Business Process Discovery: Methods, Tools and Platforms Compared

A guide for enterprise transformation and operations teams to business process discovery: what it is, the three main methods, how AI organizational discovery platforms compare, a worked example, and where to go next for each situation.

June 25, 2026Updated November 2, 202610 min read
business process discoverywhat is process discoveryprocess discovery example

The short answer

Business process discovery is the work of finding out how a process actually runs today, across the people, systems, handoffs and exceptions involved, before anyone redesigns, automates or replaces it. Horizon is an AI organizational discovery platform that does this by interviewing employees at scale and turning their answers into process maps and ranked improvement opportunities, with no event logs required.

Every enterprise process has two versions. The official one lives in diagrams, policies and system configurations. The real one includes the side spreadsheets, informal approvals, duplicated checks and team-specific shortcuts that determine how work gets done. Discovery closes the gap between the two, so the transformation that follows is built on the process people actually use.

Key takeaways

What is process discovery?

Process discovery is the structured effort to identify, document and analyze how a business process runs in practice. It answers practical questions:

Discovery is broader than documentation. Documentation records what someone already knows. Discovery finds what nobody has written down, which is usually where delay, cost and risk accumulate.

Four related terms often get mixed up:

TermWhat it meansBest used when
Business process discoveryFinding how a process actually runs across people, systems, decisions and exceptionsYou need a trustworthy current state before improving or transforming
Process mappingDrawing the steps, handoffs and decisions of a processYou need a shared visual of a process. See process mapping tools
Process miningReconstructing a process from system event logsThe process runs mostly inside systems that keep clean logs
Task miningCapturing user actions on screens to find repetitive desktop workYou need to see work between applications. See task mining software

What are the three main process discovery methods?

1. Workshops and stakeholder interviews

A facilitator brings the people involved together, in a room or one by one, and builds the process from what they describe. This method captures decision logic, local exceptions and handoff friction that no system records.

Its limit is the sample. A workshop reflects the people invited, and senior voices tend to describe the official process. The frontline staff who handle exceptions are often missing, and each additional interview adds consultant hours, so coverage stays small.

2. Process mining

Process mining tools read event logs from systems of record such as ERP, CRM and ticketing platforms. Each event needs a case ID, an activity name and a timestamp. From those, the tool reconstructs every path a case actually took, with variants, wait times and rework loops.

Its limit is the log. Work that happens in email, spreadsheets, chat approvals or between systems without a shared case ID is invisible to it, and the log shows what happened without the reason. For options in this category, see Celonis alternatives.

3. AI-led employee interviews

An AI agent interviews everyone in scope at the same time, with questions designed for each role and follow-up on each answer. The synthesis groups what hundreds of people described into findings, each linked to the conversations behind it.

This method combines the depth of an interview with the coverage of a survey, and it needs no event logs. Its limit is precision on transaction volumes and timings, which system logs measure better. For how the method works in detail, see AI employee interviews at scale.

AI organizational discovery platforms compared

AI organizational discovery platforms use AI to find how work happens across an organization, and they take four approaches. Each one reads a different layer of the same operation.

ApproachWhat it readsWhat it missesRepresentative toolsBest fit
Consulting-led discoveryAnything a person can describe to a consultantTeams outside the sample; changes after the reportTraditional consulting firmsA narrow scope with the right experts available
Process miningPaths, variants and timings inside systemsWork outside systems, and reasonsCelonis, SAP Signavio, Power Automate Process MiningHigh-volume processes with clean event logs
Work observation and task miningOn-screen activity across applicationsReasoning, off-screen work, judgmentTask mining and work-observation platformsRepetitive desktop work across tools
AI-led employee interviewsExceptions, informal approvals, reasons, improvement ideas, at scaleExact transaction volumes and timingsHorizonProcesses that cross teams, tools and countries

The four approaches complement each other. Systems record transactions, screens record behavior, and people hold the reasons. The practical question is which of those layers holds the answer you are missing.

Which process discovery platforms should an enterprise compare?

Start from the question you need answered, then compare platforms on six points:

  1. Evidence source. Does the platform read system logs, screen activity or people's answers? Match it to where the work happens.
  2. Coverage. How many people, teams and countries can it reach in one round, and how long does that take?
  3. Traceability. Can every finding be traced back to the evidence behind it?
  4. Output. Does it stop at a map, or does it produce ranked opportunities and business cases with owners?
  5. Data handling and consent. How are answers stored, reported and protected? See employee consent in process discovery.
  6. Repeatability. What does a second round cost, and does the view stay current as the organization changes?

The sixth point separates a project from a capability. A platform that makes the second round cheap turns discovery into a habit; see continuous discovery in the enterprise.

What is a process discovery example?

Trafilea, a tech-driven ecommerce group that builds direct-to-consumer brands, operates fully remote with more than 400 employees across several countries. Its process documentation had not been updated in more than two years, and teams ran on tribal knowledge. Work spread across seven or more tools with parallel updates and copied links.

