DSO (Days Sales Outstanding) is the credit manager's flagship metric. It measures, in days, the average delay between invoice issuance and actual collection. Across Europe, the average DSO hovers around 58 days according to AFDCC data (for reference only, 2024-2025 benchmarks). For an SMB invoicing $10M per year, every day of DSO represents roughly $27,400 tied up in receivables. Gaining 15 days frees more than $400,000 of cash. This article details an operational method to achieve a 30% DSO reduction in 90 days, by orchestrating a voice AI agent across the pre-due-date phase that is too often neglected.
The 4 DSO levers every credit manager must master
DSO reduction rests on four distinct phases of the receivable lifecycle, each calling for specific tactics. Companies that hit a DSO plateau often confuse these phases or concentrate their effort on the post-due-date phase alone — the most visible, but also the least effective.
Lever 1 — Pre-due-date (D-7 to D-1)
This is the most under-exploited yet most profitable lever. It means confirming, before the due date, that the invoice is received, validated, and that any latent dispute is detected before it becomes an excuse for delay. A confirmation call at D-7, followed by a D-3 reminder, converts "last-minute" payments into on-time payments. Average observed gain: 5 to 8 days on DSO.
Lever 2 — Due date (D+0 to D+5)
The week immediately following the due date is critical. A customer running 3 days late "by oversight" risks, if not contacted quickly, sliding to 30 days. A first voice contact as early as D+2 or D+3 with a neutral tone and reminder of contractual details recovers most short delays.
Lever 3 — Post-due-date (D+5 to D+45)
Active amicable collection phase. Multi-channel (SMS, email, voice call), tighter cadence, payment plan proposal if needed. Most teams concentrate 80% of their energy on this phase, which is too late. An optimized DSO must minimize the volume entering this phase.
Lever 4 — Legal (D+45 and beyond)
Escalation to judicial debt collection or receivables assignment. The credit manager arbitrates here: which receivables are worth the cost of a procedure? The operational goal is that less than 3% of the portfolio reaches this phase.
Pre-due-date playbook: the D-7 / D-3 / D-0 sequence
Here is the operational sequence that alone can win 10 to 15 DSO days on most B2B portfolios. It rests on three targeted voice touches, orchestrated by the voice AI agent across all customers whose amount exceeds a defined threshold (typically $2,000 per invoice).
D-7 — Confirmation call
The AI agent calls the debtor's accounting contact and asks four short questions: has the invoice been received? Has it been recorded in the accounting system? Is the delivery note matched? Is there a discrepancy or dispute? This call does not ask for payment: it qualifies. If a dispute is detected, the invoice is immediately routed to the order management team, which has 6 days to resolve it before due date.
D-3 — Friendly reminder
Short call, 45 seconds, positive tone. "Hello, your invoice no. X is due on [date]. Can we confirm it will be handled on time?" The agent records the Promise to Pay (PTP), which becomes the reference metric for follow-up.
D-0 — Due-date courtesy
Only for at-risk customers (score C or D — see next section). Call on the day itself, positioned as a courtesy. Recalls the day's due date, offers an immediate wire transfer or SMS payment link.
Customer scoring: differentiated orchestration
Applying the same cadence to all customers is a classic mistake. A large public account pays slowly but surely. A startup between funding rounds can flip to distress in 60 days. A historically stable customer does not deserve the same pressure as a new customer with high credit risk.
Scoring combines two axes: credit risk (credit bureau data, financial statements) and observed payment behavior (historical average delay, dispute rate, recurring lateness, amounts).
| Score | Profile | Pre-due-date cadence | Post-due-date cadence |
|---|---|---|---|
| A | Reliable payer, low risk | D-3 email only | D+5 call if needed |
| B | Occasional delays | AI call at D-3 | D+2 call, D+7 email |
| C | Recurring delays | Calls at D-7 + D-3 | Tight sequence D+1, D+5, D+10 |
| D | High risk, watchlist | D-7 + D-3 + D-0 | Human escalation from D+3 |
This differentiation frees human agents from the 70% of cases that self-regulate via the AI pre-due-date sequence, and concentrates their time on scores C and D.
KPIs to steer continuously
A playbook without a dashboard is wishful thinking. Here are the four KPIs every credit manager must track weekly, ideally on a dashboard shared with the CFO.
Overall DSO and segment DSO
Overall DSO masks sector gaps. It must be broken down by sales channel, by customer type (large account / SMB / individual), by geography. The overall reduction is the consequence of segmented optimizations.
RPC — Right Party Contact
Percentage of calls where the right person (authorized accounting contact) was reached. An RPC below 40% signals a CRM data quality problem. The AI agent typically moves RPC from 35% to over 60% thanks to multiplied calling slots.
