Invoice Approval Cycle Time: Diagnosing and Fixing AP Bottlenecks
Ask an AP manager how long it takes to approve an invoice and you will usually get a confident answer: a few days, maybe a week. Then pull the data, and the real number is often three weeks or more. The gap is not dishonesty — it is that almost nobody is actively working on the invoice for most of those three weeks. It is sitting. In an inbox, in a queue, on a project manager's pile, waiting for a second signature that nobody flagged as overdue.
Approval cycle time is the single most diagnostic metric in accounts payable. It exposes whether your workflow is a process or just a sequence of pauses. And it matters in hard dollars: slow approvals forfeit early-payment discounts, trigger late fees, and strain the subcontractors whose cash flow depends on you. This post is about measuring approval cycle time honestly, finding where invoices actually wait, and applying the levers that compress it.
Approval cycle time is the elapsed time from when an invoice enters your system to when it is fully approved and ready to pay. It is wall-clock time, not work time — weekends, evenings, and idle days all count, because the invoice is not getting paid on any of them. That distinction matters, because the instinct in AP is to measure effort ('it only takes ten minutes to approve') when the metric that affects vendors and discounts is duration.
Be precise about the endpoints. Cycle time starts at intake — the moment the invoice is captured, not the moment someone gets around to logging it. It ends at final approval, the last signature required before the invoice is eligible for a payment run. Keep it separate from two adjacent metrics: invoice processing time (the keying and coding before approval routing) and days payable outstanding (which runs all the way to the payment date and is partly a deliberate cash decision). Approval cycle time is the middle stretch, and it is the part most fully within your control.
About 0 days
Average time to process and approve a single invoice across all organizations (Ardent Partners)
Ten days is the cross-industry average, and construction routinely runs longer because of how its approvals are structured — more on that below. Top-performing AP operations clear invoices in a few days. The distance between those two numbers is almost entirely waiting, and waiting is fixable.
An invoice that takes twenty days to approve was not worked on for twenty days. It was worked on for perhaps thirty minutes and idle for the rest. Understanding the specific reasons invoices sit is the whole game, because each reason has a different fix.
An invoice arrives without a PO number, without a cost code, with an unclear scope, or referencing a job nobody can immediately identify. It cannot be routed until someone resolves the gap, so it goes into a holding pattern while AP emails the vendor or the project team and waits for a reply. This is one of the largest sources of delay, and it is upstream — the cure is cleaner intake, not faster approving.
When the workflow does not deterministically know who approves a given invoice, a human has to decide — and that decision itself is a delay. Worse, invoices get routed to the wrong person, sit unnoticed, and eventually bounce back. Construction is especially prone to this because the right approver depends on the job, the cost code, and the amount, and that logic often lives in someone's head rather than in the system.
In construction, the people who must approve invoices are project managers and superintendents, and they are on jobsites, not at desks. Invoice approval is not their primary job and it competes with everything that is. An invoice can sit on a PM's queue for a week simply because the PM has not been at a computer with a moment to spare. This is a structural reality of the industry, not a personal failing, and the fixes are mobile approval and good reminders.
Many approval workflows are strictly sequential: the invoice goes to the PM, then to the division manager, then to the controller, each step starting only when the prior one finishes. If each approver takes three days, a three-step chain takes nine days even though no single person was slow. Sequential routing is the most common structural cause of long cycle times, and often the easiest to fix.
Before optimizing anything, separate the two failure types. Some delay is people not acting (bandwidth, reminders). Some delay is the process making them wait their turn (serial chains, unclear routing). They have completely different fixes — and confusing them is why approval projects stall.
An average cycle time is a starting point, not an answer. The average tells you that you have a problem; it does not tell you where it lives. To act, you have to segment.
