Concur Invoice Capture Processing: Managed vs Client Managed

Concur Invoice Capture Processing is the OCR and machine-learning layer inside SAP Concur Invoice that auto-populates header fields (vendor, invoice number, date, totals, tax) and line-item fields (description, quantity, unit price, expense type categorisation) before each invoice enters the approval workflow. Every Concur Invoice tenant uses Capture Processing in some form. The configuration question is which of three modes runs the verification step that decides whether the OCR layer's output is trustworthy enough to advance.

The three modes:

• Concur Managed Capture is fee-based: SAP staff verify OCR output before each invoice advances in workflow.
• Client Managed Capture is free at the licence line: the client's own team verifies OCR output internally.
• Dual Support is the hybrid that routes part of the volume to SAP-staff verification while leaving the rest with the client team.

That is the snippet-shaped answer. The harder question — the one SAP's own documentation describes mechanically without helping you answer — is which to pick. SAP's help pages tell you how each mode works; partner content tilts toward upselling Managed; alternatives roundups argue replacing Concur entirely. The page-one buyer-decision walkthrough on Concur Invoice Capture Processing managed vs client managed is conspicuously absent. This article is that walkthrough.

The decision is a Concur Invoice question specifically, not a generic SAP question. Other SAP products solve invoice OCR differently — the SAP Business One invoice OCR setup options take a different shape because the product, the audience, and the integration model differ. Concur Invoice's three-mode model belongs to Concur Invoice.

What follows: each mode in turn (what it does operationally, what it costs honestly), then the trade-off matrix that lands the head-to-head, then a fourth architectural option — upstream extraction that delivers a clean structured payload to Concur via the import-invoice path — that sidesteps the three-mode choice for line-item-heavy AP shops. The final section is decision logic keyed to firm size, monthly invoice volume, AP team capacity, line-item complexity, and audit posture.

---

Concur Managed Capture: What the Fee Buys

Under Concur Managed Capture, SAP-employed staff sit in front of the Capture Processing verification queue and confirm or correct each invoice's OCR output before that invoice advances into the approval workflow inside Concur. Your own AP team picks up invoices in workflow with the header and line-item fields already verified by an SAP human. Capture-stage exception triage — the work of resolving low-confidence fields, reading awkward scans, deciphering supplier templates the OCR engine hasn't seen before — moves off your queue entirely.

SAP positions Managed verification under a contracted turnaround SLA, typically expressed in hours from receipt to verified-and-released. The specific hour count varies by contract tier and is not a public number. Overflow handling for peak volumes (month-end supplier statement waves, quarter-close flushes) is negotiated separately, often as a contracted overflow ceiling beyond which a different rate or a queueing behaviour applies.

The cost is where most reader expectations need adjusting. SAP does not publish list pricing for the Concur Capture Processing fee. The per-invoice rate is set per customer in a quote, and partner content that quotes a number is usually quoting one customer's deal in one geography in one year. There is no honest single per-invoice number to give you.

What is honest is the list of variables that drive the quote you will receive:

• Monthly invoice volume tier. Higher volumes draw lower per-invoice rates, but the tier breaks vary by contract.
• Contracted turnaround SLA. A two-hour verified-and-released SLA prices differently from a same-business-day SLA.
• Line-item depth. Header-only verification is one rate; full line-item verification on dense supplier templates is another. SAP staff verifying twenty SKU rows on a single invoice are doing more work than verifying a header.
• Geographic scope of the supplier base. Multi-language and multi-script verification commands a different price than single-region English. A supplier base spanning Latin scripts, CJK, Arabic, and Cyrillic invoice formats demands verifier coverage that English-only does not.
• Integration scope. Custom routing rules, custom exception escalation paths, and bespoke approval-handoff configurations add complexity SAP prices in.

When you take the quote conversation, the SAP Concur Invoice OCR cost discussion is more useful when you push for the rate to be unbundled. Specifically: ask for the per-invoice rate broken out by line-item depth and SLA tier; ask for explicit overflow and exception-handling pricing rather than a blended rate that obscures peak-period economics; ask for written confirmation of which supplier-base languages and scripts are covered at the contracted rate; and pin down precisely what the billing unit is, because "per verified invoice" can mean per-invoice, per-page, or per-line-item depending on the contract template, and the difference matters at scale.

