Invoice Processing Time Benchmarks:
Manual, OCR & AI (2026)

The single most common mistake in invoice-processing benchmarks is treating "9.2 days" and "12.5 minutes" as interchangeable answers to the same question. They answer different questions, move at different speeds, and are compressed by different levers.

Touch Time vs Cycle Time: The Two Metrics

Touch time (also called handle time or labor time) counts the minutes a human actually spends on one invoice — opening mail, keying data, matching purchase orders, routing for approval, filing. It is what determines staffing and labor cost. Cycle time counts calendar days from invoice receipt to payment, including every hour the invoice sits in a queue, in an inbox, or in an approver's folder. It is what determines supplier terms, late-payment risk, and cash flow. A fully manual team in Levvel's research shows the two moving together (14.6 days cycle / 10–15 min touch), but automation separates them: approval workflow tools compress cycle time to 3–5 days without touching a single field of data, while only automated capture compresses the 12.5 minutes of human effort.

Sources: IOFM (2024) — 12.5 min average touch time, stage decomposition via Nexus AP research roundup; Levvel Research (via Ascend, 2025) — manual 10–15 min / 14.6 days vs automated <2 min / 3–5 days; Ardent Partners (2025) — 9.2 days average / 3.1 best-in-class / 17.4 others, n=212; APQC Open Standards Benchmarking — top 2.8 / median 4 / bottom 7+ days.

Year Over Year: Cycle Time Is Falling, Slowly

The only cross-year series with consistent methodology is Ardent Partners' State of ePayables research. Average cycle time fell from ~11 days in 2022 to 9.2 days in 2024/2025 — a ~16% improvement over two years, driven mostly by workflow automation rather than faster data entry. The gap that matters more is horizontal: best-in-class teams (3.1 days) process invoices in roughly a third of the average time, and the bottom quartile takes 17.4 days — more than five times the best-in-class figure. The wide spread is not noise; it is the difference between teams that have automated approvals and exceptions and teams that have not.

Sources: Ardent Partners AP Metrics That Matter in 2025 (2024 data, n=212 AP professionals) — 9.2 days average, 3.1 days best-in-class; State of ePayables 2022 baseline (~11 days), via Parseur summary (2023) — labeled by year as the trend starting point; the 2022 figure is a secondary citation, not a primary report.

The two metrics compress at different speeds: cycle time is falling on the strength of approval-workflow automation (Ardent: 32.6% touchless in 2025, up as best-in-class hits 49.2%), while touch time stays flat until the data-capture step itself is automated — which is why "the invoice takes 9.2 days" and "we spend 12.5 minutes per invoice" can both be true at the same time.

Breakdown by Processing Method

The clearest lever on touch time is how data gets into the system. Every step from mailroom to archive takes the same route regardless of method — the difference is whether a human or software does the data entry, matching, and routing. The ranges below show why method, not team size, is the first question to ask: moving from fully manual to template/OCR capture cuts touch time by roughly two-thirds, and automated extraction cuts it to under two minutes per invoice.

Sources: IOFM (2024) — 12.5 min manual average (stage decomposition via Nexus AP); Levvel Research (via Ascend, 2025) — manual 10–15 min, automated <2 min; Ardent Partners (2025) — 4.8 min touch time under basic automation. Conservative midpoints used where sources disagree.

The zero-touch (straight-through processing) figures in the table are organization-level rates, not per-invoice measurements: Ardent's 32.6% average / 49.2% best-in-class applies across all methods mixed in a team. Hackett's research on AP-solution adopters reports a 60% average touchless rate for that subgroup. The <2 min AI figure is an analyst estimate (Levvel); the lowest published numbers (1.2 min vendor telemetry, 1–2 sec/page) are vendor-reported and are not used as independent facts on this page.

Breakdown by Invoice Complexity

Complexity is the second-largest driver of touch time after method. A simple single-line-item invoice with a matching PO takes a fraction of the time of a multi-line invoice that needs exception research. Industry practice groups invoices into three tiers; the IOFM average (12.5 min) sits above the simple tier precisely because the mix includes exceptions.

Simple/standard tier splits are directional industry estimates (template-based systems report 3–4 min on clean documents; Ardent's basic-automation 4.8 min is the closest measured anchor). The IOFM figures are the primary source for the average and exception rows. See Exceptions below for why the exception tier dominates real-world staffing.

