The onboarding was supposed to take two weeks. At a 14-person consultancy, it was consistently running at five, and every conversation with the team produced a different theory about why. Contracts were waiting on legal review. The client had been slow to provide information. A CRM task hadn’t been picked up in time. Each explanation was probably partly true, but none of them pointed to the same step, and none of them told the founder where to look first.
The data to settle the question was almost certainly sitting inside the CRM and project management tool already. Process mining is one way to read it.
What is process mining?
Process mining reads the event logs your business software generates as cases move through a workflow and turns those timestamps into a map of how work actually travels through your firm. Each state change, assignment, approval, or escalation leaves a record in the system. Process mining connects those records to show where cases wait the longest, where they loop back for rework, and which steps take more touches than they should.
The software doesn’t need to know what the intended process looks like. It observes what actually happened, draws the path each case took, and lets you compare the typical route against the slow outliers. If nine out of ten invoice approvals clear in a few hours and the tenth takes several days, the event log will show you what was different about those slow cases.
A process diagram drawn on a whiteboard describes the plan. A process map built from event data describes the reality. For a service firm with a repeatable workflow, that difference is often where the useful work begins.
Why does it matter for a small service firm?
For a 5-50 person service business, process mining’s value is specific: it tells you where work is actually stalling, rather than leaving you to manage competing theories from different parts of the team. Deloitte’s 2021 Global Process Mining Survey found 84% of organisations using the technology reported value from their investment. Many stopped once they had found the bottleneck. That was enough.
Fewer than half of those Deloitte surveyed named process redesign or automation as a top expected outcome. Visibility was the goal for many adopters. For a small service firm, that framing keeps the project proportionate: a diagnostic exercise with a practical scope, rather than an initiative with a board presentation and a multi-year roadmap.
The practical output, for a firm that runs a focused pilot, tends to be a short list. One approval step is causing delay. One team is doing avoidable rework because cases arrive with missing information. One software handoff is creating errors that have to be corrected manually downstream. Two or three findings of that kind are often enough to justify the exercise and identify where to act.
Where will you actually meet it?
Process mining works when a workflow repeats regularly and leaves consistent digital records. Quote-to-cash, client onboarding, invoice approval, complaint handling, job scheduling, and case or ticket management are the most common candidates in service firms. Where the same steps run dozens of times a week and each state change lands in a system with a reliable timestamp, there is usually enough data to produce a useful picture.
The qualifier matters. If the workflow lives primarily in email threads, verbal handoffs, or a spreadsheet that different people fill in differently, the event data may be too thin or inconsistent to show you anything reliable. ProcessMind, one of the platforms positioning itself for SMB use, supports data upload from Excel or system exports and offers a free trial, which gives smaller firms a lower-cost entry point than enterprise-grade tooling. The platform can only work with what the data actually contains, though.
Before committing to a process mining exercise, check whether the workflow you have in mind generates a clean log. Can you export a record of every case, with timestamps at each stage, from opening to close? If yes, you likely have enough to work with. If the answer is mostly yes but some approval steps happen by phone or message, that gap is where the analysis will start to become unreliable.
When should you look at it, and when should you not?
Process mining is worth looking at when a repeatable workflow is clearly underperforming and conversation alone isn’t showing you where the problem sits. Each member of the team will have a theory about the delay. The event data shows where cases actually waited. That gap between what people believe is happening and what the log records is often where the real issue sits, and for firms with clean digital records it can surface quickly.
Skip it when the core work is genuinely bespoke, where each case is different enough that patterns won’t emerge across the data. Skip it when work primarily happens outside your systems, when the process runs infrequently, or when the bottleneck is already obvious to everyone who does the work. If you know which step is slow and why, you don’t need a mining tool to confirm it.
There is also a practical effort question. Cleaning the data, configuring the analysis, and then getting the team to act on what it shows takes real time. For a firm under ten people running a straightforward workflow, a structured conversation with the people doing the work, a simple flow chart, and one afternoon of honest counting may cover most of the same ground at a fraction of the cost.
If you do proceed, decide what success looks like before you start. For a 5-50 person firm, the right measures are usually fewer handoffs, shorter turnaround time, less rework per case, and less admin time. Set that bar first. A process mining project without a clear success definition tends to drift into dashboards that nobody acts on.
What else should you know before you start?
Process mining and task mining are different tools that address different gaps. Process mining analyses event logs from your systems of record; task mining observes what users do on-screen to capture manual steps that don’t appear in any back-end log. If the real bottleneck in your firm is manual admin that bypasses the core system, task mining may be the closer fit, or you may need both approaches to see the full picture.
UK data protection rules apply whenever event logs include personal data, which they frequently do. Staff activity logs, customer case records, and complaint histories all qualify. The ICO requires a lawful basis for processing under UK GDPR, transparency with the individuals involved, and proportionate data minimisation. These are design decisions that belong in the scoping conversation, not questions to resolve once the pilot is already running.
The NCSC also notes that data-intensive analytical tools can introduce supply-chain and access risk. A process mining platform holds operationally sensitive data about how your business works in practice. Check where data is stored, who can view exports, and what the vendor’s security controls look like before you upload anything.
For firms operating in regulated sectors or selling into the EU, the EU AI Act, adopted in 2024, adds a further layer to consider. Process mining used as a pure diagnostic is unlikely to trigger the Act’s higher-risk categories. If the output starts feeding automated decisions about staff allocation, client prioritisation, or service delivery, it is worth assessing where that use sits within the risk framework before building a dependency on it.
