A supplier calls about an invoice that is 45 days old. AP checks the system: “Pending approval.” Pending with whom? Since when? In which entity? Nobody can say without opening three inboxes and a spreadsheet.
That moment is the real approval bottleneck. Slow approvers are a symptom. The root problem is that nobody can see where time is being lost so nobody can fix it.
Approval is now the joint biggest problem in accounts payable. In Ardent Partners’ 2026 survey of 194 AP and finance leaders, 48% said invoice and payment approvals take too long, tied with exceptions as the top challenge. Another 24% named a lack of visibility into invoice and payment data (Ardent Partners, AP in 2026, July 2026). Add more legal entities, cost centers and approvers, and the delay gets harder to locate.
This guide shows how to find approval bottlenecks across entities and departments: the timestamps to capture, the five metrics that expose delays, how to build a cross-entity view, and the fixes that work once you can see the problem.
Key takeaways
- Measure approval time per stage and per approver, not just total cycle time. A 12-day average hides a 2-day department and a 30-day one.
- Normalize data across entities first: one status model, one set of timestamps, one approver directory.
- The fastest wins usually come from routing rules, delegation and escalation — not from adding headcount.
Why approval delays hide in multi-entity and multi-department AP
In a single-entity company, a stuck invoice is usually on one person’s desk. In a group with several legal entities, shared services and department budget owners, the same invoice can pass through four or five hands — often in different systems. Each handoff is a place where time disappears without a trace.
The patterns repeat across finance teams:
- Each entity runs its own workflow. One subsidiary approves in the ERP, another by email, a third in a shared mailbox. There is no common definition of “approved” or “pending.”
- Approval happens outside the system of record. A manager replies “OK” to a forwarded PDF. The approval is real, but the timestamp is buried in someone’s inbox.
- Only the total is measured. Teams track days-to-pay, not where those days went. A slow department is averaged away by fast ones.
- Approver data is out of date. People change roles, go on leave or leave the company. Invoices keep routing to them until someone notices.
- Exceptions and approvals blur together. An invoice waiting on a PO mismatch looks the same as one waiting on a signature, so the wrong team gets chased.
- Intercompany and shared-cost invoices have no clear owner. When a cost is split across entities or departments, each assumes the other will approve.
The result is a process people experience as slow but cannot diagnose. That is why visibility comes before optimization.
What approval bottlenecks really cost
The gap between AP teams that move invoices quickly and those that don’t is large, and it shows up in cost, not only speed.
| Benchmark | Best-in-class | All others |
|---|---|---|
| Invoice processing time | 3.1 days | 17.4 days |
| Cost to process one invoice | $2.78 | $12.88 |
| Invoice exception rate | 9.0% | 22.0% |
| Invoices processed straight-through | 49.2% | 23.4% |
| Staff time spent on supplier inquiries | 13.4% | 26.9% |
Source: Ardent Partners, AP Metrics that Matter in 2025 (212 AP and finance leaders; best-in-class = top 20% on cost and cycle time). The 2026 industry average is 11.6 days and $9.90 per invoice, with a 19.9% exception rate (Ardent Partners, AP in 2026).
Slow approvals cause four expensive things:
- Missed early-payment discounts. A 2/10 net 30 term is only worth taking if the invoice is approved within 10 days. Only 19% of AP teams pay early to capture discounts (Ardent, 2026).
- Late fees and strained suppliers. Suppliers who chase payments take AP time. The average AP team now spends 29.4% of staff time answering supplier inquiries (Ardent, 2026).
- A slower close. Unapproved invoices sit outside the books, which means more accruals, more manual adjustments and a later close in each entity.
- Weaker controls. When approvals are slow, people find shortcuts: approving in bulk, paying before sign-off, or skipping the approval matrix. With 41% of AP leaders reporting more fraud attempts than the year before, those shortcuts are expensive (Ardent, 2026).
Quick test: multiply invoices stuck more than 10 days by your average invoice value and by your typical discount rate. That number is what approval delay costs you in discounts alone each month.
The five metrics that expose approval bottlenecks
Total cycle time tells you that you have a problem. These five metrics tell you where it is. Track each one by entity, department and approver.
| Metric | How to calculate it | What it reveals |
|---|---|---|
| Stage cycle time | Time between status changes: received → coded → routed → approved → scheduled | Which stage, not just which invoice, is slow |
| Approver dwell time | Time from assignment to decision, per approver (median, not average) | The individual or role holding invoices |
| Pending-approval aging | Count and value of invoices awaiting approval, in 0–5, 6–10, 11–30, 30+ day buckets | Where risk of late fees and lost discounts is building |
| Reroute and reassignment rate | Share of invoices sent to more than one approver before a decision | Broken routing rules or an outdated approver directory |
| Exception-to-approval ratio | Invoices held for mismatch or missing data vs invoices held for sign-off | Whether the delay is a data problem or a people problem |
Two practical rules make these numbers useful:
- Use the median for dwell time. One invoice left for 90 days will distort an average and hide the approver who is consistently slow.
- Report count and value together. Ten small invoices stuck in marketing matter less than one large one stuck in a subsidiary waiting on the CFO.