Breakdown by Organization Size & Performance Tier

Size matters less than performance tier. APQC's Open Standards Benchmarking and Ardent's survey both segment by outcomes (best-in-class vs median vs bottom), and the spread between tiers is far wider than any difference by company size: bottom performers need roughly 4× the resources of top performers (14.4 vs 3.3 FTEs per $1B revenue, per APQC) while taking more than five times as long per invoice.

APQC figures from Open Standards Benchmarking (invoice cycle time measures; bottom-tier 14.4 vs top-tier 3.3 FTEs per $1B revenue). Ardent figures from State of ePayables 2024 survey (n=212 AP professionals). The two sources define tiers similarly (best-in-class = top quintile) and agree within ~0.3 days.

Breakdown by Processing Stage

IOFM's 2024 benchmark is the only widely cited source that decomposes the 12.5-minute average into stages — and the stage split is where automation pays off unevenly. Data entry (4 min) and PO matching (3 min) are the stages that capture software removes; approval routing (2.5 min) is what workflow tools remove; mailroom and archiving (1.5 min each) persist in some form for nearly every team.

IOFM 2024 AP Benchmarking stage decomposition, as transcribed in the Nexus AP research roundup (the underlying IOFM report is paywalled; the decomposition is cross-checked against Levvel's manual 10–15 min range and is directionally consistent). The "removed by" column is this page's synthesis of which automation lever maps to each stage, not a source claim.

Exceptions: The Multiplier Both Numbers Miss

Neither "12.5 minutes" nor "9.2 days" captures the real cost driver. Each exception adds 15–45 minutes of human work (IOFM) on top of normal handling, and 14% of invoices are exceptions on average — 9% for best-in-class teams vs 22% for the rest (Ardent 2025). At 22%, exception handling can double the effective labor per invoice: 78 normal invoices at 12.5 min + 22 exceptions at ~30 min each = roughly 1,635 min per 100 invoices vs 1,250 min with zero exceptions — a ~30% labor premium hidden inside the "average."

The exception dimension also has its own cycle-time benchmark: APQC's exception-resolution cycle time measure (25th percentile) is 5.0 days (n=461) — meaning even the fastest quartile spends five working days resolving a single exception once flagged. This is why 53% of AP professionals list exceptions as their single biggest challenge (Ardent 2025), and why touchless/straight-through rates — the share of invoices that never hit a human — are the metric most closely correlated with both lower touch time and lower cycle time.

Impact by Role

Who the time falls on changes what the benchmark means: for clerks, touch time is the whole job; for managers, cycle time is the SLA they report; for controllers, both feed the cost per invoice.

Sources: IOFM (2024) — touch time & exception add-on; Ardent Partners (2025) — cycle time, STP, exception rate; APQC — exception-resolution cycle time (n=461); IFOL via Precoro (2025) — weekly hours; The Hackett Group (2025) — touchless productivity ratios; BLS OEWS (May 2024) — clerk wages (mean $25.75/hr). The ~38 invoices/day and ~$5.36/invoice figures are arithmetic from cited rates, not published survey numbers.

How to Use This Data

You don't need a vendor benchmark to estimate your own invoice-processing staffing. The two-metric split makes it a two-line calculation — and it works whether you're planning headcount or justifying automation.

These are rough-estimate formulas, not financial advice. Your exact numbers will vary by industry, team size, and existing processes. The goal is to move from "we don't know" to "here's a defensible range."

Frequently Asked Questions

How long does it take to process an invoice manually?

About 12.5 minutes of human touch time per invoice on average (IOFM 2024), with Levvel Research reporting a 10–15 minute range for fully manual work. That is active labor — opening, keying, matching, routing, filing — not the calendar time the invoice spends in the process.

What is the average invoice processing time?

The answer depends on which metric you mean: 9.2 calendar days from receipt to payment on average (Ardent 2024/2025, n=212), versus 12.5 minutes of human effort per invoice (IOFM 2024). Cycle time determines cash flow and supplier terms; touch time determines staffing and labor cost.

What is a good invoice processing cycle time?

Best-in-class teams process invoices in 2.8–3.1 days — APQC's top performers at 2.8 days and Ardent's best-in-class quintile at 3.1 days. The cross-industry average is 9.2 days, and bottom-quartile firms take 17.4 days.

How long does it take to process an invoice with OCR or templates?

