Tier 1 · Services 1.4–1.6

Your event logs already know where the process breaks.

Process mining reads them. Before anyone argues about how the process is supposed to work, the data shows how it actually runs — every variant, every wait, every loop back, priced.

Short answer

Process mining reconstructs how a process really runs from the timestamped records your systems already write. Orqive mines Business Central and Dynamics 365 event data — change logs, posting and document history, approval entries, Dataverse audit — to expose bottlenecks, rework loops, and conformance breaks, then quantifies what each one costs.

Built on Power Automate Process Mining, Process Advisor, Dataverse, and Power BI or Fabric. Output is always delivered inside the stack the mining runs on — mine in Power Automate Process Mining and the maps, analysis, and monitoring land in your own Power Automate and Fabric tenant. Two to five weeks for a first process.

Three services

Discover the real process, price the waste, then redesign it.

1.4 / Discover

Process Discovery & Mining

We build an event log from your live systems and mine it, so the process map is evidence rather than a workshop artefact — including the variants nobody documented.

Event-log mining Task mining Conformance checking

Deliverable: as-is process map + variant analysis

1.5 / Analyse

Bottleneck & Cost Analysis

Where the process stalls, how long each variant takes, what the waiting and rework cost per month — and which steps are stable and expensive enough to hand to an agent first.

Cycle-time analysis Cost of inefficiency Automation scoring

Deliverable: priced bottleneck register + ranked backlog

1.6 / Optimise

Process Optimization & Monitoring

The redesign the evidence argues for — steps removed, approvals collapsed, exceptions codified — then dashboards that keep measuring cycle time, rework, and conformance after the change.

Process redesign Continuous monitoring

Deliverable: to-be design + Power BI / Fabric monitoring

What it typically finds

Four patterns show up in almost every mid-market process.

None of them are visible in a standard report, because a report measures the outcome rather than the path taken to reach it.

Variant sprawl

One documented process, forty real paths. Most volume runs through a handful; the long tail is where errors and manual effort concentrate.

Rework loops

Records that go back a step — re-approved, re-priced, re-keyed. Each loop is invisible in cycle-time averages and expensive in practice.

Waiting, not working

Most elapsed time in a mid-market process is queue time between steps, not processing time inside them. Mining separates the two.

Conformance breaks

Steps skipped, thresholds bypassed, approvals granted out of sequence. The same evidence a governance review needs.

How it runs

Four steps, two to five weeks

The long pole is almost always the event data, not the analysis. Where audit logging and change tracking are already switched on, the whole thing compresses.

Process mining sits inside Tier 1, so its output feeds directly into the agent blueprint rather than sitting in a separate report.

Step 01 · Scope

Pick the process and define the case

One process, one case identifier — a sales order, a purchase invoice, a requisition. Getting the case definition right is what makes the resulting map readable.

Step 02 · Extract

Build the event log

Change logs, document and posting history, approval entries, job queue records, and Dataverse audit data become one table of case, activity, timestamp, and actor. Task mining recordings fill in the desktop work no system records.

Step 03 · Mine

Map, compare, and price

Variants, cycle time, waiting time, rework rate, and conformance against the designed process — then a cost per bottleneck using your own labour and volume figures.

Step 04 · Decide

Redesign, then rank for automation

A redesign workshop against the evidence, and an automation opportunity score per step — volume, stability, cost, and rule clarity — that becomes the input to the agent blueprint.

Mined with

Power Automate Process Mining Process Advisor Dataverse Power BI Microsoft Fabric Business Central change log D365 audit & telemetry

Or in the mining tool you already license

SAP Signavio Celonis Mavim

Plenty of mid-market organisations already carry a Signavio, Celonis, or Mavim licence — often bought for a past transformation programme and under-used since. Where that is the case we mine in it rather than standing up a parallel tool, and the Business Central and Dynamics 365 event log we build feeds that platform instead. Output is delivered in the stack the mining ran on, in your own tenant: mine in Power Automate Process Mining and the process maps, analysis, and monitoring live in your Power Automate and Fabric environment; mine in Celonis and they live in Celonis. The engagement follows your process, not a product roadmap.

FAQ

Process mining questions

On what it is, what it needs from your systems, and how it relates to agents.

What is process mining?

Process mining reconstructs how a business process actually runs by reading the timestamped event records your systems already write. Instead of interviewing people about how the process is supposed to work, it shows every real variant, how often each occurs, how long each step takes, and where work loops back — as evidence rather than opinion.

Can you do process mining on Business Central?

Yes. Business Central and Dynamics 365 write the event data process mining needs — change logs, document and posting history, approval entries, job queue records, and Dataverse audit data. We extract those into an event log keyed by case, activity, and timestamp, then mine it in Power Automate Process Mining or Fabric.

How is process mining different from a Power BI dashboard?

A dashboard reports on outcomes you already chose to measure. Process mining reconstructs the sequence of events, so it surfaces variants, rework loops, and waiting time nobody had thought to measure — including the process paths that only exist because of workarounds.

Which tools does Orqive use for process mining?

Primarily Power Automate Process Mining and Process Advisor, with Dataverse for the event store and Power BI or Microsoft Fabric for continuous monitoring. Where a client already licenses SAP Signavio, Celonis, or Mavim, we mine in that tool instead rather than duplicating the licence. Either way the output is delivered in your own tenant, in the same stack the mining ran on.

Do we need process mining before deploying AI agents?

Not always, but it removes the guesswork. Mining gives an evidence-based ranking of which steps cost the most and which are stable enough to automate, so the agent blueprint targets the expensive, repetitive paths rather than the ones that are simply most visible.

How long does a process mining engagement take?

Two to five weeks for a first process, depending on how accessible the event data is. Extraction and event-log construction is usually the longest part; analysis and the optimization workshop follow within days once the log is clean.

Start with the readiness grade

Find out whether your data can carry a mining project yet.

The OARI grades Business Process Readiness and Data & Integration among its six dimensions — the two that decide whether an event log can be built cleanly. Eighteen questions, about ten minutes, no account.

Take the free assessment →

How it worksHow an agent gets built in 90 daysFive phases, what you receive, and who does what. ComparisonAgentic AI vs CopilotWhere Copilot stops and an agent starts, side by side. GlossaryMining terms, definedEvent log, task mining, conformance checking, variant.