Pick one process you own. An approval chain, customer onboarding, month-end reconciliation, a claim, an intake queue. Answer one question without contacting anyone: how many items it holds now, how long the oldest has waited, and which step it is stuck on.
Most executives in companies between sixty and two hundred million dollars in revenue cannot answer that for at least one process they own. Not because the company is badly run. The answer does not exist anywhere it can be looked up. It is spread across the memories of several people, and assembling it is what the standing meeting is for.
Twelve filings over the trailing twelve months use the exact phrase "manual processes and spreadsheets" to describe this condition. That count is small because most companies in the same state describe it less plainly, or they do not describe it at all.
The Honest Versions Of "No"
The negative answers are useful because each one points at a different missing piece.
- "I can find out." Somebody can assemble it. The information exists; it has no home.
- "There is a report, from last night." The system knows about the records but not about the work. It can tell you what exists, not what is waiting or on whom.
- "The tracker is current if Ana updated it." The record is a document maintained by hand, so its accuracy is a person's discipline rather than a property of the process.
- "That depends what you mean by open." The process has no agreed states, so two people counting it produce two numbers and both are defensible.
All four are the same finding: the work is performed, and the process is not recording it.
The Process Leaves Documents, Not Events
The process mining manifesto (van der Aalst et al., 2012) sets out what is needed to discover, check or improve a real process: an event log where each event belongs to a case, refers to an activity, carries a timestamp, and ideally names the person who performed it. Without those four fields there is nothing to analyze, however much data the company has.
A process run on spreadsheets and email produces artifacts instead: a file without a case identifier, an email with a timestamp and no activity, a folder whose modification date records the last save instead of the last decision. Each of these artifacts is evidence that something happened, and none of them can be counted, sorted, or aged.
A system that stores your documents is not the same as a system that knows what your process is doing. Most companies own the first and assume it is the second.
The process cannot be measured; every claim is an assertion rather than a measurement.
What The Missing Record Costs
The cost is usually described as the meeting, and the meeting is the smallest part of it.
- Coordination absorbs the senior hours. The study of coordination in real development work (Kraut & Streeter, Communications of the ACM, 1995) found teams relying heavily on informal, person-to-person communication for coordination precisely where the formal mechanisms were weakest — and that informal coordination scales with the number of people involved, not with the amount of work being coordinated.
- The interruptions are not free. The cost of interrupted work (Mark, Gudith & Klocke, CHI, 2008) found interrupted tasks were completed in less time, but at the price of higher stress and effort. Asking four people for a status update is four interruptions.
Together those explain a pattern most operators recognize. The process is not slow because the work is hard. It is slow because it waits, and nothing announces that it is waiting.
Adding The Record Is Smaller Than Replacing The Process
The reaction this diagnosis usually provokes is that the company needs a system, and that a system means a year and a large budget. That conclusion is wrong in a specific and useful way. What is missing is not the ability to store the work. It is a record of the work moving.
| What exists today | What has to be added |
|---|---|
| A shared folder of files | A case identifier that travels with the item |
| An email approving something | A recorded approval, by a named person, at a time |
| A spreadsheet column called "Status" | A defined set of states the item can be in |
| A weekly meeting that produces the count | A query that produces the count |
Four fields per step: which item, which step, when, and who. That is the whole addition, and it is why this work is measured in weeks rather than in quarters. It sits beside the enterprise system, in the part of the process the enterprise system was never built to cover.
Two things follow once the record exists. The status question is answered by a query instead of a meeting, and the process becomes something an auditor can test.
First Steps
- Ask the question about three processes you own, in the same week. What state is each in now, without asking anyone. Note which of the four negative answers you got.
- For the worst one, write the list of states an item can be in. Five or six words each. If those who run the process cannot agree on it in one meeting, that is the finding.
- Count the standing meetings that exist only to assemble status. Attendees times hours times weeks is the cost, and it is usually larger than the cost of adding the record.
Answer It With A Query
The question separates a merely manual process from an invisible one. Manual is often fine. Invisible makes every decision about it slow and every improvement unmeasurable.
A fifteen-minute call takes one such process apart: what states it has, what record each step should produce, what has to be built, and what it costs. If the record already exists in your systems and the work is retrieval and reporting, that is the answer you get.
Where something does have to be built, that work runs as a monthly engineering partnership, with the targets for each release agreed before it starts. Paid-for deliverables transfer to you monthly. The engagement runs in TopDo, ML LABS's own system, so you can see its state daily: the same test, applied to ML LABS. After that, $10,000 a month puts one accountable person on keeping it running, with a written monthly report of what ran and what changed.
References
- van der Aalst, W. M. P., et al. Process Mining Manifesto. Lecture Notes in Business Information Processing, 2012.
- Kraut, R. E., & Streeter, L. A. Coordination in Software Development. Communications of the ACM, 1995.
- Mark, G., Gudith, D., & Klocke, U. The Cost of Interrupted Work: More Speed and Stress. Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, 2008.