Template/OCR-assisted processing drops human touch time to 3–4.8 minutes per invoice — Ardent's basic-automation measurement is 4.8 min, industry template systems report 3–4 min on clean documents — versus 10–15 minutes manually. Cycle time drops from ~14.6 days to 3–5 days once workflow tools are added.

What percentage of invoices are processed without any human touch?

On average, 32.6% of invoices go through straight-through processing with no manual handling (Ardent 2025); best-in-class teams reach 49.2%, and companies that have adopted AP solutions average 60% (Hackett). The remaining invoices are where exceptions happen.

How many invoices can one AP clerk process in a day?

At the IOFM average of 12.5 minutes of touch time per invoice, one clerk covers roughly 38 invoices per 8-hour day of pure processing time — fewer in practice once exceptions (15–45 min each) and meetings are added. APQC's Open Standards Benchmarking points the same direction: best-in-class teams process ~32 invoices per FTE per day vs ~3 for laggards (directional figure, reported via industry roundups).

What is the difference between invoice touch time and cycle time?

Touch time is minutes of human effort per invoice; cycle time is calendar days from receipt to payment. An invoice can sit for eight days waiting for approval (cycle time) while only 12.5 minutes of actual work goes into it (touch time). The two respond to different levers: approval automation compresses cycle time, while only automating data capture compresses touch time.

How much time do invoice exceptions really add?

Each exception adds 15–45 minutes of handling on top of normal processing (IOFM), and exceptions affect 14% of invoices on average — 9% for best-in-class vs 22% for other firms (Ardent 2025). Even the fastest quartile takes 5.0 days to resolve a flagged exception (APQC exception-resolution measure, n=461).

How much time can automating invoice processing save?

Academic research from the Politecnico di Milano International E-Invoicing Observatory (via Billentis, 2026) finds that converting PDF invoices to structured digital data cuts processing time by up to 88%; the Finnish government's business case shows a paper invoice costing 30–50 EUR to process falling to 1 EUR fully automated. On the touch-time axis, the published levers are 12.5 min → 4.8 min (template/OCR) → under 2 min (AI extraction).

Methodology & Sources

How This Page Was Built

This page aggregates benchmarks from 13 independent studies and reports spanning 2015–2026. Sources were selected based on three criteria: (1) publicly documented methodology with stated sample size where available, (2) data collected within the last 3 years (older sources are labeled by year), and (3) relevance to business invoice/AP workflows. Where multiple sources report the same dimension, conservative ranges (lowest–highest reported values) are used — we prefer to under-claim. Where sources disagree significantly (e.g., 9.2 days vs 14.6 days cycle time), the disagreement is explained by measurement definition (blended average vs fully-manual environment) rather than averaged away. The 12.5-min stage decomposition comes from a paywalled IOFM report transcribed by a vendor research roundup; it is cross-checked against Levvel's 10–15 min range for consistency.

