You're Chasing the Wrong Customers

If you have been following this Net Working Capital series, last week we fixed the DSO gauge with the countback method so it measured collection speed instead of revenue timing. If you landed here from a search, start with Article 1 for why working capital is a system and not a single metric. Either way, the question this week is short and uncomfortable: open your aging report, look at the five customers holding most of your overdue cash, and ask whether you can actually say why each one is late.

The standard playbook for overdue receivables is diligence. Send reminders earlier. Send them more often. Automate the dunning cycle. Train your AR clerks better. Escalate faster.

Read any best-practice guide on accounts receivable and that's the message. Get diligent and you'll get paid on time.

It doesn't work.

Not for lack of effort. The diligence framework assumes overdue is one problem, solvable by more communication. It isn't. The customer who has paid at 60 days every quarter for two years isn't waiting for a more urgent email. The customer with a stuck credit memo isn't waiting for a reminder. They're waiting for the credit memo. The customer whose master data shows the wrong terms isn't even late. And the customer who can't pay isn't going to be persuaded by anything you write.

In Atradius' 2025 North America survey, overdue B2B credit sales sat in the low 40s: 43% in the US, 44% in Canada, 41% in Mexico. When a major survey puts overdue around the low 40s, "chase harder" stops being a useful portfolio answer. What needs fixing is the category of action you take on each name.

Open your AR aging report. Sort by overdue amount. Take your top 5. Now ask one question: for each of them, do you know why they're late?

Not "they always pay late." That's a habit, not a reason. Why. Which step broke. Which conversation never happened. Which line in the master data is wrong. Which contract clause was negotiated outside the ERP and never made it in.

If you can't answer that for the 5 customers holding most of your overdue cash, the aging report is doing nothing for you. It's a list of symptoms with no diagnosis attached. And every action you take based on it lands on the wrong intervention for 3 of the 4 categories of late.

This article is about why the standard advice fails by design, not by execution. And what to do instead.

The chase-harder fallacy

In every AR portfolio I've worked with in twelve years of manufacturing finance, the same monthly pattern shows up. Aging report goes out. Finance lead reviews top overdue customers. AR team sends a wave of reminder emails. A week later, a few small invoices clear, the top customers stay on the report. Stronger reminders go out. A week later, same picture. Calls go out. Two more small invoices clear. The top customers stay.

Next month's aging report shows the same names at the top.

The honest read on this cycle isn't "we need to be more diligent." It's that diligence isn't what's missing. The reminder is the wrong intervention for the customer it's being sent to.

To make this concrete: the customer who's been paying at 60 days on Net 30 for two years has internalized that the real payment date is day 60. Your reminder at day 35 is noise. Your stronger reminder at day 45 is noise. Your call at day 55 might generate a polite acknowledgment, and they'll still pay at day 60. They are not confused. They've made a decision. No email is going to undo it.

The other three patterns at the top of an aging report fail diligence for different structural reasons. A customer with a stuck credit memo is waiting for the credit memo, not the reminder. A customer whose ERP terms differ from the signed contract isn't actually late, so the reminder is incorrect. A customer heading into financial distress can't pay regardless of the wording, and sharper language risks creating problems rather than solving them.

Four customers. Four different stories. One reminder template applied to all four. The template is not the leading intervention for any of them.

The diligence framework is designed for one category of late: the customer who pays after one or two reminders, with no underlying pattern, who simply needs a polite nudge. That category exists. It's most of the long tail of small overdue balances. For that population, the standard playbook works.

But the long tail isn't where the cash is. The cash is in the 10 to 15 percent of customers at the top of the aging report. And for those customers, every category except the long-tail-like one fails the diligence test for a different structural reason.

Once you see this, you can't unsee it. The standard playbook isn't wrong because AR teams execute it poorly. It's wrong because it answers the wrong question for the customers who matter most.

Pareto on AR: where the cash actually sits

The first cut is Pareto on overdue balance. Sort descending. Look at the top.

In most AR portfolios I've worked with, the curve is brutal: a small number of customers hold most of the overdue cash. Your exact split might be 60/10, 70/15, or 80/20. The ratio is less important than the shape. Take a $3.5M total AR with $1.2M overdue across 200 customers. The top 8 customers hold $720K of that, about 60 percent. The remaining 192 customers split $480K, averaging $2,500 each.

