adjust limit, reclassify, escalate or exit relationship
Low volume / SMB
- Manual credit checks via Creditsafe / D&B
- Single credit limit per customer, no tiering
- Excel-based exposure tracking
- Owner/manager approval for new accounts
- Prepayment or personal guarantee for unknowns
- Quarterly portfolio review
Mid volume / regional
- ERP-integrated credit limits with hard/soft blocks
- Risk tiering (A–D) driving payment terms
- Automated credit-bureau feed (monthly refresh)
- Exception workflow for order holds
- Credit insurance for top-exposure accounts
- Trade-credit-insurance portfolio management
High volume / global enterprise
- Predictive ML credit scoring (PD / LGD models)
- Real-time bureau integration (D&B, Allianz Trade)
- Group exposure view across legal entities
- Dynamic credit limits (behaviour-adjusted)
- Credit-committee governance for strategic accounts
- Concentration-risk analytics & country-risk overlay
| Technique | Description | When to use | Tools / sources |
|---|
| Credit-bureau pull | External credit score, payment history, legal events from bureaus | All new customers; periodic refresh for existing | D&B, Creditsafe, Atradius, Experian, Graydon |
| Internal scoring model | Proprietary score combining payment behaviour, ageing, order history | Mid–high volume; after 6–12 months of trading data | SAP FI-AR, Salesforce Revenue Cloud, Python ML models |
| Trade-credit insurance | Policy covering receivables against customer insolvency | High concentration risk, export markets, strategic accounts | Allianz Trade, Atradius, Coface |
| Letters of credit / guarantees | Bank-guaranteed payment instrument for high-risk counterparts | New entrants, high-risk geographies, large single orders | SWIFT MT700, bank treasury teams |
| Prepayment / proforma | Payment required before goods/services delivered | Unknown customers, small orders, high-risk profiles | ERP order management, payment gateway |
| Dynamic limit adjustment | Algorithm auto-adjusts limit based on payment-behaviour signals | High-volume, data-rich environments | HighRadius, Cforia, Billtrust, custom ML pipeline |
| Concentration-risk analysis | Monitors % of total AR exposed to single customer/sector/country | Portfolio-level risk governance, credit-committee reporting | Power BI, SAP Analytics Cloud, Tableau |
Collections is a segmented, prioritised, relationship-sensitive discipline. Blanket dunning destroys commercial relationships. Best practice segments the AR portfolio by risk, value and behaviour — and applies tailored, escalating outreach accordingly. The emphasis is always on proactive engagement before invoices fall overdue.
T–7d
Pre-due reminder
Friendly statement / invoice copy. Confirm receipt, flag upcoming due date
T+1d
First dunning
Automated email/portal notification. No escalation language. Check for dispute
T+10d
Analyst call
Personal outreach. Identify reason for delay. Agree payment commitment
T+20d
Senior escalation
Team-leader / manager involvement. Consider credit hold. Payment plan
T+45d
Commercial loop-in
Sales / account manager engaged. Relationship vs risk decision. Formal notice
T+60d+
Legal / agency
Formal demand letter. Collection-agency referral. Legal proceedings / write-off
Portfolio segmentation matrix
| Segment | Profile | Strategy | Channel mix | Priority |
|---|
| Strategic / key accounts | High value, low risk, proven payer | Proactive relationship management; dedicated collector; executive level if needed | Phone, email, personal visit | HIGH |
| High value / high risk | Large balances, poor payment behaviour or deteriorating credit | Early escalation, credit-hold triggers, commercial alignment, payment plan | Phone, formal letter, legal hold | CRITICAL |
| Mid-market standard | Mid-value, average risk, irregular payers | Structured dunning cadence; payment-commitment tracking; dispute check | Email + portal + phone | MEDIUM |
| Long tail / small customers | Low value, high volume, mixed behaviour | Fully automated dunning; self-serve portal; agency for chronic delinquents | Email, SMS, portal, automated call | LOW (AUTO) |
| Disputed / pending | Any value; payment withheld pending resolution | Pause collections; fast-track dispute resolution; collect undisputed portion | Dispute-workflow tool | HOLD |
| Payment plan / restructured | Agreed instalment schedule | Monitor adherence; auto-alert on missed instalment; renegotiate if needed | System-driven reminders | MONITORED |
Low volume — manual
- Excel ageing report, manual outreach
- Outlook-templated dunning emails
- Owner/bookkeeper makes calls
- FreshBooks / Xero auto-reminders
- Local collection agency if needed
Mid volume — semi-automated
- ERP dunning levels (SAP FI, D365, NetSuite)
- Collector workqueues with daily prioritised task lists
- CRM call logging and commitment tracking
- Customer self-service payment portal
- Promise-to-pay tracking & breach alerts
High volume — intelligent automation
- AI-prioritised worklists (propensity-to-pay scoring)
- Omnichannel outreach: email, SMS, WhatsApp, IVR
- Predictive cash forecasting from commitment data
- RPA for statement generation & account reconciliation
- Global agency-network management
- Real-time DSO dashboard & portfolio-health analytics
Golden rule: always collect the undisputed portion
When a customer raises a dispute on part of an invoice, do not pause collections on the entire balance. Separate the disputed amount into a defined dispute bucket and continue actively collecting the undisputed portion. This single discipline is one of the highest-impact practices to protect cash flow in complex B2B environments.
