Your ERP holds years of complete operational history: every approved purchase order, every invoice entered, every stock discrepancy corrected. Yet your dashboards show only aggregates. They tell you how many invoices were processed — not how many took a detour, required three approvals instead of one, or sat in a manager’s inbox for eight days before being reopened.
That is exactly what process mining reveals. By reading your ERP’s event logs directly, it reconstructs the real path of every transaction, compares it to the intended process, and maps the deviations, bottlenecks, and workarounds that cost time and money every day.
This guide explains what process mining means in an ERP context, how to choose between the three market leaders (Celonis, SAP Signavio, IBM), which processes to audit first, and how to structure a first project in six weeks.
What Is Process Mining in an ERP Context?
The Difference from Classic BI
A Business Intelligence tool aggregates data: total sales, on-time delivery rate, average payment delay. It answers “how many?” but not “how?”. You know that 15% of your orders are delivered late. You do not know where in the process the delay occurs, or whether it is always the same type of order involved.
Process mining answers “how?”. It reconstructs real workflows from the traces left in the ERP database: every document status change, every actor, every timestamp. The result is a map of the real process, layered over the intended process. Deviations become visible, measurable, and prioritisable.
How the Tool Reads ERP Event Logs
In SAP, every management document (purchase order, invoice, production order) generates an event trail stored in dedicated tables. In SAP ECC and S/4HANA, tables CDHDR and CDPOS record every change. In Oracle Cloud ERP, Microsoft Dynamics 365, or NetSuite, equivalent mechanisms journal transactional activity.
Process mining tools connect to these tables via specialised extractors without modifying any ERP configuration. They do not touch settings, do not affect end users, and require no ERP-side development. The extraction is read-only. This is a decisive advantage over traditional process redesign projects that demand workshops, interviews, and months of manual mapping.
The 3 Market Leaders: Celonis, SAP Signavio and IBM Process Mining
Celonis: The Multi-ERP Independent
Celonis is the process mining market leader by revenue and customer base. Its platform offers more than 100 process connectors covering all major ERP systems (SAP, Oracle, Microsoft Dynamics, Salesforce, ServiceNow) as well as supplementary data sources such as SQL databases, Azure, and PostgreSQL. More than 600 teams worldwide use Celonis’s SAP extractor, which supports SAP ECC and S/4HANA on HANA, Oracle, IBM, and Microsoft backends.
Its primary strength is independence from ERP vendors. An operations director whose application landscape combines SAP for finance, Dynamics for procurement, and Salesforce for CRM can plug Celonis into all three systems and map cross-ERP processes that are invisible in any of the native tools.
Worth noting: a legal dispute emerged in 2024–2025 between SAP and Celonis over data access rights for shared customers. A provisional agreement was reached so that SAP would not disrupt customers’ access to their own data via Celonis during proceedings.
Positioning: Multi-ERP, mid-market and enterprise, large-scale programmes.
SAP Signavio: Native S/4HANA, Zero Friction
SAP acquired Signavio in March 2021 for approximately €949 million according to SAP’s SEC filing (SAP Form 6-K, SEC.gov), widely described as a “billion-dollar” acquisition in the trade press. Signavio is now integrated into SAP Business Process Intelligence, available on the SAP Business Technology Platform (BTP).
For SAP customers, the integration is native: SAP Signavio accesses S/4HANA data directly without a third-party extractor, with pre-configured process models for Order-to-Cash, Purchase-to-Pay, and Record-to-Report cycles. Signavio’s process repository also links real processes to the target processes defined during design, simplifying post-deployment governance.
The limitation is straightforward: SAP Signavio is built for and by the SAP ecosystem. If your ERP is not SAP, the added value is considerably lower.
Positioning: SAP S/4HANA customers, RISE migration projects, continuous post-go-live governance.
IBM Process Mining: Accessible Option, ERP and CRM Integration
IBM Process Mining is the third significant player in the market. The platform integrates with major ERP sources (SAP, Oracle Cloud Financials) and CRM systems, and positions itself on accessibility and integration with the IBM Automation ecosystem. It is also available on AWS Marketplace.
IBM Process Mining fits organisations already working with the IBM stack (Automation, Cognos Analytics, watsonx) that want a solution consistent with their existing infrastructure without managing a specialist third-party vendor.
Positioning: IBM ecosystem, regulated industries (banking, insurance, healthcare), integration with IBM Automation.
Quick Decision Matrix
| Criterion | Celonis | SAP Signavio | IBM Process Mining |
|---|---|---|---|
| Multi-ERP | Yes (100+ connectors) | Limited (optimised for SAP) | Yes (SAP, Oracle, CRM) |
| Native SAP integration | Yes (certified extractors) | Yes (native BTP) | Yes (connectors) |
| Integrated process governance | Via Celonis EMS | Yes (BPM repository) | Via IBM Automation |
| Target profile | Multi-ERP mid-market / enterprise | SAP customers | IBM ecosystem |
| Implementation complexity | Medium | Low for pure SAP | Medium |
The 5 ERP Processes to Audit First
Order-to-Cash (O2C): Where Are You Losing Time Between Order and Payment?
The O2C cycle traces a customer order from creation to cash receipt. Process mining identifies orders that pass through more approval steps than intended, credit-check hold-ups, delays between delivery and invoice issuance, and manual chasing that could have been automated.
The key metric to watch is DSO (Days Sales Outstanding): the average number of days between invoicing and payment. Reducing DSO by even a few days delivers a direct, measurable cash flow benefit.