That alone reframes the work. You don't have a 200-customer problem. You have an 8-customer problem plus a long tail of small balances.

Amount is the right first cut because cash impact is what you're managing. Age and credit risk come next: a $30K invoice at 180 days can need escalating before a $90K one that's three days past due. Sort by money, then let age and risk tell you what's actually deteriorating.

So far this is textbook Pareto. Sort by impact, focus on the top. Useful, incomplete.

The completion comes from looking at the top 8 a second time and asking a different question. Not "how much do they owe." That you already know. The second question is "why are they late."

That second question is where the framework moves from priority list to action list.

The eight-customer example below is a composite. The customers are invented; the shape is one I've mapped repeatedly across twelve years of manufacturing finance, and it's the same first move I make when I map a close cycle on GoFast.Finance: sort by impact, then ask what's actually causing it, one name at a time.

Four kinds of late

For the top customers on the overdue list, there are usually four kinds of late. Each has a different cause. Each has a different owner. Each needs a different intervention. The reminder template works for none of them.

• Behavioral late. The customer has a consistent payment pattern, and that pattern is not the contractual terms. Customer A is on Net 30 and has paid at 58 to 62 days for eight consecutive quarters. The customer has unilaterally renegotiated payment terms, and nobody on the seller's side has acknowledged it. Why diligence fails: the customer is not confused. The customer has decided. Sending the reminder harder, more often, from a more senior signatory, none of that changes the decision. The actual fix is a commercial conversation with Sales. Acknowledge the customer has effectively moved to Net 60. Decide whether to renegotiate formally and price the credit, or enforce Net 30 and accept the relationship risk. The current state, informal Net 60 with no acknowledgment, is the worst of the three options.

• Contractual late. The customer is paying on time. The ERP master data says they're late. Terms agreed with Sales in last year's renewal were Net 60. Terms in the master data are Net 30. The customer pays around day 58 and the aging report flags it as 28 days overdue. Why diligence fails: there's no diligence problem to solve. The customer is paying on the terms they agreed to. The aging report is wrong about the deadline. Sending reminders for "late" payment on invoices that are actually on time damages the relationship and tells the customer your Finance function isn't talking to your Sales function. The fix is master data correction with Sales confirming the actual contract terms.

• Process late. The customer wants to pay but something specific is blocking the invoice. A credit memo was promised three months ago and never issued. The PO references the wrong line item. The buyer left and the new buyer hasn't been onboarded. The address changed and the invoice keeps going to the old one. Sometimes the cash already arrived and sits unapplied because the remittance didn't match the invoice numbers, so a paid customer still reads as overdue while the money is on your own books. Why diligence fails: the customer isn't refusing to pay. They're waiting for a specific thing your organization promised and didn't deliver. The reminder doesn't deliver the missing credit memo. The reminder doesn't update the contact. The fix is root cause on the specific blocker, often a 30-minute call between AR, Sales operations, and the customer's AP team. Once the blocker clears, the invoice pays.

• Risk late. The customer has a liquidity problem. They aren't paying because they can't. Missed payments, restructuring talks, deteriorating financial signals. This is the only category where the reminder makes any sense at all, because at least it documents that you've requested payment. But the reminder isn't going to produce the payment. The customer doesn't have the cash. And if the language gets sharper, inaccurate, or inconsistent with what's actually owed, you risk creating legal or commercial problems rather than solving the financial one. The fix is credit review, an order hold or credit-limit restriction where it's contractually and commercially available, payment plan negotiation if recovery is possible, provisioning and write-down planning if it isn't.

Same line in the aging report. Four completely different stories. Four different owners. Behavioral and contractual both route through Sales (one to renegotiate terms, the other to align master data). Process routes through AR plus whichever operations function owns the blocker. Risk routes through Credit Management.

The reminder email is the right intervention in exactly one case: the long tail of small overdue balances paying after one or two reminders. None of your top 8 is in that category.