Disputes are not just collection problems — they are process-quality signals. Every dispute category points to a failure upstream: wrong price, delivery issue, damaged goods, PO mismatch. Best-in-class organisations resolve disputes fast, track root causes systematically and feed insights back into O2C process improvement.
Common dispute categories & root causes
| Dispute type | Root cause | Resolution owner | Target cycle time | Prevention lever |
|---|
| Pricing / rate dispute | Incorrect price on invoice vs contract or quote | Credit team + Commercial | < 5 days | Contract-to-invoice price-validation automation |
| Short / non-delivery | Goods not received or incomplete shipment | Logistics / Ops | < 10 days | Proof of delivery (POD) capture; track-and-trace |
| Quality / damage | Goods received but not to specification or damaged | Quality / Supply Chain | < 15 days | Improved packaging, carrier SLAs, QC inspection |
| Duplicate invoice | Same invoice issued twice in different periods/systems | AR / Billing | < 2 days | Duplicate detection at invoice creation (ERP rule) |
| PO / contract mismatch | Invoice references wrong or expired PO | AR + Procurement (customer side) | < 7 days | PO validation at order entry; pre-invoice confirmation |
| Unauthorised deduction | Customer deducts discount, rebate or claim not agreed | Credit + Commercial | < 10 days | Rebate-management discipline; deduction-tracking system |
| Tax / VAT dispute | Incorrect VAT rate, exemption-certificate issue, cross-border complexity | Tax + AR | < 20 days | Tax-engine integration (Vertex, Avalara) |
| Payment-application error | Payment received but applied to wrong invoice | Cash Application / AR | < 1 day | AI-powered cash application; remittance matching |
Dispute lifecycle — end-to-end process
PHASE 1
Capture & log
Multi-channel intake: email, portal, phone, EDI. Auto-create case with all attributes
PHASE 2
Validate & triage
Classify type, validity check, assign priority, route to resolution owner
PHASE 3
Investigate
Gather evidence: POD, contracts, price files, photos. Set SLA clock
PHASE 4
Resolve
Credit note, re-invoice, reject with rationale, or escalate. Customer comms
PHASE 5
Close & record
Post resolution, update AR, release collections hold, confirm with customer
PHASE 6
Root-cause & prevent
Tag root cause, escalate patterns, feed process-improvement backlog
Low volume (<50 disputes/mo)
- Shared inbox + Excel dispute log
- Manual email communication with customer
- Owner or senior AR handles all types
- No formal SLA tracking
- Root cause verbal / ad hoc
Mid volume (50–500 disputes/mo)
- ERP dispute case management (SAP UDM)
- Type-based routing to resolution teams
- SLA tracking in CRM/ticketing tool
- Customer self-service dispute submission
- Monthly root-cause reporting
High volume (500+ disputes/mo)
- Dedicated dispute-management platform
- AI auto-classification and evidence matching
- Auto-resolution for defined dispute types
- Real-time SLA dashboards with breach alerts
- Pareto analysis → cross-functional fix sprints
- Dispute-prevention KPIs tied to upstream teams
The dispute-to-prevention flywheel
Resolution without root-cause analysis is just keeping score. The real value lies in tagging every dispute with a structured root-cause code, aggregating patterns monthly, and running cross-functional improvement sprints. Organisations that do this consistently reduce dispute volume by 30–50% within 18 months — without any additional headcount.
Measuring the right things drives the right behaviours. A well-designed KPI framework covers four dimensions: cash performance, process efficiency, risk quality and customer experience. Avoid vanity metrics — every KPI must be actionable.
