How to Set Up a Water Routine a Customer Will Actually Keep
Why this matters
Most water routines handed to a customer are technically excellent and dead within two months. The tech writes a thorough sheet, walks the building engineer through six readings, and leaves. The sheet gets filled in for four or five weeks and then stops, and nobody mentions it again until something fails and the sheet is produced as evidence that the customer did not follow the program.
That outcome is a design failure, not a customer failure. A routine survives exactly when the effort of doing it is smaller than the resistance to doing it, and every extra reading raises the effort side while nothing on the sheet ever lowers the resistance side. So the design act is subtraction: find the smallest set of readings that can still detect a change, and move everything else, the interpretation, the judgment, the decision, off the person holding the pen.
The arithmetic that should decide the length of the sheet
Take a six-reading routine at about 3 minutes each. That is 18 minutes a week, which is roughly 15.6 hours a year if it is done every week. It almost never is. Call it five weeks before it lapses, which is 5 readings taken in a year and about 90 minutes actually spent.
Now take a three-reading routine at about 2 minutes each. Six minutes a week, about 5.2 hours a year, and this one gets done because it fits inside a walk somebody already does. Fifty-two readings in a year, about 312 minutes actually spent.
Compare like with like and the picture is honest in both directions. The short routine costs more time actually spent, about 3.5 times as many minutes as the long one, because it is the one that happens. And it produces about ten times the observations, 52 against 5. Minutes spent is the wrong scoreboard here. Observations per year is the number that decides whether a change gets caught, and a routine that is not performed produces zero of them regardless of how good it was.
This is why the sheet gets shorter rather than more thorough when a customer stops using it.
Three tests a candidate reading has to pass
Not every useful reading belongs on a customer routine. Run each candidate through all three:
Is it cheap in time and unambiguous to record? A number off a meter or a gauge passes. A judgment about clarity, color or smell does not, because it is not comparable between two people or between two months.
Does it move early? The reading has to change before the failure, not with it. A makeup meter moves the day a leak starts. A corrosion finding moves years later.
Can it be taken without protective equipment or a tool the customer does not own? If a reading requires goggles, a glove class, a hot sample cooled before handling, a ladder, or opening anything pressurized, it does not belong on a customer's routine at any price. That reading belongs to you on your visit. A routine that quietly makes a building engineer reach across a hot line or crack a sample port on a pressurized system has created a hazard for a person who was handed a clipboard, not training, and the fact that it fails the resistance test as well is almost beside the point.
That third test eliminates more candidates than the other two combined, and it is the one most often skipped, usually on a domestic hot water or boiler sample point where the water at the port is at scald temperature.
What goes on the card, and why each field is there
The artifact is one page, laminated, hung at the reading point rather than filed in an office.
| Field | Why it exists |
|---|---|
| System name and the exact reading point | Two gauges 6 feet apart are two different readings, and next year nobody will remember which one was meant |
| The reading, its tool, and its unit | A number with no unit is not comparable to anything |
| Target band, printed | So the person reading does not have to interpret |
| Action if inside the band | Almost always "record and move on", which needs to be stated or the reader invents work |
| Action if outside the band, with the phone number on the card | The trigger goes to somebody whose job is the interpretation |
| Date and initials | Attribution, and it makes a lapse visible |
| One line: anything changed on this system since last time | The single field that makes an out-of-band reading explainable later |
Two rules about the layout matter as much as the fields. Blank lines stay blank rather than being backfilled, because a blank line is real information about the routine's health and a guessed entry is worse than nothing. And the card is attached to something that already happens on that schedule, an existing rounds sheet, a Monday walk, a filter check, rather than being a new event on somebody's calendar. New events get dropped. Additions to existing events survive.
The two gates, stated so they cannot be argued with
Every reading gets a band and two triggers, and the triggers are deliberately different shapes because they catch different things.
The spike gate is a single reading well outside the band. Unit of analysis is one reading on one system. One occurrence triggers the call.
The drift gate is a smaller excursion that persists. Unit of analysis is consecutive readings on one system, and it requires two in a row above the drift threshold, not one. A single reading slightly out is noise on most instruments, and a routine that calls on noise trains the customer to stop calling.