How to classify in practice. For each top overdue customer, check five fields before assigning the category. Contractual terms (what does the master data say versus what does the signed contract say). Last four-quarter payment pattern (consistent timing or one-off). Open disputes, deductions, or pending credit memos. Unapplied cash sitting on the account. Current credit-risk signals from public sources or your own monitoring. If the terms in the ERP don't match the contract, it's contractual. If payment behavior is stable but later than terms across multiple cycles, it's behavioral. If a specific document, contact, PO, or process step is blocking payment, it's process. If payment capacity is deteriorating, it's risk. The check takes about three minutes per customer once you have the data in front of you.

The worked example, and the conversation with Sales

Back to the $3.5M total AR, $1.2M overdue, top 8 = $720K. Here are the 8 customers, classified.

• Customer A: $180K, behavioral. Eight quarters of paying between 58 and 62 days on Net 30. Sales conversation about formal terms or enforcement.

• Customer B: $145K, process. Disputed invoice from January, credit memo promised, never issued. Internal fix, one call to close.

• Customer C: $115K, process. AP contact rotated in February. Invoice keeps going to a person who no longer works there. Update master data, resend.

• Customer D: $90K, contractual. Sales agreed Net 60 in last year's renewal. ERP still says Net 30. Master data correction, retroactive aging adjustment.

• Customer E: $75K, behavioral. Smaller pattern of paying 10 days late, consistently. Sales conversation, less acute version of Customer A.

• Customer F: $50K, process. PO mismatch on three invoices. Sales operations reissues.

• Customer G: $40K, risk. Two missed payments, news of restructuring talks. Credit review.

• Customer H: $25K, process. Address change six months ago, invoices still going to the old building. Update master data, resend.

Eight customers, $720K of overdue cash, eight specific actions. Two Sales conversations (A, E). Four process fixes (B, C, F, H). One master data alignment (D). One credit review (G).

The default response to this list is a stronger reminder to all eight and a follow-up call next week. That treats the work as a calling list. The four-category split treats it as a triage list with named owners. The first is busy. The second moves cash.

Now the conversation that changes things.

You walk into the next Sales pipeline meeting. You bring the classified list. You start with Customer A. You say:

"Customer A has been paying at 60 instead of 30 for two years. We're carrying about $180K of incremental AR with this customer at all times because of the lag. At our 10% cost of capital, that's roughly $18K a year, $4.5K a quarter, of financing cost that's nowhere in this customer's margin. Do we want to renegotiate to Net 60 and price for it, or enforce Net 30?"

The Sales director's response is predictable. There are usually three forms.

• Pushback 1: "They're a strategic customer." Counter: strategic customers can be repriced. The question isn't whether to keep them. It's whether the current terms reflect the credit you're extending. Strategic doesn't mean unprofitable in ways nobody sees.

• Pushback 2: "They'll go to a competitor." Counter: if they would leave over a small price uplift on Net 60 that formalizes what they're already doing, the relationship was never as strong as you thought. And if the competitor offers the same terms without the uplift, the competitor is making the same mistake we are. Either way, the unpriced credit is a problem we can name now or absorb forever.

• Pushback 3: "We agreed this informally, that's how we keep them." Counter: informally agreed Net 60 isn't a relationship asset. It's a discount nobody at our company is tracking. The arithmetic: 30 extra days of credit at 10% annual cost is about 0.82% of invoice value. A 1% price uplift on Net 60 covers the financing cost with a small margin for admin and default risk. A formal Net 60 with a 1% uplift is better for both sides, because at least we know what we're trading.

What's happening in this exchange: Finance has reclassified the issue. The framing has moved from "AR collections is asking us to chase a strategic customer" to "Finance is asking Sales to acknowledge a commercial concession that has cost the company money for two years." That's a different conversation. Sales can have it. Collections can't.

The deeper unlock: in most comp plans, Sales is rewarded on bookings, not on payment timing. Once the sale is booked, there's no incentive to enforce terms. The org design produces exactly the result you're looking at on the aging report. Naming this in Finance leadership is part of what unlocks the conversation. It isn't a personal failure of any Sales director. It's the predictable outcome of how the comp plan is wired.