Beyond the verification labour itself, Managed shifts a few less-visible costs off your AP function. Capture-stage exception management goes with it — your team stops triaging low-confidence-score invoices each morning. The OCR-tuning feedback loop (the cycle of corrections that, over time, improves recognition for your specific supplier base) is run by SAP rather than by your verifiers. And invoices typically arrive in approval workflow with cleaner header and line-item data, which often produces a measurable reduction in downstream rework — coding corrections, three-way-match exceptions, payment-block resolutions traceable to a mis-captured field.

None of this makes Managed automatically the right answer. It is a labour-and-accountability transfer with a price tag, and whether the price tag is worth what it transfers depends on what your alternative looks like.

---

Client Managed Capture: Free at the Licence Line, Real at the FTE Line

Under Client Managed Capture, the OCR layer extracts the same header and line-item fields as it does under Managed — the engine is identical — but your own AP team works the verification queue. Verifiers see each invoice in Concur's verification UI, with field-level confidence scores and a source-document overlay that lets them confirm or correct each field against the original PDF or image. Once a verifier releases the invoice, it advances into the approval workflow. Both modes end at human-verified output; the only difference is who the human is.

That equivalence at the output side is why Client Managed is "free." There is no separate fee on the licence line for the verification step. The cost lives entirely in your team's time, which is why "free" needs immediate qualification.

A defensible way to think about the FTE-hours load is to multiply seconds-per-invoice by monthly volume, with the seconds-per-invoice band reflecting your supplier mix. For straightforward header-only invoices on familiar templates with high OCR confidence, verifiers typically spend somewhere in the range of thirty to sixty seconds confirming and releasing. For line-item-heavy invoices (manufacturing supplier statements, professional services breakdowns, logistics line detail), the band stretches to ninety seconds to three minutes, sometimes more. The variance is driven by line-item count, OCR confidence on the specific supplier templates, and how familiar the verifier is with the supplier base. At a monthly volume of 5,000 mixed invoices, the math lands somewhere between a half and a full FTE just on capture verification. The numbers are illustrative — your actual figure depends on supplier-base complexity, OCR-tuning maturity, and verifier experience — but they make the point that the licence-line "free" understates the operational reality.

The seconds-per-invoice math also misses a few second-order costs that show up as soon as the function moves past steady state. Multi-region invoice arrival means shift coverage, which means either staggered shifts or a verifier population large enough that capture stays below an SLA-relevant queue depth across the working day. Verifier turnover means training cycles, and verifiers reach competence on supplier-template recognition over weeks, not days. The OCR-tuning feedback loop is now your loop — every correction is a signal someone in the AP function has to manage if recognition is going to improve over time. And the audit trail of who-verified-what becomes part of your SOX evidence chain rather than something SAP keeps for you.

The throughput ceiling is the team's throughput. Client Managed handles whatever volume your verifiers can process in their working hours, no more. Steady-state volume is fine. Peak-period spikes — month-end supplier statement waves, quarter-close invoice flushes, annual-contract-renewal billing surges — hit the verification queue first. If the team is sized for steady state, the queue depth at peak becomes an approval-cycle bottleneck that propagates downstream into payment-cycle delay and supplier-relationship friction.

None of this is a "free is bad" framing. Client Managed is a legitimate operational choice for firms whose AP teams are already sized for verification work, whose supplier base is consistent enough that OCR tuning produces compounding gains over time, and whose audit posture wants verification kept inside the firm rather than handed to an external party. It is also worth saying that the heaviness of the verification load is not purely a function of volume — it is also a function of how documents reach the verification queue in the first place. The decisions involved in designing a digital mailroom for AP intake (channel rules, classification before capture, removing remittance advice and email cover sheets at the door rather than at the verifier's screen) shape how much of the capture queue is real verification work versus filtering work the intake architecture should have done.