Core KPI framework
| KPI | Definition | Good | Best-in-class | Dimension |
|---|
| DSO (days sales outstanding) | AR balance ÷ (revenue ÷ days). Measures the average collection period | 30–45 days (B2B) | < terms + 5 days | Cash |
| CEI (collection effectiveness index) | % of collectable AR actually collected in period. Better than DSO for seasonal businesses | > 85% | > 95% | Cash |
| Overdue AR % | Overdue balance as % of total AR | < 20% | < 8% | Cash |
| Bad debt % | Write-offs as % of revenue | < 0.5% | < 0.1% | Risk |
| Credit-limit utilisation | Exposure vs approved limits; concentration monitoring | < 80% portfolio average | Dynamically managed | Risk |
| Dispute rate | Disputes raised as % of invoices issued | < 3% | < 1% | Quality |
| Dispute resolution time | Average calendar days from dispute open to close | < 15 days | < 5 days | Quality |
| First-contact resolution % | Disputes resolved without escalation beyond initial contact | > 60% | > 80% | Quality |
| Promise-to-pay kept % | % of payment commitments honoured on the agreed date | > 70% | > 88% | Process |
| Collector productivity | Accounts managed per FTE (varies by segment) | 300–500 | 800–1200 (with automation) | Process |
| Cash-forecast accuracy | Predicted vs actual collections within rolling 30-day window | ±15% | ±5% | Cash |
| Customer satisfaction (CX) | NPS / CSAT from customers who experienced collections contact | Neutral | Net positive | Experience |
Organisations evolve through four maturity stages. Knowing where you are is the starting point for any transformation roadmap. The jump from Level 2 → 3 (structured → proactive) typically delivers the highest ROI; Level 3 → 4 (proactive → predictive) requires significant technology investment but unlocks sustained competitive advantage.
| Level | Label | Credit risk | Collections | Disputes | Technology |
|---|
| Level 1 | Reactive | No formal limits; react after loss | Chase when cash is needed; no cadence | No tracking; informal resolution | Excel, email |
| Level 2 | Structured | Credit limits set; bureau checks at onboarding | Dunning levels in ERP; ageing reports | Dispute log; type classification | ERP (SAP, D365, NetSuite) + basic CRM |
| Level 3 | Proactive | Risk-tiered portfolio; ongoing monitoring; TCI | Segmented strategy; PTP tracking; pre-due outreach; portal | SLA-tracked; root-cause reporting; prevention KPIs | Collections platform + dispute tool + analytics |
| Level 4 | Predictive | ML credit scoring; dynamic limits; concentration dashboards | AI worklists; propensity-to-pay; cash forecasting; omnichannel | Auto-classification; auto-resolution; flywheel prevention | HighRadius / Cforia / Billtrust + AI + RPA + S/4HANA |
Credit-policy essentials
- Credit-limit matrix — authority levels by risk tier and exposure size
- Payment-terms standard — net 30/45/60 mapped to risk tier
- Order-hold criteria — when to block, who can release and under what conditions
- Review frequency — annual minimum; event-triggered for high-risk
- Write-off authorisation — thresholds and approvers clearly defined
- Escalation matrix — from analyst → team lead → credit manager → CFO
- Trade-credit-insurance rules — coverage thresholds, uninsured-limits policy
Collections-policy essentials
- Dunning cadence by segment — defined days, channels, tones per tier
- PTP management — max extensions, commitment logging, breach triggers
- Payment-plan authority — who can agree, maximum duration, discount limits
- Agency-referral criteria — age, balance, last contact outcome
- Legal-action thresholds — minimum exposure, evidence requirements, approval chain
- Commercial-alignment protocol — when and how sales are looped in
Dispute-management policy
- Dispute-acceptance rules — what constitutes a valid dispute
- SLA targets by type — acknowledged within 48h; resolved per category table
- Credit-note authority matrix — up to €X without approval; €X–Y requires manager
- Root-cause code taxonomy — standardised across all teams
- Dispute-hold protocol — how to flag and exclude from active dunning
- Deduction-management rules — valid deductions vs unauthorised; recovery process
Governance rhythms
- Daily: collector worklist review; order-hold queue; dispute SLA-breach alert
- Weekly: collections pipeline review; PTP adherence; ageing flash
- Monthly: DSO / CEI / bad-debt review; dispute root-cause Pareto; credit-risk portfolio
- Quarterly: credit-limit review; policy-compliance audit; KPI benchmarking
- Annual: full policy review; credit-insurance renewal; team-capability review
Collections escalation & governance timeline
Pre-due contact T–7 to T–3 days
Automated statement, invoice copy or proactive call to key accounts. Confirm receipt and payment intent. Analyst authority. No escalation needed.
Standard collections T+1 to T+20 days
Dunning levels 1–2. Analyst-managed. PTP obtained and tracked. Dispute check mandatory. Payment plan possible with team-lead approval up to 60 days.
Senior escalation T+21 to T+45 days
Credit manager involved. Credit hold applied to new orders. Sales / account manager notified. Formal payment-demand letter. Extended payment plan requires manager sign-off.
