I have been on both sides of this problem as a former SaaS CFO responsible for cash flow and forecasting, and now building Paraglide for teams that are tired of running accounts receivable out of an inbox.
If you are leading finance in B2B SaaS, you already know the uncomfortable truth: Recurring revenue does not automatically mean recurring cash.
Late payments, billing disputes, missing PO numbers, and slow accounts payable processes can quietly turn a “predictable” subscription business into a working-capital headache.
That’s exactly why dunning exists.
A modern dunning process is not just a sequence of reminder emails. It’s a structured accounts receivable workflow that connects billing, subscription management, ERP systems, and collections operations — while protecting customer relationships.
This guide explains how to automate dunning for SaaS companies, why B2B SaaS dunning is uniquely complex, and how AI agents are changing the way collections workflows run.
What is dunning?
Dunning is the structured accounts receivable process of following up on overdue invoices to secure payment. It typically begins as soon as an invoice passes its due date and continues until the invoice is paid or formally resolved. A dunning process usually includes staged reminders that become more direct over time. It also includes escalation steps when a customer remains unresponsive or when payment continues to be delayed.
How dunning works in SaaS environments
Dunning in SaaS works differently from many other industries because overdue invoices are rarely “just late” they’re usually stuck in a subscription, billing, or accounts payable workflow. In SaaS environments, dunnning sits inside quote-to-cash and order-to-cash:
Quote-to-cash: pricing → contracting → invoicing → payment
Order-to-cash: invoicing → collections → cash application → reconciliation
In a SaaS environment, dunning typically starts the moment an invoice passes its due date and triggers a staged follow-up sequence. Early-stage dunning is usually polite and process-driven, focused on confirming the invoice was received and sent to the right person. As invoices age, the workflow becomes more structured, with tighter follow-up cadences, escalation to finance leadership, and sometimes account-level actions depending on the customer segment.
Unique challenges in B2B SaaS dunning
B2B SaaS dunning looks simple until you run it at scale. The complexity comes from payment methods, subscription models, and the fact that the real bottleneck is often inbound, not outbound.
Credit card dunning (high automation, high failure volume)
Credit card dunning is common in SMB and product-led SaaS.
It’s automatable because billing platforms can:
Retry failed payments
Send automated reminders
Pause service automatically
But the root cause of card failures is often data, not willingness to pay:
Expired card
Card replaced
Wrong billing contact
Email no longer active
Missing billing address fields
Credit card dunning often turns into a contact-data cleanup exercise.
Direct debit dunning (lower volume, compliance-heavy)
Direct debit (SEPA, ACH, Bacs) tends to have fewer failures.
But when it fails, exceptions can be harder:
Mandate not signed
Mandate expired
Bank rejection codes
Account details changed
Direct debit dunning is less about chasing and more about procedural correctness.
Bank wire dunning (enterprise reality)
For mid-market and enterprise SaaS, bank wire dominates.
This is where most CFOs feel the pain — because the customer controls payment through internal workflow:
Invoice approval chain
PO matching
Vendor onboarding
Monthly payment runs
There is no “retry payment.” There is only: follow up, unblock, escalate.
Why ERP dunning letters aren’t enough
Most ERP and billing systems can generate dunning letters.
But ERP dunning is outbound-only. It does not manage the operational mess that happens next.
In B2B SaaS, the biggest delays come from the finance inbox:
Replies to reminders
Dispute explanations
missing PO requests
“Please resend invoice”
Vendor onboarding forms
Payment confirmations
This is why many SaaS companies even with modern ERP and billing stacks — still run collections through:
billing@company.com
ar@company.com
finance@company.com
And shared inboxes are where dunning processes go to die.
PO numbers: the silent cause of SaaS invoice delays
One of the most common reasons enterprise SaaS invoices go overdue is simple:
The invoice needs a PO number.
No amount of automated dunning letters will fix that.
The resolution requires:
Identifying the missing PO
Requesting it from the right person
Updating ERP/billing
Reissuing the invoice if needed
Confirming it’s approved
This is a workflow problem — not a reminder problem.