Client Managed is real work, sized realistically. Read against a Managed quote, the comparison gets honest only when both sides are loaded honestly.

---

Dual Support: When the Hybrid Actually Applies

Dual Support is the configuration that runs both modes side by side inside one Concur Invoice tenant. Routing rules — defined at setup, not invoice by invoice — direct part of the volume to SAP-staff verification under Managed Capture, while the rest stays in the client's verification queue under Client Managed. SAP's own documentation names Dual Support; what it doesn't do is walk through when Dual Support actually earns its complexity. Four operational patterns make the hybrid worth running.

Geographic split. Some country or regional supplier bases route to SAP-staff verification; the rest stay Client Managed. The pattern fits firms with one or two regions where the supplier-base mechanics make Managed disproportionately valuable — local-language tax documents that the in-house verifier population is not staffed to read, or jurisdictions where invoice-format variation defeats template-based OCR tuning — while a dominant English-language supplier base in the home region handles efficiently with internal verification.

Volume overflow and peak coverage. Steady-state volume is verified by an in-house AP team sized for that load. Above a contracted threshold, additional invoices route automatically to SAP-staff verification. This protects approval-cycle SLAs through month-end and quarter-close without permanently sizing the AP team for the peak-month volume that only happens four times a year. The economics are particularly clean for firms whose volume curve is sharply seasonal rather than steadily increasing.

High-complexity invoice subset. Line-item-heavy invoice categories — manufacturing supplier invoices with dozens of SKU rows, professional-services invoices with detailed time-and-materials breakdowns, logistics invoices with per-shipment line breakdowns — route to SAP-staff verification, while header-only or low-complexity supplier invoices stay Client Managed. The pattern targets the per-invoice spend at the invoices where line-item OCR is hardest, without paying the Managed rate on the long tail of simple invoices where in-house verifiers are perfectly adequate.

Supplier-tier routing. Strategic-supplier invoices, where a delayed or mis-captured invoice damages a relationship or breaches a contract SLA, route to Managed. Long-tail supplier invoices stay Client Managed. The criterion is business risk on individual invoice accuracy rather than invoice complexity per se.

The cost of running Dual Support is operational, not just per-invoice. Routing rules need maintenance as the supplier base evolves — new suppliers have to be classified, sunset suppliers cleared out, jurisdictional changes (a German subsidiary spun off, a UK supplier reclassified post-Brexit) reflected in the rules. Billing reconciliation gets more complex because SAP invoices the client only for the Managed share, and the contract has to define exactly which invoice volume falls into that share when an invoice's routing is rule-driven rather than manual. Exception-handling paths fork by route, so the team needs separate playbooks for "exception on a Managed invoice" and "exception on a Client Managed invoice." Dual Support is not free hybrid; it is two operational models running in parallel inside one tenant, and the firm pays for that parallelism in operational overhead even before the per-invoice fees land.

Dual Support is the wrong answer when the volume is small-to-mid and the supplier base is homogeneous. The routing complexity outweighs any per-invoice saving, and the firm ends up paying operational overhead for hybrid mechanics it doesn't actually need. It is also the wrong answer when the AP function explicitly wants a single verification model for audit clarity — auditors testing a single verification model walk through one control; auditors testing Dual Support have to test two control models and the routing logic that decides which invoice goes through which.

The trade — service-handled capture verification on part of the volume, internal verification on the rest, with routing logic deciding what goes where — is not unique to Concur. Oracle Fusion Intelligent Document Recognition for AP has analogous service-vs-self-serve structure, and looking at how another enterprise AP platform frames the same hybrid choice is useful for sanity-checking that the underlying decision is platform-shape, not vendor-marketing.

---

The Trade-Off Matrix: Cost, Throughput, Accountability, Audit Posture

The head-to-head between Concur Managed Capture and Client Managed Capture lives in four dimensions: cost, throughput, accountability for capture errors, and audit posture. Each dimension carries its own weight, and the right configuration is the one that lands the trade in the right place across all four for your firm. There is no winner-take-all axis.

