“You find yourself arriving at the clinic around 05H45 or 06H45,” a 48-year-old man in South Africa’s North West Province told researchers in 2016, “but you only finish late around 16H00 or 15H00.” He was one of 14 people with HIV, all recently out of hospital, whom field workers followed through 15 clinic visits. At every step the field workers noted the time a wait began, the time the service began and the time it ended.
When the researchers added it up, the typical visit took 3.3 hours inside the clinic. Of that, 2.9 hours went to standing or sitting in queues, five of them on a typical visit. The time spent with clinic staff, counting even fetching the patient’s file, came to 25 minutes. The case has the rest.
Time the gaps, not just the steps
Draw every step a piece of work goes through, and time the waiting as well as the work.
That is a value stream map. The Lean Enterprise Institute defines it as “diagraming every step involved in the material and information flows needed to bring a product from order to delivery,” and says it grew out of what Toyota calls a “material and information flow diagram.” For each step the map records two kinds of time: how long the work itself takes, and how long the work sits before or between steps. The total from start to finish is the lead time.
The point of following a single item, a patient or an order, is that waiting rarely belongs to any one desk. The clinic study didn’t judge how quickly the nurse, the pharmacist or the clerk worked. It measured the patient’s day, and by that measure the typical patient’s time with staff was about a seventh of the time spent in queues.
What the map turns up
Laid end to end, the clinic visits showed more than slow lines. Three patients had to queue for a nurse or doctor a second time because their file lacked prescription details when they reached the pharmacy. Two were turned away by specialists because the referral note didn’t say enough. One woman was told to come back the next day for her medicines, her third visit in three days. Each of these is a small slip at one step that sends someone back through the line.
The researchers judged that the inefficient processes behind many of the delays “appear to be correctable.” They suggested training staff on what a prescription needs, getting lab results into the record promptly, and streamlining registration, which they estimated could save about an hour. These were proposals; the study mapped the visits but didn’t test the fixes. It was also small, 15 visits in one district.
A map doesn’t say what to change. It shows where the time goes, so people stop guessing.
Imagine a small landscaping firm where a customer’s request for a quote takes two weeks to answer. Writing the quote takes maybe 40 minutes. The rest is waiting: for the owner’s site visit, for a supplier’s prices, for a sign-off. A faster quoting template might save a few minutes and leave the two weeks almost untouched.
AI in the value stream
Google’s DORA research program, which studies how software teams deliver, put this at the center of its 2025 report on AI. From nearly 5,000 survey responses and more than 100 hours of qualitative data, it concluded that AI acts as “an amplifier,” magnifying “the strengths of high-performing organizations and the dysfunctions of struggling ones.” It found that “while AI adoption on its own shows a modest impact, the effect is dramatically amplified in organizations with strong VSM practices,” meaning value stream management.
The report gives an example. A team “may discover, through mapping, that code reviews are a significant bottleneck.” It could then use AI to speed up reviews, “rather than using AI to simply generate more code that will only exacerbate the bottleneck.” The finding comes from a survey, so it shows which organizations do better, not that mapping causes the gain.
In the North West clinics, the map showed where the day went: 25 minutes with staff, and most of the rest spent waiting.