R2R Journal Entries Automation at a Global BPM Shared Services Provider — 99.99% STP Accuracy, 45% AHT Reduction
DocGenie + Workflow Planner + Workflow Automator automating manual and accrual journal entries from email, Excel and PDF into SAP — with exception-handling and duplicate detection in a single workflow.
The challenge
The provider — a global business-process-outsourcing firm running record-to-report shared-services centres for multiple enterprise customers — was operating a manual journal-entry workflow that absorbed substantial workforce capacity. Inbound journal-entry requests arrived via email (the bulk of volume), Excel attachments (the typical format for the recurring journals), and PDF attachments (the typical format for the supporting-documentation-led journals). Each request was read by a human accountant, who then identified the relevant account codes, made the journal entry in SAP, attached the supporting documentation and emailed the requestor a confirmation.
Accruals were the more painful component. Month-end and quarter-end accruals required the information-providers across the customer's business to submit accrual estimates with back-up; the accountants then compared against the prior period, sought clarifications through informal email loops, compiled the final accruals, and uploaded the accrual journals to SAP. The end-to-end accrual cycle absorbed several days of senior-accountant time per closing period, and the back-and-forth clarification loops introduced delays that compressed the rest of the closing workflow.
Operational risk was a meaningful concern beyond the cost. Duplicate journal entries (the same journal posted twice through the manual workflow's edge cases) had been a recurring audit finding. Period-end posting cutoffs were missed when the closing-workflow compression caught up with the accountants. The provider's customers had begun to ask whether the manual workflow was structurally compatible with their growth ambitions.
The approach
MindMap deployed an R2R automation platform composed of DocuMage (Dm) as the email-and-attachment-intelligence layer, Workflow Planner (Wp) for the per-journal-type workflow orchestration, Workflow Automator (Wa) for the SAP-posting integration, and Anomaly Detector (Ad) for the duplicate-entry monitoring.
Phase one was the journal-source cataloguing. We catalogued every inbound source of journal entries across the customer base — the recurring monthly journals (the standard accrual-reversal-and-re-recognition entries), the period-end allocations (the cost-allocation journals), the recurring same-amount journals (the standard depreciation, the standard provisions), the variable-amount recurring journals (the variable-cost accruals, the revenue-recognition adjustments), and the one-off journals (the corrections, the reclassifications). The catalogue identified roughly 80% of journal-volume as fully-automatable.
Phase two was the email-and-attachment classification workflow. An email-auto-forward bot runs against the customers' inbound journal-request mailboxes, identifies emails that should generate a journal entry through a rules-and-LLM-driven classification approach, and forwards them to a centralised processing mailbox. From the centralised mailbox, DocGenie reads the email content and attachments to extract the structured journal-entry data — the account codes, the cost-centres, the amounts, the period assignments, and the narrative-text.
Phase three was the SAP-posting workflow. The extracted journal data flows through Workflow Planner, which validates the journal against the SAP chart-of-accounts, the cost-centre master, and the period-controls. Validated journals are posted to SAP through Workflow Automator's SAP integration with the supporting documentation auto-attached. Posting confirmations and exception notifications are sent to the original requestor.
Phase four was the accruals automation layer. The accruals-automation workflow combines an information-provider-facing guided interface (with automatic reminders based on the closing-calendar) and a back-end automation that compares the submitted accruals against the prior period, surfaces variances above threshold for review, compiles the consolidated accruals file and posts the accrual journals to SAP. Anomaly Detector runs continuously against the posted-journals stream to identify potential duplicates and flag them for review.
The pre-built building blocks
Rather than commission a ground-up build, the engagement leaned on MindMap's pre-built accelerator library — production-tested components that compress what would otherwise be a six-to-nine-month build into weeks.
DocuMage
Email-and-attachment journal extraction with LLM classification
Workflow Planner
Per-journal-type validation-and-posting workflow orchestration
Workflow Automator
SAP-posting integration with document-attachment workflow
Anomaly Detector
Duplicate-entry monitoring across the posted-journals stream
The architecture
The platform runs on the provider's managed cloud environment with per-customer tenant isolation. Each customer's journal-processing happens in its own tenant with the appropriate access-controls and audit-trail handling per the customer's data-handling commitments.
DocGenie's email-and-attachment classification combines a deterministic rules layer (the recurring-journal patterns are matched against known templates) and an LLM-driven classification layer (the variable-format journal requests are classified through structured prompting against the journal-type catalogue). The classification produces both the journal-type identification and the extracted journal-data fields.