Cost. Managed carries a per-invoice fee that scales linearly with volume — the variables driving the quote sit in Section 2 above. Client Managed carries no licence-line fee but a real FTE-hours cost that scales with both volume and supplier-base complexity, sketched in Section 3. The honest comparison is between the Managed quote you negotiate and the fully-loaded internal cost of verifier headcount: salary plus benefits plus shift coverage plus training plus turnover plus the manager time spent supervising the function. Many firms model this dimension incorrectly. The trap is comparing a Managed quote against the marginal cost of an existing verifier's time — "we already have the people, so the in-house option is free at the margin" — when the realistic comparison is against the fully-loaded cost of the headcount the function actually requires once volume is taken seriously. The marginal-cost framing flatters Client Managed; the fully-loaded framing tells the truth.

Throughput. Managed throughput is contracted via the SLA, and the contract makes peak handling SAP's problem rather than yours. You don't size for peaks because the contract obligates SAP to. Client Managed throughput is your team's throughput. Peak periods become approval-cycle bottlenecks unless the team is sized for them, and sizing for peak means carrying capacity that sits idle through the rest of the month. Firms with seasonal volume curves feel this dimension hardest; firms with steady volume can absorb it more easily.

Accountability for capture errors. Under Managed, when a line-item field is mis-captured and downstream coding or three-way-match fails because of it, the verification step that should have caught it is SAP's. The accountability sits with the contracted service. Under Client Managed, capture errors stay internal — the verifier owns them, the AP function is accountable for OCR-stage rework, and root-cause analysis on a recurring miscapture pattern is your work to do. Neither framing is intrinsically better; the question is which accountability shape your operating model actually wants.

Audit posture and segregation of duties. This is where Managed and Client Managed diverge most sharply for SOX-relevant firms. Under Client Managed, the verifier of OCR output is internal staff, and depending on team size and role definition, that same verifier may also touch downstream approval steps — coding, voucher review, payment release. The combination raises a segregation-of-duties question your audit function has to address. Under Managed, the verifier is external SAP staff with no path to downstream approval; the separation is clean by construction.

The clean separation under Managed does not make Client Managed disqualifying for SOX firms — it just shifts the work onto the firm to design and document compensating controls. PCAOB Auditing Standard 2201 acknowledges that segregation-of-duties limitations may force a company to implement alternative controls that the auditor must evaluate for effectiveness. In Client Managed terms, the alternative controls that typically show up in SOX-tested AP functions are mandatory dual-review on capture corrections (a second verifier signs off on any OCR field correction above a materiality threshold) and rotation of verification duty across staff (no single verifier sees a given supplier's invoices over consecutive periods). The controls add operational cost; the point is that they are designable and auditor-acceptable, not that the dimension is moot.

This same trade-off shape — service-handled capture verification on one side, client-team verification on the other, with audit and cost implications that follow — is not unique to Concur. The Coupa invoice processing workflow and pain points play out across analogous dimensions on a different enterprise AP platform, and seeing the same decision framed in another vendor's vocabulary helps confirm that you're reasoning about the underlying operating model rather than reacting to Concur-specific configuration mechanics.

What is not a meaningful trade-off dimension between Managed and Client Managed is OCR-layer accuracy. Both modes use the same OCR engine, both end at human-verified output, and the accuracy ceiling is similar. The trade is about who does the verification work and what controls wrap it, not about whether verification happens or how good the underlying recognition is. Reading partner content that frames Managed as "more accurate" or Client Managed as "good enough on accuracy" is reading marketing — the accuracy comparison between the two modes inside Concur is a non-question.

---

The Fourth Path: Upstream Extraction Into Concur

There is an architectural option that does not appear in SAP's three-mode framing because it lives outside the framing entirely. Instead of choosing between Managed, Client Managed, or Dual Support for Capture Processing verification, the firm extracts invoice data upstream of Concur using a dedicated extraction layer, then loads the resulting structured payload into Concur via the platform's import-invoice paths. The Capture Processing mode setting still exists in tenant configuration, but the volume routed through Concur's native OCR drops sharply — for line-item-heavy invoice categories, often to zero.