How to build visibility across entities and departments: 6 steps
You cannot compare entities that measure different things. These six steps build one view of approvals across the group, starting with the data and ending with a routine.
- Agree on one status model. Define the same statuses for every entity, for example: Received, In coding, Exception, Awaiting approval, Approved, Scheduled, Paid. Map each entity’s local statuses to it.
- Capture a timestamp at every status change. The time an invoice enters and leaves each status is the raw material for every metric above. If approvals happen by email, this is where they get lost.
- Move approvals into one system of record. Approvals given in email, chat or on paper can’t be measured. Approvers should act in one place, ideally from a link or mobile app, so the decision and its time are logged automatically.
- Build and maintain one approver directory. List every approver by entity, department, cost center and limit. Assign an owner to update it when people join, move or leave. Most reroutes start here.
- Create a cross-entity approval view. One dashboard showing pending invoices by entity, department and approver, with aging buckets and value. Group controllers see the whole picture. Entity and department heads see their own slice.
- Set a weekly review rhythm. Fifteen minutes a week: review the oldest 20 pending invoices, the slowest approvers by median dwell time and any entity trending worse. Assign one action per finding.
Start small. If six steps feel like a project, start with steps 2 and 5 for your largest entity. One month of clean timestamps usually shows where the biggest delay sits.
How to fix bottlenecks once you can see them
Match the fix to what the data shows. Pushing every approver to “go faster” rarely works; changing the rules around them does.
| What the data shows | Likely cause | Fix |
|---|---|---|
| One approver has high median dwell time | Too many invoices or unclear responsibility | Raise their auto-approval threshold for low-risk spend, or split the queue |
| Delays spike during holidays or travel | No backup approver | Set automatic delegation when an approver is out of office |
| High reroute rate in one entity | Outdated approver directory or routing by name | Route by role and cost center, not by person |
| Invoices wait on second and third approvals | Approval chains longer than needed | Use value-based tiers: one approver below a threshold, more only above it |
| Many invoices age past 10 days | No reminders or escalation | Automatic reminders at day 3, escalation to the manager at day 7 |
| Exceptions outnumber approval holds | Data problem, not a people problem | Fix PO matching, vendor master data and invoice capture before touching approvals |
| Intercompany or split invoices stall | No clear owner | Assign a default approver per cost allocation rule |
Keep controls intact. Faster approvals should never mean weaker ones. Keep segregation of duties, keep approval limits by role, and log every delegation and override. A clean audit trail is the proof that speed didn’t cost you control.
Where automation helps. AP automation platforms such as DOKKA can apply routing rules, delegation and escalation consistently across entities, and record every step with a timestamp. That gives you the metrics above without building them by hand.
Self-audit: can you see your approval bottlenecks?
Answer these in 10 minutes. Each “no” marks a blind spot.
- Can you say, right now, how many invoices are awaiting approval in each entity, and their total value?
- Do all entities use the same approval statuses?
- Is every approval recorded in one system, with a timestamp?
- Can you name the five approvers with the longest median dwell time last month?
- Do you know what share of “pending” invoices are waiting on an exception rather than a signature?
- Is there a named owner for the approver directory, and was it updated this quarter?
- Does every approver have a backup for absences?
- Do overdue approvals escalate automatically?
- Does someone review approval aging every week?
Scoring: 8–9 yes means you have visibility; focus on fine-tuning. 5–7 means partial visibility; start with the step you’re missing. Fewer than 5 means you are managing approvals by memory and email — start with timestamps and a single system of record.
FAQ
What is an approval bottleneck in accounts payable? It is any point in the approval workflow where invoices wait longer than they should. It is usually a specific approver, department, entity or approval tier. It differs from an exception, where an invoice is held because of a data mismatch.
How long should invoice approval take? There is no single standard; it depends on invoice volume and payment terms. For context, best-in-class AP teams process an invoice end to end in 3.1 days, against 17.4 days for everyone else (Ardent, 2025). A practical target: approve fast enough to capture early-payment discounts. For common 2/10 terms, that means within 10 days of receipt, with routine invoices approved in a few days.
How do I compare approval times across entities with different ERPs? Map each entity’s statuses to one shared status model. Then pull status-change timestamps from each system into one report, or use an AP platform that sits across all entities. Compare medians, not averages.
Should I measure approvers individually? Yes, but use the data to fix routing and workload, not to blame people. A slow approver often has too many invoices, unclear authority or no backup.
Will faster approvals weaken internal controls? Not if you keep role-based approval limits, segregation of duties and a full audit trail. Clear rules and automatic logging usually make controls stronger, because fewer approvals happen informally.
Sources
- Ardent Partners, AP in 2026: Performance, Payments, and the Path Forward (July 2026; 194 AP, finance and P2P leaders; sponsored by Bottomline)
- Ardent Partners, AP Metrics that Matter in 2026, Pt. 1: Reporting and Analytics Move to the Top of AP’s Agenda (October 2026)
- Ardent Partners, AP Metrics that Matter in 2025 (February 2025; 212 AP and finance leaders; sponsored by Pagero)
