Unstructured Work
Last updated
Unstructured work is the execution that happens outside the systems that log it: ad-hoc work in spreadsheets and email, project work that is not a linear process, exceptions handled manually, and legacy applications with no modern logs.
What Unstructured Work actually means
Process mining captures what structured systems record. In most enterprises that is the minority of execution. The majority sits in spreadsheets and email, in project-based work that does not follow a repeatable path, in the exceptions that get handled by hand precisely because the system could not handle them, and in legacy applications that produce no usable log.
That share is often estimated at around seventy percent, and its composition matters more than the number. It is not a random remainder. It is disproportionately made up of exceptions, judgement calls, and the workarounds people invented when the system did not fit, which means it concentrates exactly the bottlenecks, error sources, and automation opportunities a program is looking for.
This produces a systematic bias in tools that depend on logs. They report confidently on the part of the process they can see, and the part they cannot see does not appear as a gap. A process that looks clean in an event log can be held together by three spreadsheets and a standing weekly call, and nothing in the log will say so.
Capturing unstructured work requires observing the work rather than querying the systems. The consequence for AI targeting is direct: a candidate list built only from logged work systematically under-counts the addressable surface, and tends to point at processes that are already the most structured rather than the ones where the most time is going.
Examples
The spreadsheet holding the process together
An event log shows a clean three-step reconciliation. In practice a spreadsheet sits between steps one and two, maintained by one person, reconciling the two systems that do not talk to each other.
Exceptions as the real workload
A claims process is automated for standard cases. The automation reports high coverage. The team time is going almost entirely to the exception queue, which the automation routes out and the log records only as an exit.
Work in a system with no logs
A pricing step runs in a legacy application with no modern logging. To every log-based tool, that step does not exist, so the process appears to jump from quote to approval with nothing in between.
Frequently asked questions
Related terms
Process Mining
Process mining reconstructs how a process ran from the event logs that systems recorded. It is usually bought as the discovery phase of a larger program: find the inefficiencies, then redesign, automate, and measure.
Execution Data
Execution data is the record of how work actually moves through an organization: the layer between the inputs a business captures and the outputs it reports. It covers workflow patterns, collaboration structures, and where work slows down.
Work Intelligence
Work intelligence is the operational layer that provides a current, evidence-based picture of how work actually moves through an enterprise. It is used to target AI deployment, verify post-deployment outcomes, and continuously update the baseline as the business changes.