Multiply this across the portfolio. If you have 5 to 10 behavioral late payers averaging $100K each, you're carrying $500K to $1M of unpriced credit at all times. At a 10% cost of capital, that's $50K to $100K a year of finance cost nobody is tracking. Most companies have never put a number on this. Most companies have never had the conversation.

The second story: a process blocker that keeps coming back

The behavioral case is the one that makes the slide. The process case is the one that frees up cash fastest, and it fails diligence for a completely different reason. Worth walking through one in full, because the mechanics, the owner, and the fix all differ from the Sales conversation above.

Customer B sits at $145K, more than 90 days, flagged for escalation three months running. On a pure aging read, this is the customer you escalate hardest. Senior signatory, formal demand, the works. Every one of those moves is wrong, because Customer B is right.

What a 30-minute call surfaces, with AR, Sales operations, and the customer's AP team on the line: back in January, a pricing dispute was settled in the customer's favor. A credit memo was agreed to clear the difference. The sales rep approved it verbally on the call. It never got entered, because the credit-memo workflow needs a second approval that nobody owned, so it sat in a queue. The customer is holding the full invoice until the memo lands, exactly as any disciplined AP team would. From their side, your company agreed to something and didn't deliver it.

Issue the credit memo. The invoice clears inside a week. No reminder produced that. No escalation produced that. A reminder would have told a customer who is behaving correctly that your house isn't in order, which is the message Customer C and Customer H are already getting while their invoices route to people who left and buildings the company moved out of.

Here's the part that matters more than the $145K. Fixing Customer B's invoice is a correction. The credit memo gets issued, the cash arrives, the line drops off the report. The blocker that caused it is still there. Next quarter a different customer will settle a different dispute, a different memo will sit in the same unowned queue, and a new six-figure line will appear at the top of the aging report wearing the same disguise as a collections problem.

The countermeasure is to give the credit-memo approval step an owner and a deadline, so a settled dispute turns into an issued memo without anyone chasing it. That fix lives in process design, not in collections. It's the difference between bailing the boat and finding the hole. AR can bail. Only the process owner can patch.

This is why the category label matters and "they're just slow" doesn't. Behavioral late needs a recurring commercial decision and lives with Sales. Process late needs a one-time clearance plus a structural fix and lives with whoever owns the broken step. Same dollar figure on the same report, opposite work.

Methodology, and the monthly cadence that actually works

The Lean tool is Pareto. The 80/20 rule applied to AR balance gives you a priority list. Useful, but it stops one step short.

The completion is the classification step. Sort by impact first. Then classify by cause. The classification turns a priority list into an action list with a specific owner per item. Two steps, not one. The first step (Pareto) tells you where to focus. The second step (classify) tells you what to do.

Most AR tooling optimizes inside step one. Dunning workflows focus on who gets reminded, when, and with what wording. Better platforms go further: risk-based segmentation, dispute and deduction tracking, reason codes, AI-prioritized worklists. Real improvements over a manual aging spreadsheet. But a reason code is not an owner. Without someone accountable for the action, a deadline, and a tracked cash result, even a smartly prioritized worklist is a faster version of the same chasing loop. The cadence gets cleaner. The customers who hold most of your overdue cash still don't pay, because the intervention they need isn't a message at all.

This is the standard form of a Lean tool failure: applying optimization to a step that doesn't answer the underlying question. Lean asks "is this activity adding value" before "can we do it faster." Step one alone, no matter how well optimized, isn't where the value gets created on the top of the aging report.

The same tool appears in CC-AA01, Month-End Close Optimization, applied to close tasks ranked by cycle time impact. Same shape, different data. There the question is which tasks cost the most hours and what causes the cost. Here it's which customers hold the most overdue cash and what causes the lateness. In both cases Pareto is the first cut. The classification is what makes the second cut actionable.

To put this in practical terms, here's a monthly cadence that works.

Once a month, 30 minutes. Four people in the room. AR Lead, who owns the data and the classification. Credit or Finance Manager, who owns the decisions. Sales Lead, who owns the commercial conversations. Operations Lead, who owns the process blockers.

The agenda is three questions:

1. Of last month's top 10 actions, which were completed and what did each one move? Cash impact tracked per action, not in aggregate.