How to automate the dunning process for SaaS
Automation does not simply mean sending reminders. True dunning automation means creating a workflow that reliably moves invoices from overdue → resolved → paid, with minimal manual work.
At a practical level, an automated dunning process needs:
Clear stages (early, mid, late overdue)
Customer segmentation (value, risk, tier)
Integration across ERP + billing + subscription systems
Measurable outcomes (DSO, collection rate, disputes)
Professional services firms like PwC and EY consistently emphasize that working capital performance depends on disciplined AR processes, data quality, and governance — not heroics.
How AI agents can automate dunning workflows
This is where things get interesting.
Traditional automation handles outbound sequences.
AI agents handle the messy middle: the inbox, the replies, the exceptions, and the ERP updates.
That matters because in real-world SaaS collections, the majority of AR time is not spent writing reminders. It’s spent handling everything that happens after the reminder.
What AI agents do in dunning
An AI agent in accounts receivable can:
Read inbound emails to billing@ / ar@
Classify intent (dispute, PO, resend invoice, promise-to-pay, etc.)
Extract structured data (invoice number, PO number, payment date)
Draft replies that match policy and tone
Flag and update bad contact data
Update ERP / billing systems
Route exceptions to humans with context and suggested actions
Gartner has consistently highlighted the shift toward AI-enabled finance operations, including intelligent automation and autonomous workflows in AR. The direction is clear: finance teams are moving from manual triage to AI-supported operations.
Traditional AR automation vs AI-driven dunning workflows
Comparison of legacy reminder-centric automation against modern AI agent-driven workflows that handle inbound exceptions, dispute categorisation, and workflow updates.
Area | Traditional dunning automation | AI agent-driven dunning workflow |
Outbound reminders | Scheduled emails | Scheduled + personalized reminders |
Inbound replies | Manual handling in inbox | Auto-triage + suggested/automated responses |
Disputes | Logged manually | Detected + categorized + routed automatically |
PO number capture | Manual back-and-forth | Extract PO from email + update ERP |
Promise-to-pay tracking | Spreadsheet | Automatically tracked + follow-up triggered |
ERP updates | Manual | Automated actions or guided updates |
Reporting | Often incomplete | End-to-end workflow visibility |
The two workflows AI agents unlock
From my CFO seat, I think about dunning as two workflows running in parallel:
Outbound dunning letters (the predictable part)
Inbound billing and collections support (the real work)
AI agents can automate both.
Common inbound AR emails and how an AI agent handles them
Examples of typical customer replies that stall cash collection and how AI classification, extraction, and workflow actions can resolve them.
Customer email / reply | What it means | What an AI agent can do |
“Can you resend the invoice?” | Invoice lost or wrong contact | Find invoice, resend, update contact |
“We need a PO number” | AP cannot pay without PO | Ask for PO, extract it, update ERP |
“This amount is wrong” | Billing dispute | Detect dispute, route to billing owner |
“Paid yesterday” | Payment confirmation | Check ERP/bank sync, update status |
“Send vendor onboarding documents” | Vendor setup blocker | Respond with required docs + track completion |
“We pay monthly on the 30th” | Payment run schedule | Log promise-to-pay date + follow up automatically |
“Please contact AP” | Wrong stakeholder | Identify AP contact and reroute reminders |
AI agents as collection agents and billing support agents
In SaaS, the collections team is often doing billing support.
That’s not because they want to — it’s because billing blockers prevent payment.
AI agents can function as:
a collection agent (follow-up, escalation, promise-to-pay)
a billing support agent (resends, PO capture, contact updates)
And crucially: they can do it consistently, 24/7, without inbox chaos.
Dunning considerations across different countries
If you operate internationally, dunning is not one-size-fits-all.
Collections regulations and cultural norms differ significantly. Even within B2B, what is considered “standard escalation” varies.
United Kingdom
In the UK, the Late Payment of Commercial Debts Act allows statutory interest and fixed compensation on overdue invoices.