Purchase-to-Pay (P2P): Duplicates, Delays, and Non-Compliance
The P2P cycle covers procurement from requisition to supplier payment. This is process mining’s natural territory: discrepancies between purchase order, goods receipt, and invoice generate disputes, payment delays, and supplier penalties. Automatic detection of invoices without matching purchase orders, duplicate payments, or bypassed approvals is among the most documented use cases.
Record-to-Report (R2R): Accelerating the Monthly Close
The monthly close is often seen as an unavoidable constraint. Process mining reframes it: it identifies the 20% of manual journal entries that cause 80% of close delays — a pattern frequently reported by finance teams that have run this type of analysis. Reprocessing patterns, pending validations, and late arbitrations become visible, and therefore addressable.
Hire-to-Retire (H2R): Streamlining HR Processes
HR processes are often fragmented across the HRIS, payroll, and ERP. Process mining on the H2R cycle maps onboarding delays, hold-ups in leave-approval workflows, and discrepancies between HR data and system access rights. For organisations with high turnover or in intensive recruitment phases, fluency gains come quickly.
Incident-to-Resolution (I2R): Monitoring Post-Go-Live ERP Support
Often overlooked, this cycle is critical during the 12 to 24 months following an ERP deployment. Process mining on support tickets identifies which processes generate the most incidents, which user types most frequently bypass workflows, and which modules need configuration review. It is a powerful tool for the ERP post-deployment centre of excellence.
Case Study: Launching a Process Mining Project on SAP S/4HANA in 6 Weeks
This timeline applies to a proof-of-concept project on a single target process (for example, the Purchase-to-Pay cycle) using SAP Signavio or Celonis.
Weeks 1–2: Event Log Extraction and Connector Setup
The project team (two to three people: a process mining consultant, an SAP Basis administrator for access, and a business key user) configures the extractor on the SAP environment. Target tables are identified based on the chosen process (for P2P on SAP: tables EKKO, EKPO, RBKP, BKPF). An initial extraction is run on 12 to 24 months of historical data.
No changes to SAP configuration are required. The extraction is read-only and has no impact on end users.
Weeks 3–4: Modelling and Deviation Identification
The process mining platform automatically generates a real-process map from the extracted event logs. The consultant analyses variants: how many different paths does a supplier invoice actually take? Which variants are most frequent? Which are the most time-costly?
A two-hour workshop with the finance team qualifies the deviations: which are legitimate exceptions, which are workarounds to correct, and which point to configuration issues.
Weeks 5–6: Action Plan and KPI Baseline
The team establishes baseline indicators before any action is taken: average time per process variant, conformance rate to the ideal flow, number of superfluous approval steps. These metrics form the baseline that will measure the impact of corrective actions in the months ahead.
The final deliverable is a prioritised action plan: two or three high-impact actions on the audited scope, with required resources, an owner, and a target date.
Expected ROI and Key Metrics
What Process Mining Lets You Quantify
Unlike traditional process redesign projects, process mining produces objective metrics before any action is taken. Gains typically identified in a first audit include:
- DSO reduction on the O2C cycle: identifying recurring bottlenecks in the invoicing flow enables targeted action (automated chasing, approval simplification) whose impact on DSO is measurable in days.
- P2P automation rate: on the Purchase-to-Pay cycle, the target is typically 70–80% of invoices processed without manual intervention. Process mining identifies which invoice types cannot yet be automated and why.
- Monthly close acceleration: identifying recurring manual journal entries enables converting them into automatic rules in the ERP, reducing the number of close days.
For a deeper look at how process mining and automation complement each other, see our guide on RPA and ERP: automating repetitive tasks.
What Process Mining Does Not Replace
Process mining identifies and quantifies problems. It does not solve them. The real value lies in the combination: deviation analysis via process mining, then action through RPA, ERP reconfiguration, or user training. Without a structured action plan, the process map remains a diagnosis without follow-through.
Getting Started: Requirements, Budget, and Team
Technical Prerequisites
- Read-only extraction rights on ERP transaction tables (to be confirmed with your SAP Basis administrator or equivalent)
- Access to a test or sandbox environment for connection testing
- At least 12 months of transaction history on the target process (24 months preferred to detect seasonal patterns)
Typical Project Team
A process mining proof of concept typically mobilises two to three people:
- A process mining consultant (external or internal CoE) for extraction, modelling, and analysis
- An ERP administrator for access configuration and validation of source tables
- A business key user (finance, procurement, or operations depending on the process) to qualify deviations
Indicative Budget
Annual licence costs for process mining platforms vary by vendor and scope covered. According to published benchmarks (KYP.ai Process Mining Software Comparison 2026), Celonis premium connectors are priced between $10,000 and $30,000 per connector per year. SAP Signavio integrated solutions are priced through BTP based on data volume and activated modules. IBM Process Mining offers per-process or per-volume pricing models.
For a proof of concept on a single process, budget between €20,000 and €60,000 all-in (POC licences, consulting, extraction, analysis workshops). This produces a deviation map and a prioritised action plan.
To place this approach within a broader governance framework, see our article on ERP current-state audit before migration: process mining can usefully complement a classic ERP diagnostic.
Taking Action
To test a process improvement hypothesis, run a 6-week proof of concept on a target process (Purchase-to-Pay or Order-to-Cash). Typical budget: €20,000 to €40,000. Output: a real-process map, deviations quantified by time and frequency, and a prioritised action plan with baseline KPIs. Not an impressionistic diagnosis, not a sales pitch — data extracted from your own ERP.
To go further:
- Our guide ERP Audit: Assessing Your Current System Before Migrating to objectify the overall diagnostic
- Our article RPA and ERP: Automating Repetitive Tasks to understand how to combine process mining and automation
- Our article on ERP Post-Deployment Governance and Centre of Excellence to integrate process mining into your ongoing supervision