2. This month's top 10 overdue customers, classified into the four categories. Who owns each next step.

3. Any escalations needed. Customer A in renegotiation deadlock. Customer G needs a credit committee call this quarter.

Keep DSO as the portfolio gauge, but it isn't the meeting's metric. DSO moves slowly and confounds collection behavior with revenue timing (which Article 2 of this series covered). What the meeting tracks is actions completed and cash moved per action. That's the leading indicator. DSO is the lagging one.

Two compound effects show up after three to six monthly cycles. First, the classification gets faster, because the same names appear repeatedly until the underlying issue is fixed. Second, the long tail starts to shrink, because behavioral late payers either renegotiate to longer terms with pricing, or come back to contractual compliance once the conversation has happened.

Then run Pareto a second time, on the causes instead of the customers. If 40% of your top overdue dollars are process-blocked and half of those are stuck credit memos, you don't have a collections problem. You have a credit-memo workflow problem, and the fix lives upstream, the same countermeasure logic that cleared Customer B. That's the move from faster triage to a system that stops generating the same overdue lines.

By month six, the aging report stops being a list of symptoms and starts being an artifact of a system that's working. The names that remain are the ones you're actively working, with specific owners and specific dates. The rest paid, because the rest could pay.

Common pushback

When this framework gets proposed inside a finance function, four objections come up reliably. Each is reasonable on its face. None survives contact with the arithmetic.

• "This is more work than just sending reminders." The classification is about three minutes per customer for your top 10. Thirty minutes, once a month. The reminder loop runs weekly and produces nothing on the customers who hold most of the cash. You aren't adding work. You're stopping work that doesn't move anything and redirecting a fraction of it to work that does.

• "Our AR platform already prioritizes by balance and risk." That's step one, optimized well, and the good platforms go further than sorting: dispute tracking, reason codes, risk segmentation. The gap usually isn't the software. It's the operating model. A reason code doesn't move cash until it becomes a named owner, a next action, and a deadline. A risk-ranked worklist still routes every top customer to the collections team, and three of the four categories can't be closed by collections. Classification without ownership is just better reporting.

• "Isn't this Credit Control's job?" Credit Control owns one of the four categories. Behavioral routes to Sales, contractual routes to Sales plus master data, process routes to whoever owns the broken step. The reason the top of the aging report never clears is precisely that the work doesn't belong to the team holding the report. Putting it all on Credit Control is how it stays stuck.

• "Our portfolio is too concentrated for Pareto to mean anything." Concentration makes the classification more valuable, not less. If three customers are 70% of your overdue cash, the cost of sending all three the wrong intervention is higher, not lower. With a concentrated book you can skip the sorting step almost entirely and go straight to classifying the handful of names that matter.

FAQ

• How do I get Sales to actually show up to a monthly AR review? Lead with the cost-of-capital number on one customer, not with a request for their time. "We're spending about $18K a year financing Customer A's habit, and I need fifteen minutes of your read on whether we reprice or enforce." Sales shows up for revenue and margin conversations. Framed as collections, the meeting is an imposition. Framed as a margin leak with their name on the relationship, it's their problem too. Start with one customer, show one number move, and the second meeting is easier to fill.

• What if Sales refuses to have the terms conversation at all? You don't need their permission to put a number on it. Finance can quantify the unpriced credit across every behavioral late payer, book the financing cost against the relevant commercial owner in your internal reporting, and bring the stacked total to leadership. The conversation that Sales won't have at the customer level becomes a portfolio-level decision at the leadership level. Naming the comp-plan incentive openly, that bookings are rewarded and payment timing isn't, usually does more to unlock it than pressing any individual director.

• Does this work for a services business or a small AR portfolio? Yes, because the four categories are defined by cause, not by industry. A consultancy still has behavioral late payers, disputed deliverables that mirror process blockers, terms that drift from the engagement letter, and clients in trouble. A small portfolio skips the Pareto step almost entirely. With 20 customers you don't need to sort by impact first. You classify all of them and act, which is faster than a large book, not slower.