Source List

1. Ardent Partners — AP Metrics That Matter in 2025 (eBook based on State of ePayables 2024). Survey of 212 AP professionals. Provides average cycle time 9.2 days, best-in-class 3.1 days vs others 17.4 days, touchless/STP 32.6% average / 49.2% best-in-class, exception rate 14% (9% BIC vs 22% others), 53% citing exceptions as top challenge, 24% of organizations with full invoice automation.
2. APQC — Open Standards Benchmarking. Nonprofit benchmarking organization (1,000+ member organizations). Provides cycle time by tier (top 2.8 / median 4 / bottom 7+ days) and exception-resolution cycle time (25th percentile 5.0 days, n=461), plus resource gap data (bottom 14.4 vs top 3.3 FTEs per $1B revenue). Official measures page is bot-protected; figures cross-checked via Planergy, Stampli, and Auxis summaries.
3. IOFM — AP Benchmarking + Ask the Expert. Institute of Finance & Management, the leading AP association. Provides the 12.5-min average touch time with stage decomposition (mail 1.5 + entry 4 + PO match 3 + routing 2.5 + archive 1.5), +15–45 min per exception, and ~9.7 days receipt-to-ready-to-pay discussion. Stage decomposition transcribed via Nexus AP (paywalled report).
4. Levvel Research (now Deloitte) via Ascend Software. Industry analyst research roundup. Provides manual 5–6 invoices/hour (≈10–12 min each) vs 30+ automated; manual touch 10–15 min → <2 min automated; manual cycle ~14.6 days → 3–5 days automated; ~30% industry touchless vs 60–80% for high performers.
5. The Hackett Group — AP Solutions Research (2025). Consulting/benchmark firm. Provides 60% average touchless rate among AP-solution adopters, 3.5× AP productivity at ≥30% touchless, 59% cycle-time improvement, and Digital World Class finance teams at ~80% automated AP workflows with 42% fewer FTEs.
6. US Bureau of Labor Statistics — OEWS May 2024 / OOH. Official US government data. Provides the wage anchor for touch-time labor cost: Bookkeeping/Accounting/Auditing Clerks (43-3031), 1,373,680 employed, mean $25.75/hr / $53,560 annual, median $24.36/hr (May 2024); OOH median $49,210.
7. Politecnico di Milano / Billentis — 2026 E-Invoicing Report. Academic research (International E-Invoicing Observatory) cited in industry report. Provides the up-to-88% processing-time reduction from PDF → structured digital format, plus the Finnish Treasury cost case (paper 30–50 EUR → semi-auto 10 EUR → fully auto 1 EUR) and global invoice volume estimates (~600 billion invoices, 29% B2B e-invoicing).
8. Sterling Commerce (legacy research, ~2015, via Symtrax). Legacy industry study. Provides historical context on manual invoice errors (1.6% per invoice) and correction costs ($53.50 avg) — used only as background on the error side of manual processing, labeled by year.
9. Stampli × Probolski Research — AP Processing Times Survey (2023). Industry media survey of finance leaders. Provides 67.7% of respondents processing invoices in under 1 week and 51.6% handling 500+ invoices/month — a real-world cross-check on the cycle-time distribution.
10. IFOL — AP Automation Trends Survey (2025, via Precoro). Institute of Financial Operations & Leadership survey summarized by industry media. Provides 63% of AP teams spending 10+ hours per week on invoice processing in 2025, up from 52% in 2024 — the time-allocation trend.
11. Nexus AP — Invoice Processing Time Benchmarks research page (vendor roundup). Vendor aggregation page. Used only to cross-check the IOFM stage decomposition and Ardent 4.8-min basic-automation figure; its own AI-native telemetry (1.2 min) is vendor-reported and not used as an independent fact on this page.
12. Ardent Partners — State of ePayables 2022 (via Parseur summary, 2023). Industry research baseline reported through a vendor summary. Provides the ~11-day average cycle time (2022) used as the year-over-year trend start point; labeled by year and by secondary-citation status.
13. Qvalia / Billentis — 2026 Key Report Findings. Industry research summary. Provides the global e-invoicing adoption framing (29% B2B e-invoicing; Latin America 78% vs Europe 64% e-invoicing share) used for context on why the 88% reduction figure is reachable at scale.

Limitations

• Geographic coverage: The majority of sources are US/North America-based (Ardent, APQC, IOFM, Stampli, IFOL, BLS); the academic and e-invoicing anchors are European (Politecnico di Milano, Billentis, Finland). No reliable published benchmarks were found for invoice processing time across APAC, Latin America, or emerging markets.
• Methodology constraints: Several key figures (IOFM 12.5-min decomposition, APQC tier splits) come from paywalled or bot-protected primary reports transcribed by third parties — the secondary route is stated next to each figure. The AI/extraction touch time (<2 min) is an analyst estimate; the lowest published numbers are vendor telemetry and are labeled as such. Stampli/IFOL/Ardent figures are self-reported surveys, not measured timings. The "removed by" column in the stage table is this page's synthesis, not a source claim.
• Temporal gaps: The most recent primary data is 2024–2025 (Ardent, IOFM, BLS, IFOL). The Sterling Commerce error-rate figure dates to ~2015 and is used only for background. The only consistent cross-year series is Ardent (2022 → 2024/2025); the 2022 baseline is a secondary citation. No reliable year-over-year series exists for touch time itself.
• What we could not find: (1) Independent (non-vendor) measurements of AI-native touch time per invoice; (2) time benchmarks segmented by industry (APQC publishes cost by industry but not time); (3) multi-page vs single-page invoice time differences; (4) geographic variance in touch or cycle time. These gaps are real; we state them rather than fill them with marketing numbers.

Related references: Document Processing Cost per Record · Manual Data Entry Error Rates · OCR Accuracy by Document Type

Related reading: Cost to extract 100 invoices per month · Scaling invoice processing without headcount