The architectural case is specific. Concur's native OCR is strongest on header fields and weakest on dense line-item tables. Invoice categories where line-item OCR strain shows up are well-defined: manufacturing supplier invoices with dozens of SKU rows, professional-services invoices with detailed time-and-materials breakdowns, logistics invoices with per-shipment line breakdowns, energy and utility invoices with metered-period detail. For these invoice categories, both Managed and Client Managed verification are doing heavy correction work on line items. SAP-staff verifiers under Managed earn their per-invoice fee by re-keying line-item fields the OCR layer didn't capture cleanly; in-house verifiers under Client Managed do the same work without the fee but at the FTE-cost reality Section 3 walked through. Either way, line-item capture is where the verification load concentrates.

An AI-powered invoice data extraction layer purpose-built for line-item capture changes that economics. The upstream layer takes the invoice files, extracts header and line-item fields into a structured spreadsheet (Excel, CSV, or JSON), and the structured file is what Concur ingests via its import-invoice path. The interaction model is prompt-driven: a finance operator specifies header columns and per-line-item columns once, and the same specification produces the same column-shape across whatever batch volume the AP function is processing. That stability is the property that matters for downstream import — Concur's import paths consume a consistent payload, not a payload whose shape depends on the supplier template the OCR engine happened to see.

What changes operationally inside Concur is meaningful. The AP team focuses Concur's capabilities on the work Concur is genuinely strong at — approval routing, three-way match against POs and receipts, GL coding rules, payment scheduling, supplier-master controls — rather than spending verifier time on capture. Capture-stage exceptions land at the upstream layer, with their own verification and correction UI, rather than queueing up inside Concur's verification queue and competing with workflow-stage exceptions for the AP team's attention. The audit trail forks cleanly because capture and approval happen in different systems with their own logs — capture logs at the extraction layer, workflow and approval logs in Concur — which auditors generally find easier to walk than a single-system trail where capture verification and approval steps share UI.

The trade-off is honest: the upstream layer is a system to procure, integrate, and operate. The integration is a defined Concur import path — depending on the tenant's configured paths, that means CSV or XML upload via Concur's import UI, an API push to the import endpoint, or a scheduled pickup from a watched location — but it is still an integration to design, monitor, and maintain. The case for the layer is strongest where line-item-heavy invoices are a meaningful share of total volume. For header-mostly invoice profiles (high-volume utility bills with minimal line detail, simple service invoices, repeating subscriptions), the three-mode choice inside Concur is usually the right scope; the upstream-extraction architecture earns its complexity at line-item depth and falls below the line at header simplicity.

The framing that matters is keep-Concur-and-supplement, not replace-Concur. The reader of this article is staying on Concur — Concur Invoice is the AP platform of record, the workflow and approval and payment infrastructure live there, the supplier master and the integration to the GL and the bank live there. The upstream extraction layer offloads one specific weakness of Concur's native capture for the invoice categories where that weakness costs the most, without disturbing Concur's role anywhere else. This is not a Concur replacement.

The line-item OCR pattern is a platform-OCR pattern rather than a Concur-specific defect, and seeing the same shape on adjacent platforms confirms that. The Bill.com Invoice Coding Agent line-item extraction limits play out the same architectural argument on a different platform — the AP tool is good at workflow and pricing-tier-appropriate at header capture, and an upstream extraction layer earns its place where line items are the value-density of the invoice. The pattern generalises across enterprise and mid-market AP tools because the underlying recognition challenge — long, variable, high-precision line-item tables on inconsistent supplier templates — generalises across them.

A note on search intent. Searches for "Concur Invoice OCR alternatives" mostly land on replace-Concur content from vendors that compete with SAP at the platform level. That is a different decision and not what this section is about. The architectural option here is the layer that sits in front of Concur for line-item-heavy capture, not the platform that sits in place of Concur. A firm whose AP function is built around Concur Invoice and wants to address its line-item OCR weakness without switching platforms is the firm this fourth path is for. A firm considering platform replacement is reading a different article.