Both thresholds are stated as multiples of that system's own established baseline rather than as absolute numbers, because the baseline is a property of the system and the multiple travels. And the baseline is established by measurement over the first several readings, not by the tech's estimate on day one.
The filled-in card
A small commercial closed hydronic loop with an automatic fill. Reading point is the fill line's water meter and the fill-side pressure gauge, both at the mechanical room wall, both reachable standing, neither requiring anything to be opened.
| Reading | Tool | Band | In band | Out of band |
|---|---|---|---|---|
| Makeup meter, total gallons | The meter face | Weekly increase at or under 2 gallons, the baseline established over the first 4 weeks | Record the number | Spike gate: any week above 3 times baseline, call. Drift gate: two consecutive weeks above 1.5 times baseline, call |
| Loop pressure, cold, at the wall gauge | The gauge | Within the fill setting range noted on the card | Record | Outside range on any reading, call |
| Inhibitor level | The supplier's test kit, on a cold sample from the designated tap | The supplier's stated range, printed on the card | Record | Below range on two consecutive readings, call |
Note what is not on the card. No pH. No conductivity. No visual clarity judgment. Not because those are unimportant, but because they belong to the visit, where the tools and the training are.
Four weeks of that card, read
Meter readings recorded on four consecutive Mondays: 41,204, then 41,206, then 41,208, then 41,211, then 41,219.
Deltas: 2, 2, 3, 8 gallons. Baseline is 2 gallons a week.
Week one and two sit at baseline. Week three at 3 gallons is 1.5 times baseline, which is at the drift threshold rather than above it, and the drift gate requires two consecutive weeks above it in any case, so the correct action is to record it and move on. That is not the routine failing to notice; it is the routine behaving as designed, and a card that had triggered a call there would have burned credibility on noise.
Week four at 8 gallons is 4 times baseline, above the 3 times spike gate, and it is a call the same day.
Eight gallons in a week on a loop that normally takes two is a leak, a weeping relief valve, or a fill valve passing. All three are ordinary problems and all three are cheap while they are small. What matters is the calendar: this was caught in week four by a reading that took about 30 seconds, taken by somebody who does not know what an inhibitor is.
The comparison version writes itself. On the six-reading routine that lapsed in week five, the same fill valve passes for months. The makeup water it brings in carries dissolved oxygen and dilutes the inhibitor with every gallon, so what eventually surfaces is not a leak at all: it is an inhibitor result out of range at the next scheduled visit, or corrosion product in a strainer, months downstream, with no record that could tell anyone when it started. The mechanism by which makeup water drives that outcome is its own article and is not re-derived here.
The part that keeps it alive after month three
A routine dies when the readings go into a folder nobody opens, because the person taking them works out within a season that the numbers do not go anywhere.
So the routine has two halves and the second one is yours. Someone reviews the sheet on a defined cadence, monthly is a reasonable default and quarterly is the floor, and the review is visible to the person who filled it in: a note on the card, a line in the visit report, a sentence at the next visit naming a specific week's reading. Naming one actual number from their sheet is what proves it was read, and it costs a minute.
The review also checks completeness rather than only values. Count the filled lines against the expected count. If a four-week month has two entries, the finding is not a water finding, it is that the routine is failing, and that gets addressed before the next out-of-band reading is missed. A routine at half compliance is not half a routine; the gaps land wherever they land, and the week the leak starts is as likely to be a blank as any other.
One more thing to hand over at setup: what happens when the person who takes the readings leaves. Name the reading point on the card precisely enough that a stranger can find it, keep the card at the point rather than in a folder, and put the successor's first four readings on your own calendar to re-check against the established baseline, because a new reader taking a reading from a slightly different point resets the series without anyone noticing.
References
- Treatment supplier documentation for test kit ranges, sampling instructions and product-specific target bands
- Equipment manufacturer documentation for normal fill pressure and makeup expectations on a closed system
- See related: Why Makeup Water Is the Variable That Decides Everything; How to Find Out How Much Makeup Water a System Is Really Taking; What a Water Management Plan Is For