PTP — Promise to Pay and promise-kept rate
Every call resulting in a PTP must be logged. The promise-kept rate (actual payment by the promised date, with a +3-day tolerance) is a debtor reliability indicator and a predictor of future DSO.
Pre-due-date dispute detection rate
KPI specific to AI orchestration. A dispute detected at D-7 is resolved on average in 4 days. The same dispute detected at D+20 takes 18 days to resolve and generates 14 additional DSO days. Target a pre-due-date detection rate above 80% of total disputes.
Case study: $50M revenue industrial mid-market firm (illustration)
For illustrative purposes, take a European industrial mid-market firm with $50M annual revenue, about 1,800 invoices issued per month, an average ticket of $2,800 and a starting DSO of 64 days. The assumptions below are indicative and must be adjusted to each real context.
Initial situation: 2 in-house credit managers, 1 collections assistant, email reminders at D+15 and D+30, sporadic human calls at D+45. Effective contact rate: 28%. Receivables >60 days: 22% of portfolio.
AI agent deployment: pre-due-date orchestration D-7 + D-3 on 100% of invoices >$2,000, A/B/C/D scoring applied to the full portfolio, automated human escalation on C-D scores and detected disputes. Deployment in 7 days, calibration in 30 days.
Results observed at 90 days (indicative order of magnitude):
- DSO moved from 64 to 45 days (-30%)
- Freed cash: roughly $2.6M
- Receivables >60 days: 22% → 8%
- Pre-due-date dispute detection rate: 83%
- Human time on mass reminders: -65%, reallocated to complex cases and large-account negotiations
"Our DSO dropped below 45 days for the first time in 8 years. The real change is not the decline itself, it's predictability: we now know in the first week which invoices will be problematic." — CFO, regional industrial group
Building your monitoring dashboard
An effective credit management dashboard fits on a single screen and shows, in real time:
- Overall DSO with a 12-month curve and target
- DSO breakdown by age bucket (0-30, 31-60, 61-90, 90+)
- Top 10 at-risk receivables (amount × probability of non-payment)
- Promise-kept rate over the last 30 days
- Alerts: detected unprocessed disputes, broken promises, pending escalations
- AI-freed capacity (calls placed, qualified cases)
This dashboard must be consultable by the CFO autonomously, without credit manager intervention, to feed weekly cash committees.
Classic mistakes to avoid
- Trying to automate everything at once. Start with the most voluminous and standardized segment. Expand later.
- Eliminating human intervention. AI handles volume, humans handle complexity. The two are inseparable.
- Neglecting CRM quality. A wrong phone number is an RPC collapse. Clean data before deployment.
- Forgetting compliance. Legal hours (8 a.m. to 9 p.m. weekdays per US TCPA; 8 a.m. to 8 p.m. weekdays in France), systematic opt-out, GDPR traceability. See our GDPR and voice AI guide.
- Not measuring. Without a dashboard, impact cannot be demonstrated to leadership, and the project loses momentum.
FAQ
What is DSO and how is it calculated?
DSO (Days Sales Outstanding) measures the average customer payment delay. Formula: (trade receivables / total sales incl. tax) × number of days in the period. A DSO of 58 days means it takes on average 58 days to collect an issued invoice.
Is a 30% DSO reduction realistic?
Yes, on a high starting DSO (>65 days) and with well-designed pre-due-date orchestration. Gains come from three sources: proactive D-7 confirmation, scoring-differentiated reminders, and immediate handling of disputes detected vocally.
Which KPIs should I prioritize?
Four critical KPIs: DSO overall and by segment, RPC (Right Party Contact), PTP (Promise to Pay), 7-day promise-kept rate. Add the pre-due-date dispute detection rate to measure preventive impact.
Does customer scoring replace the credit manager's analysis?
No. Scoring automates behavioral segmentation (average delays, repeat offenders, amounts). The credit manager remains the decision-maker on sensitive cases, limit reviews and legal escalations.
How long before the first gains appear?
Typically 60 to 90 days. The first month calibrates scripts and scoring. The second month shows a 5-10% DSO reduction. At 90 days, a 20-30% gain becomes visible if pre-due-date orchestration is systematized.
Going further
This playbook works provided it is calibrated to your real portfolio, sector and cash seasonality. A strategic audit identifies, in 30 minutes, the two or three priority levers in your context and the realistic DSO gain expected over 90 days.
Further reading: Debt collection: voice AI doubles callback rates · Deploy a voice bot invoice collection in 7 days · Debt Collection Industry · About Vocalis AI.