---

Choosing Among the Four: A Decision Logic Keyed to Your Firm

There is no single right answer. The decision is keyed to your firm's specific shape across five variables, and the right configuration is the one that lands all five honestly rather than the one a partner pitch or a SERP-favourite blog declares best.

Monthly invoice volume. Below roughly 1,000 invoices per month, the absolute Managed fee is small in dollar terms and the FTE cost of Client Managed verification is also modest. Either mode works; the choice tips on the other variables. From 1,000 to roughly 10,000 invoices per month is where the comparison gets sharp — the Managed fee scales meaningfully, and the verifier-team requirement starts to demand dedicated headcount rather than a shared task on someone's plate. Above 10,000 invoices per month, both pure modes are operationally serious, and the conversation usually moves to Dual Support routing or upstream extraction rather than one mode end-to-end.

AP team capacity and shape. A firm with an established AP team already sized for verification work — verifiers in role, supplier-template familiarity built up, manager bandwidth to run the function — is well positioned for Client Managed. A firm running a lean AP function, transitioning from a manual AP shop, or scaling through a period of supplier-base churn is often better served by Managed in the short term, then re-evaluating as the team matures. Buying time for team maturation is a legitimate reason to pay the Managed fee for a defined period.

Supplier-base complexity. Homogeneous, single-region, single-language supplier bases are exactly the conditions where Client Managed verifier training pays off and OCR tuning compounds. Multi-language, multi-region, multi-currency supplier bases tilt the balance the other way — verifier-staff training cost scales with complexity, and the geographic-split pattern of Dual Support often becomes more attractive than either pure mode. The supplier-base shape is the variable buyers most often underweight in their own analysis.

Line-item complexity. Header-mostly invoice profiles — utility bills, simple service charges, recurring subscriptions, fixed-price professional services — fit any of the three Concur-native modes cleanly. The verification work is light because the OCR engine handles header fields well. Line-item-heavy profiles are where the upstream-extraction fourth path earns its complexity. The threshold question is what share of monthly volume comes through line-item-heavy supplier categories. Below something like a quarter, the architectural cost of the upstream layer is hard to justify against the in-Concur modes; above something like half, the line-item correction load makes the fourth path the most operationally rational answer.

Audit posture. SOX-relevant firms with strict segregation-of-duties expectations have two coherent paths. One is Managed, where the external-verifier separation is clean by construction and the audit testing is straightforward. The other is Client Managed with documented compensating controls — the dual-review-on-corrections and rotation-of-verification-duty patterns Section 5 walked through — where the controls are real operational cost but the audit posture is defensible. Firms without SOX exposure have more freedom on this dimension and can let the other four variables drive the call.

A few representative shapes the patterns combine into:

• High-volume manufacturing firm with line-item-heavy invoices and a mature AP team. Upstream extraction into Concur, leaving Concur's native modes for the residual header-only invoice flow. The line-item correction load is the dominant operational cost, and the upstream layer addresses it directly.
• Global enterprise with a multi-region supplier base and SOX exposure. Dual Support with geographic routing — Managed for the regions where local-language supplier complexity is highest, Client Managed for the home region with documented compensating controls. The hybrid earns its complexity because the supplier-base profile is genuinely heterogeneous.
• Mid-market firm on Concur Invoice with steady mid-volume header-mostly invoices and a small AP team. Managed, for the labour transfer. The Managed fee is the price of not having to staff a verification function the firm doesn't otherwise need.
• Mid-market firm with the same volume profile but a larger established AP team. Client Managed for cost control, with documented compensating controls if SOX applies. The team is already sized for the work.

The one durable instruction across all of these: get the Managed quote with the variables unbundled (per Section 2), model the fully-loaded Client Managed cost honestly without the marginal-cost trap (per Section 3), and put the upstream-extraction architecture on the table for line-item-heavy volumes (per Section 6). The right answer is the one defensible against those three numbers at your firm's actual volume and supplier-base profile.