• How do I tell a behavioral late payer apart from an early-stage credit risk? Look at the shape of the timing, not the size of the delay. Behavioral late is stable. Day 58 to 62, quarter after quarter, predictable enough to forecast. Risk late is moving. The delay was 35 days, then 50, then a skipped payment, often alongside external signals like a credit downgrade or news of restructuring. A consistent late payer is a pricing problem. A deteriorating one is an exposure problem, and the right move there is an order hold and a credit review, not a repricing conversation. When in doubt, the trend direction tells you more than the absolute number of days.

Ask This Next Month

For your top 5 overdue customers: which are behavioral, which are contractual, which are process failures, which are credit risks? And for each, who owns the next step?

Controller Move

Export AR aging. Sort by overdue amount descending. Take the top 10 customers. For each one, check the five fields from the classification check (contractual terms, four-quarter payment pattern, open disputes, unapplied cash, credit-risk signals) and assign behavioral, contractual, process, or risk.

For the behavioral ones, estimate the unpriced credit: average incremental AR balance you're carrying because of the timing lag, times your annual cost of capital. For a customer paying 30 days late on a $180K recurring balance at 10% CoC, that's about $18K a year. Stack the numbers for all behavioral late payers in your portfolio. The total is usually the most uncomfortable slide you've ever brought to a Finance leadership meeting.

Then set the cadence. 30 minutes, monthly, four roles in the room (AR Lead, Credit Manager, Sales Lead, Operations Lead). Three questions per meeting (what got completed, what's next, what needs escalation). Track actions completed and cash moved, not DSO.

The first month is heavy because you're classifying for the first time and most of your top customers have multiple issues stacked. Month two is lighter. By month four, the meeting is mechanical and the aging report has structurally changed.

Last week, countback DSO showed you actual collection speed, separated from revenue timing noise. The gauge stopped lying.

This week, the aging report stopped being a calling list and became a triage list with four categories, each with a named owner.

Next week we cross to the other side of the balance sheet. Payables. The question shifts from "who owes us cash and why are they late" to "how much cash left our own account before anyone decided it should."

Go further

This article gives you the framework: the classification, the cadence, the conversation with Sales. The course gives you the tool that runs it from your ERP export. CC-AA02: Net Working Capital Optimization on GoFast.Finance. The AR view sorts your aging by overdue balance, applies the four-category logic to data you already have, and feeds a consolidated action list you can take straight into the monthly review.

This article originally appeared in the Practical Lean Finance newsletter on LinkedIn. Read the newsletter version. For shorter takes between Monday articles, follow me on LinkedIn.

Sources

1. Atradius, B2B Payment Practices Trends in North America 2025. Overdue B2B credit sales sat in the low 40s across the region in the 2025 survey: 43% in the US, 44% in Canada, 41% in Mexico. Stated reasons for late payment include customer liquidity pressure, payment-process delays, and invoice disputes, mapping respectively to the risk and process categories used here.

2. Allianz Trade, Accounts Receivable Aging: Importance, Method and Strategies. Aging analysis identifies overdue invoices, recurring late payers, and credit-risk signals, and helps prioritize collection effort. It surfaces the symptom. This article adds the operating layer the aging report does not supply on its own: cause and owner.

3. HighRadius, Guide to Dunning Management. A reference for what modern dunning platforms claim to do: structured reminders, escalation, risk-based segmentation, dispute tracking, and reason-code reporting. The methodology section credits these features while arguing they sit inside step one, and that a captured reason code still needs an owner, an action, and a deadline before it moves cash.

4. Stripe, Dunning Management. Baseline definition of dunning as a structured communication process to recover overdue payments. Used here to set the boundary of what dunning is: one legitimate intervention, not a diagnosis.

5. Working Capital Hub, Accounts Receivable: Complete Guide. Confirms aging analysis should explain drivers at transaction and customer level (invoice accuracy, disputes, terms misalignment, collection capacity) and notes that poor cash application leaves phantom receivables that distort AR balances, the unapplied-cash case in the process category.

6. John G. Salek, Accounts Receivable Management Best Practices (Wiley, 2005). Practitioner reference on prioritizing collections effort, resolving disputes, and treating late payment by cause rather than blanket follow-up. Available on Amazon.

As an Amazon Associate, GoFast.Finance earns from qualifying purchases. This adds nothing to your price and helps keep the newsletter free.

#finance #controlling #accountsreceivable #creditmanagement #collections