Segment approval cycle time along these dimensions
- By stage — measure the dwell time at each approval step, so you can see exactly which handoff invoices die in
- By approver — find the individuals or roles where invoices consistently sit longest
- By project and project manager — cycle time often varies enormously between jobs depending on the PM
- By vendor or invoice type — subcontractor pay applications, material invoices, and equipment rentals behave very differently
- By amount — large invoices may legitimately route through more approvers; small ones should be quick
- By distribution, not just the mean — the average can look fine while a long tail of invoices takes forty days
Per-stage dwell time is the most valuable cut. It converts a vague complaint ('approvals are slow') into a precise finding ('invoices wait an average of nine days at the division-manager step'). That finding points straight at a fix. A capable AP platform timestamps every state change automatically, so this analysis is a report rather than a stopwatch exercise — Covinly tracks dwell time per stage and per approver out of the box, which turns cycle-time diagnosis into something you can do continuously rather than once a year.
Once you know where invoices sit, a small set of levers does most of the work. Apply them to the specific bottleneck your segmentation found, rather than across the board.
Get AP insights in your inbox
A short monthly roundup of construction AP + accounting posts. No spam, ever.
No spam. Unsubscribe anytime.
Parallel Approvals
If two approvers do not depend on each other's decision, route the invoice to both at once. A PM confirming the work was performed and a controller confirming budget availability are checking different things; there is no reason the controller must wait for the PM. Converting a serial chain to parallel where the logic allows can cut days off the cycle without removing a single control. Reserve true sequence only for steps that genuinely depend on the prior outcome.
Not every invoice needs the full approval chain. A small invoice that matches a purchase order and receipt within tolerance carries little risk and adds little value when a person stamps it. Routing low-value, fully matched invoices straight through removes a large volume of items from the queue, which both shortens those invoices' cycle time and frees approvers to attend faster to the invoices that genuinely need judgment.
Most of the PM-bandwidth delay is not refusal — it is that the invoice fell off the radar. Automated reminders bring it back. Set a clear service-level expectation per step, send a nudge as it approaches, and escalate to a backup approver or a manager when it is breached so a single unavailable person cannot stall a payment indefinitely. Reminders are low-effort and high-yield: they often produce the biggest single improvement.
Because missing information is such a large source of delay, anything that improves intake quality compounds. Capture PO numbers and cost codes at the point of entry, validate that required fields are present before an invoice is allowed into the approval workflow, and route incomplete invoices into a dedicated exception path so they do not silently clog the main queue.
Resist piling on more approval steps as a control instinct. Every approver added lengthens the cycle and dilutes accountability — when five people sign, no one feels they truly own the review. Fewer, well-placed approvals beat a long chain.
Approval cycle time is not a vanity metric. It connects directly to money and to relationships, which is why it earns the attention.
The clearest financial link is early-payment discounts. A 2% discount for paying within ten days is worth capturing — annualized, it is a return few other uses of cash can match — but it is unreachable if an invoice takes three weeks just to clear approval. Slow approval forfeits the discount before the payment decision is even on the table. The same delay drives avoidable late fees and interest on invoices that should have been paid comfortably on time.
The relationship cost is just as real. Subcontractors run on tight cash, and a general contractor with a reputation for slow payment is a contractor that gets higher bids, less scheduling priority, and the weaker subs. Predictable, prompt payment is a competitive advantage in the trades. Approval cycle time is the lever that makes prompt payment possible — you cannot pay quickly what you have not yet approved.
“We always blamed our approvers for slow invoices. Then we measured dwell time by stage and saw most of it was three sequential signatures that did not actually depend on each other. We ran them in parallel and the cycle dropped by a week — same people, same controls.”
— Controller, regional general contractor
Long approval cycle times are almost never an effort problem. They are a structure problem — invoices sitting in inboxes, waiting in serial chains, stalled on a missing cost code, forgotten on a project manager's queue. The path forward is methodical: measure cycle time as honest wall-clock duration, segment it by stage and approver until you know exactly where invoices wait, then apply the matching lever — parallelize independent approvals, auto-approve the low-risk volume, add reminders and escalation, tighten intake. Done well, approval cycle time falls from weeks to days, and that compression shows up as captured discounts, fewer late fees, and subcontractors who put you at the top of their list.
Written by
Sarah Blake
Head of Product
Former AP Manager at a $200M construction firm, now leads product at Covinly. Writes about what AP teams actually need from automation — beyond the marketing promises.
View all posts