In practice, many SaaS companies avoid enforcing it aggressively to protect relationships. UK dunning tends to stay professional and measured until later stages.
United States
In the US, B2B collections is less centralized than consumer collections, but contractual terms typically govern interest and penalties.
Enterprise SaaS collections often rely on internal escalation before involving external collection agencies.
Nordics
In the Nordics, formalized collection steps are common and culturally accepted.
It’s more normal to move invoices to structured collection processes earlier, especially in Sweden and Finland. Reminder fees and escalation steps can be more standardized.
France
France has stricter invoicing and compliance requirements. Administrative correctness matters more:
legal entity
invoice format
mandatory fields
PO accuracy
Small errors can create disputes and delay payment significantly.
When collection agencies make sense in SaaS
In SaaS, collection agencies are usually a last resort, but they can make sense when:
The invoice is significantly overdue
The customer is unresponsive
Internal escalation has failed
The relationship is no longer strategic
The key is having a dunning workflow that clearly defines when escalation moves outside your team.
Benefits of automating dunning
When CFOs ask me “does this actually matter?”, my answer is always:
Yes, because dunning is one of the few levers in SaaS that improves cash flow without changing your product, pricing, or sales strategy.
Reduced DSO
Even small improvements in Days Sales Outstanding can unlock meaningful working capital.
A 5–10 day DSO improvement can fund hiring, reduce reliance on credit facilities, or extend runway.
Lower bad debt
Automation helps you identify disputes, risk accounts, and chronic late payers earlier.
That reduces write-offs and prevents invoices from silently aging into bad debt.
Time savings and scalability
Manual dunning does not scale.
Automation allows a small AR team to manage:
More invoices
More customers
More complexity
…without growing headcount at the same pace.
Better customer relationships
This is the part many finance teams miss.
Good automation reduces:
Duplicate chasing
Inconsistent messaging
Wrong escalation timing
Frustration caused by billing errors
In SaaS, where retention matters, dunning should feel like a professional process — not a scramble.
How Paraglide automates dunning for Saas companies with AI agents
Paraglide is designed for SaaS finance teams that already have an ERP and billing system, but still run collections through a shared inbox and manual workflows.
Instead of replacing your ERP, Paraglide acts as the workflow and AI agent layer across:
accounts receivable
billing support
dunning replies
exception handling
ERP updates
Where Paraglide fits in the dunning workflow
This table shows how Paraglide sits on top of your existing ERP and billing tools to automate the real dunning work.
Dunning workflow step | What usually happens today | How Paraglide helps |
Reminder emails | Automated in ERP/billing | Works alongside, adds workflow + intelligence |
Replies to reminders | Manual inbox triage | AI agent categorizes + drafts responses |
Missing PO numbers | Manual back-and-forth | Extracts PO + updates ERP/billing |
Disputes | Logged inconsistently | Detects dispute + routes with context |
Resend invoice requests | Manual | Auto-resend + correct recipient |
Promise-to-pay | Tracked in spreadsheets | Tracks dates + triggers follow-ups |
ERP updates | Manual | Suggests or automates updates safely |
Reporting | Fragmented | End-to-end visibility across workflow |
If you want to see what this looks like in practice, the best next step is simple: map your current billing@ inbox workflows and compare them to what an AI agent can take over.
Common mistakes finance teams should avoid in dunning
Over the years, the same mistakes show up again and again in SaaS dunning.
Automating reminders without automating exceptions
If you only automate outbound messages, your team still gets buried in replies.
Treating all customers the same
Enterprise customers require a different workflow than SMB subscriptions.
Ignoring the inbox
Most dunning delays come from inbound queries and replies, not from the reminder schedule.
Poor integration between ERP and billing
When invoice status is inconsistent, automation creates confusion and mistrust.
Escalating too aggressively
Over-escalation can damage renewal relationships, especially when the issue is administrative.
Not measuring the process
If you don’t track dispute causes and stage-level performance, you can’t improve.