Before discoveryAfter discovery
Process documentationNot updated in more than two yearsStandardized documentation ready for the internal knowledge base, audits and due diligence
Status trackingDuplicated across two project tools, at 80 to 100 requests a month218 hours a month of duplicate tracking quantified, and a decision to consolidate into one source of truth
Creator pipelineSeven or more tools with parallel updates and copied linksIntake and approvals standardized into a single workflow
ReworkUp to three reshoots per creator, about a week of delay eachReshoots capped and feedback service levels set
Time to mapAbout a year, by the process specialist's own estimate, to map manually43 asynchronous interviews across two teams in two weeks

The round produced more than 10 actionable findings, identified 50 to 90% automation potential across key workflows and eliminated more than 129 hours of discovery work compared with traditional process mapping. Day-one ROI was 310%.

That is one engagement under specific conditions, and it shows the typical shape of discovery: the decisions that came out of it were structural, and they would have been hard to reach from any single tool's data.

How do you run business process discovery?

Ten steps cover a discovery effort on a process important enough to improve, automate or redesign:

  1. Start with the business question. "Why does onboarding take too long?" sets a better scope than "map onboarding."
  2. Define the boundaries. Set where the process starts and ends.
  3. Identify the evidence sources. People who do the work, approvers, downstream teams, systems, spreadsheets and documents.
  4. Capture the current state. Use logs for system-heavy steps and interviews for everything else.
  5. Map the main flow and the variants. Include exception paths, rework loops and manual workarounds.
  6. Find the root causes. A delay is a symptom; the cause may be unclear ownership, missing data or a duplicated approval.
  7. Size the impact. Estimate frequency, time cost, business value, evidence strength and fix complexity.
  8. Prioritize initiatives. Turn findings into a short list with a named problem, change, impact, owner and next step.
  9. Validate with the teams doing the work. Confirm the map and the fixes before locking the roadmap.
  10. Keep discovery current. Schedule the next round before the first one ends.

When the discovery precedes a system change, the order of these steps matters even more; see process discovery before a system migration. When the goal is automation, see automation vs process discovery for why discovery comes first.

Process discovery questions to ask

Use these in workshops, one-to-one interviews or AI-led conversations.

Flow: What triggers the process? Where does it usually slow down? Where does work loop back?

Ownership: Who owns it end to end? Where does ownership change, and which handoffs cause confusion?

Systems and data: Which data gets entered more than once? Which system do people trust as the source of truth? Where do side spreadsheets exist?

Exceptions: What are the most common exceptions? Which approvals get skipped, duplicated or escalated informally?

Improvement: Which step would you remove if you could? Which issue creates the most avoidable rework?

Discovery guides by situation

This hub connects Horizon's guides on process and organizational discovery. Start with the one closest to your situation:

See how one of your processes really runs, in one week. Run Horizon's one-week challenge: AI interviews with everyone who touches the process, an AS-IS map and a ranked list of what to fix first. Start the one-week challenge

Where Horizon fits

Horizon is an AI organizational discovery platform built for the human layer of process discovery, the part system logs cannot see.

Discovery Cycles run AI-led interviews across functions and countries at the same time, with questions designed for each role. The Process Library turns the conversations into structured, navigable process documentation that grows with each cycle. The Insights Dashboard groups findings by process, system and cause and ranks them by impact and effort, and the Initiatives Dashboard turns priorities into business cases with owners.

When a process runs mostly inside systems with clean logs, process mining measures it with precision, and the two combine well: mining shows where cases wait, and interviews explain why.

FAQ

What is process discovery?

Process discovery is the work of finding out how a business process actually runs today, including the people, systems, handoffs, exceptions and workarounds involved. It produces a trustworthy current state that leaders can use to decide what to improve, automate or redesign.

What are process discovery tools?

Process discovery tools fall into three groups: process mining tools that read system event logs, task mining tools that capture on-screen activity, and AI interview platforms such as Horizon that gather evidence from employees at scale. Diagramming tools then document the result.

What is the difference between process discovery and process mining?

Process mining is one method of process discovery. It reconstructs a process from system event logs. Process discovery is the wider effort, and it also draws on interviews, documents and observation to cover work that happens outside systems.

Can you do process discovery without event logs?

Yes. Workshops and AI-led employee interviews build the process from what the people doing the work describe. Trafilea mapped work that spread across seven or more tools through 43 interviews in two weeks.

How long does business process discovery take?

A manual effort on a complex process can take months. With AI interviews, a first round on one process can run in a week, and Trafilea completed 43 interviews across two teams in two weeks.

Who should be involved in process discovery?

Everyone who touches the process: the people who do the work, the managers who approve exceptions, the downstream teams who receive the output, and the process owner who will act on the findings.

Discovery ends in a decision

Business process discovery is worth the effort when it changes what the organization does next. The best discovery work combines evidence from people and systems, maps the variants along with the main flow, and ends with a short list of initiatives that each have an owner.

See it. Fix it. Lead it.

Ready to transform?

See Horizon in Action

Discover how AI-powered organizational discovery can uncover hidden opportunities in days, not months.

Get Started

Related Resources