Everyone in this category quotes an "under a minute" number. We wanted to know what the minute is made of — because you can only design against a number you've broken apart.
The test that separates pool-native software from generic field-service tools is whether a technician can log chlorine, pH, alkalinity and stabiliser, see the calculated dosage, capture a photo and complete the visit in under a minute. The category leader does it consistently in about forty-five seconds. That is a design number, not a backend one — no database is the bottleneck at that scale.
So we did the boring thing. We built the flow the way most systems build it — a vertical stack of text inputs, a product picker, an add button per line item, then a submit — and watched people use it with a stopwatch running.
Thirty-one seconds of an eighty-second median went into entering four numbers. Not thinking about them — entering them. The keyboard covers the form, the field loses focus, a wet thumb hits the wrong key, and 7.4 becomes 74 with no complaint from the software.
The other two costs were smaller but more annoying: twelve seconds finding the right account when the route order didn't match the driving order, and nine seconds of mental arithmetic converting a reading gap into an ounce count for a pool whose volume the technician had to remember.
None of the three costs were about the technician being slow. All three were the software asking a person to do something a computer already knew.
Three changes, in the order they paid off:
Median went from eighty-one seconds to forty-two. The interesting part is that the perceived improvement was much larger than the measured one — people described the new flow as "not really data entry", which is the actual goal. A technician who doesn't feel like they're doing paperwork will log the visit at the pool instead of in the truck at the end of the day, and that is what makes the data trustworthy for everyone downstream.
Forty runs in controlled conditions is not a season on a real route. We don't yet know how the flow holds up in the last hour of a forty-stop day, or how often the prefilled value gets accepted when it shouldn't be. Both are things we'll only learn from pilot operators, and both are reasons the number in our marketing is forty-two seconds and not thirty.