Workflow Planner's per-journal-type orchestration encodes the validation-and-posting workflow per journal-type. Standard recurring journals flow through a streamlined validation-only-then-post path; variable-amount accrual journals flow through a more involved validation path with prior-period comparison and variance-surfacing; correction-and-reclassification journals flow through an enhanced review path with mandatory human approval.
Workflow Automator's SAP integration uses SAP's standard inbound BAPI and IDoc interfaces where the provider's customers have configured them; for customers where direct integration was not feasible, the integration uses screen-level automation against the SAP GUI with appropriate transaction-level recovery. The integration includes the document-attachment workflow that preserves the supporting documentation against the posted journal in SAP's document-management module.
Anomaly Detector runs the duplicate-monitoring workflow against the posted-journals stream. The detection uses a combination of exact-match rules (same amount, same account, same date), fuzzy-match rules (similar amounts within tolerance, similar narrative-text) and pattern-match rules (sequences of journals that resemble historical duplicate patterns). Flagged candidates route to a review workflow where an accountant confirms or dismisses the duplicate concern.
Reporting-and-management dashboards provide the per-customer journal-volume statistics, the per-journal-type processing metrics and the exception-and-duplicate trend analysis. The audit trail captures every journal-lifecycle event with the full context preserved for the customer's regulatory and internal-audit requirements.
The numbers behind the story
Straight-through-processing accuracy on automated journal entries has stabilised at 99.99% on a rolling measurement, meeting the provider's customer-facing SLA commitments with substantial headroom. The 0.01% exception rate is dominated by genuinely-ambiguous cases that genuinely require human judgement.
Average handling time on automated journal types has dropped 25-45% depending on the journal complexity. Standard recurring journals — the bulk of volume — process in seconds; the previous workflow had absorbed several minutes per journal. The accrual workflow's end-to-end cycle has compressed from several days to under one day, with the closing-workflow ripple effect reducing pressure on the downstream consolidation-and-reporting workflow.
The R2R workforce has been substantially redirected. The journal-entry-keying capacity has been absorbed into closing-workflow oversight, exception-handling and the genuine-value-add R2R analytical work (variance investigations, account reconciliations, ad-hoc analyses for the customer's business teams) that the previous keying-load had crowded out.
Duplicate-entry incidents have been substantially reduced. The Anomaly Detector workflow has caught several hundred potential-duplicate cases across the customer base over the operational period; the absence of duplicate-entry findings in the customers' recent audits validates the workflow's effectiveness.
Customer satisfaction has improved measurably. The provider's customer-satisfaction surveys on the R2R workflow have shifted from being a structural area of concern to being a strength. Several customers have expanded their scope-of-engagement on the basis of the automated R2R workflow's demonstrated performance.
“Our R2R workflow had reached the structural limits of a manually-keyed model. MindMap delivered ninety-nine-point-nine-nine per cent straight-through accuracy on automated journals, twenty-five to forty-five per cent AHT reduction across journal types, and the duplicate-entry findings that had been a recurring audit concern have been eliminated. Our customers see the R2R workflow as a strength now rather than as a structural concern.”— Chief Operating Officer· Global BPM Shared Services Provider
Why MindMap was chosen
The provider had evaluated two specialist R2R-automation vendors. Both had strong RPA-tooling capabilities for the SAP-posting workflow but limited LLM-driven email-and-attachment understanding capability, which was the structural challenge in the variable-format journal-request workflow.
MindMap's accelerator-composition approach — bringing DocGenie's LLM-driven extraction together with Workflow Planner, Workflow Automator and Anomaly Detector around the SAP estate — was the structural differentiator. We could demonstrate the email-and-attachment classification working on the provider's actual customer-journal-request samples within a two-week proof-of-concept.
Our embedded R2R-operations expertise on the delivery team (two former R2R-shared-services leads and a former CPA from a Big-4 accounting firm) was the third factor. The provider's COO felt that the team understood the operational reality of running a multi-customer R2R workflow at scale, rather than approaching it as an abstract automation engineering problem.
Related deployments
Background Checks at Scale
A unified RPA + IDP stack handling identity, employment, education, criminal and credit checks across 40 countries — 3,000 checks per day, straight-through.
RPA-to-Agentic Migration
A migration from 38,000 brittle RPA bots to 240 agentic workflows — handling more volume at a fraction of the maintenance burden.
Big-4 AI Delivery Practice
MindMap's platform underpins the Big-4 firm's AI delivery practice — 2,400 consultants delivering AI engagements to the firm's enterprise customer base.
Want an outcome like this?
Start with a 2-week AI Readiness Sprint. We deliver a prioritised use-case backlog and business case grounded in what's actually buildable with our accelerator library.