Commercial & executive loop-in T+46 to T+60 days
CFO / director involved for strategic accounts. Joint commercial/credit decision on relationship vs recovery. Final demand with legal warning. Stop-supply active.
Legal / agency / write-off T+60 days +
Handover to collection agency or legal counsel. Bad-debt provision created. Write-off approved per authority matrix. Root cause recorded for future onboarding decisions.
AI is moving from buzzword to material advantage in AR. The highest-impact use cases are not replacing collectors — they are making collectors dramatically more effective by eliminating low-value work, surfacing the right accounts at the right moment, and predicting cash flow with precision.
Credit-risk AI
- ML credit scoring — predictive PD models using internal + bureau data, sector signals, macroeconomic overlays
- Dynamic limit adjustment — real-time limit recalculation based on payment velocity and order patterns
- Fraud & anomaly detection — unusual order patterns, new-entity risk flags
- News & event monitoring — NLP scanning for bankruptcy filings, M&A, sanctions hits
Collections AI
- Propensity-to-pay scoring — ranks accounts by likelihood and urgency; tells collectors where to focus
- Optimal contact timing — predicts best channel and time window per customer
- Cash-flow forecasting — predicts collections inflow from commitment data and behaviour history
- Automated escalation rules — AI triggers escalation before SLA breach based on risk signals
- Conversational AI — chatbot & IVR for low-risk account self-service payment
Dispute AI
- Auto-classification — NLP reads dispute email/portal text → assigns type and owner automatically
- Evidence matching — auto-retrieves relevant POD, PO, price file from connected systems
- Auto-resolution — for defined dispute types (duplicate, small pricing delta) → credit note or rejection without human touch
- Root-cause pattern detection — cluster analysis surfaces systemic upstream failures
| Automation use case | Technology | FTE saving | Cash / quality impact | Complexity |
|---|
| Dunning-letter generation & dispatch | RPA / ERP workflow | High | Faster dunning start | Low |
| Statement & invoice resend | RPA / document management | Medium | Removes payment excuse | Low |
| AI cash application & remittance matching | HighRadius, Billtrust, Esker | Very high | Reduces unapplied cash; frees dispute | Medium |
| Propensity-to-pay collector worklist | HighRadius, Sidetrade, ML model | Medium | DSO –3 to –8 days | Medium |
| Credit-limit recommendation engine | ML + bureau API + SAP | Medium | Reduces bad-debt risk | High |
| Dispute auto-classification & routing | NLP + workflow engine | Medium–High | Cycle time –40% | Medium |
| Auto-resolution of defined dispute types | AI + ERP credit-note API | High | Cost per dispute –60% | High |
| 30-day collections cash forecasting | ML + commitment data | Low | Treasury & working-capital optimisation | High |
Global Business Services and Shared Service Centre design for credit, collections and disputes requires deliberate decisions on centralisation vs local presence, work-allocation models, language and timezone coverage, and the balance between efficiency (scale) and effectiveness (customer relationships).
What to centralise
- Credit-risk assessment & limit setting (policy-driven, consistent)
- Automated dunning execution & statement dispatch
- Cash application & unapplied-cash resolution
- Dispute logging, classification & routing
- Collections analytics, KPI reporting & dashboards
- Credit-bureau feed management & monitoring alerts
- Write-off processing & bad-debt provisioning
What to keep local / decentralised
- Strategic-account relationship management & executive calls
- Local legal & regulatory compliance (jurisdiction-specific)
- Language-specific collector calls for complex negotiations
- Local agency and legal-counsel relationships
- Sales interface for commercial holds / disputes on key accounts
- Country-specific tax and VAT dispute handling
Hub-location criteria
- Language coverage for target markets
- Timezone alignment (near-shore preferred)
- Talent availability & cost arbitrage
- Legal right to collect cross-border
- Data-privacy compliance (GDPR etc.)
Work-allocation models
- Geographic: collector owns a region or country portfolio
- Customer tier: specialist team for strategic accounts
- Risk-based: high-risk bucket handled by senior collectors
- Hybrid: automated for the tail, human for high-value/risk
Continuous improvement
- Monthly root-cause review with upstream owners
- Lean / Six Sigma for dunning & dispute workflows
- Quarterly automation-opportunity scan
- Benchmarking vs APQC / IOFM standards
- Collector NPS as an internal CX metric
The GBS credit leader's mandate
In a high-performing GBS, the credit, collections and disputes function operates as a revenue-protection partner — not a cost centre. The best leaders combine deep process knowledge with commercial awareness, data fluency, and the ability to influence upstream teams (Sales, Ops, IT) to remove the root causes of credit loss and dispute volume at source. The function should be measured on cash impact, not just activity metrics.