When a City Writes Down Its Water Loss, a Purchase Is Coming
Across the United States, roughly one in five gallons of treated drinking water never gets billed, costing US utilities about $6.4 billion a year. The water leaks out of old pipes, or it slips past worn meters that undercount what customers actually use. The industry calls the gap non-revenue water. A city just calls it money it treated, pumped, and never got paid for.
If you sell to governments, here is why that matters: a city cannot close that gap without buying something. New meters. A smart-meter network. A leak-detection program. So the moment a city writes its loss number into a budget or a set of board minutes is the earliest moment the coming purchase is visible to anyone paying attention.
Almost nobody is paying attention, because the obvious way to watch is broken. Set a keyword alert on "water loss" and your inbox fills with lawn care: a Harris County, TX utility district telling residents to water after 7 p.m., springtime landscaping tips from Hidden Hills, CA, pool covers from Contra Costa Water District, drought messaging from Castle Rock, CO.
So we sorted it. We pulled 46 public records from 44 agencies across 21 states, budgets, board minutes, capital plans, and press releases, and separated the conservation advice from the real thing. The phrase is noise. The number is the tell. At least 15 agencies have already written theirs down.
Finding 1
The signal is a number, not a phrase
The disclosed figures run from a low target of 8 percent to a measured 26.06 percent. Oreana, IL reported 26.06 percent in a meeting packet. Gunnison, CO's budget puts its loss at 20 to 25 percent. Newton, NJ recorded 21 percent. Delta, CO reported losses up to 14 percent. Zion, IL measured 12.1 percent. Lanai City, HI carries 12 percent. Warren, MI budgets to hold losses at or below 10 percent. Oklahoma City set a target below 10 percent, as did Channahon, IL. Santa Clara, CA's target is 8 percent.
For scale, here is the national picture: Bluefield Research estimates 19.5 percent of treated drinking water in the US is lost or improperly billed, costing utilities $6.4 billion a year. A utility publishing a 21 or 26 percent figure is telling you it sits at or above the national average, in its own documents, under its own letterhead.
One town even priced the problem. Countryside, IL's budget book calculates that every 1 percent of non-revenue water costs the city just over $33,000. That is a buying signal with a dollar sign already attached.
The exact phrase matters less than the number. A "water loss audit" line item, as in Dallas, TX, Ferdinand, IN, Cedar Lake, IN, and Canal Winchester, OH, is the same signal by a different name. Full audit tables in the style states require appear in Lexington, TN and Dyersburg, TN annual reports.
Finding 2
The number moves years before any solicitation does
Two proof cases show the conversation forming in public, step by step, long before any purchase.
In Lanai City, HI, Maui County board minutes from October 2020 introduce the concept to the board, with staff telling members "you might not be familiar with that term" while explaining non-revenue water. Nineteen months later, the department put its figure on the record: 12 percent, reduced over the prior five to ten years. Neither step is visible to a bid feed. Both are in the public record.
Zion, IL did all of it in one meeting. Its council minutes from November 2016 record that losses had fallen from 18.9 percent to 12.1 percent after leak repairs, and set targets of no more than 12 percent by 2017 and 10 percent by 2020. That is a starting number, a result, and a schedule of future targets, written down four years before the last target date.
Finding 3
When the number appears, the meter follows
Once a utility names its number, the fix is almost always metering, and that is where the buying starts. Ten agencies in our sample tie their loss figure directly to new meters, an AMI network (the smart-meter systems that read usage automatically and accurately), or a leak-detection program.
Chandler, AZ's capital plan funds meter replacement explicitly to cut non-revenue water loss. Lake Charles, LA credits its automated meter system with "reducing unaccounted-for water". New Port Richey, FL budgeted a smart-meter water loss and revenue analysis. Newburyport, MA tied a citywide meter upgrade to "reduced undetected water loss". Corona, CA reported its AMI network 94 percent installed. Flora, IN blamed "old meters that have slowed down". Hadley, MA swapped manual-read meters for tamper-proof units, and San Bernardino County, CA uses cellular-read meters for leak detection.
Then comes the step a bid feed finally sees. North Chicago, IL authorized public works to advertise a leak-detection bid specifically to cut non-revenue water. That authorization is the first moment this deal would appear in any solicitation feed. Everything before it, the percentage, the audit, the meter line items, was already on the record.
Finding 4
The audits are multiplying, so the signals will too
This category is about to get louder, because more utilities are being required to produce the number.
The standard method is a water audit built on the industry rulebook, AWWA M36, and adoption is spreading unevenly: an industry review this year counts only about 10 states that currently require utilities to use these auditing standards, while Bluefield Research notes that states including California, Georgia, Indiana, and Texas are moving to standardize loss reporting and validation. Every new requirement turns a private internal number into a public, dated record, which is exactly the signal this report is built on.
The infrastructure math points the same direction. US water systems lose roughly 6.75 billion gallons of treated water a day across 2.2 million miles of pipe, with a main break estimated every two minutes. The loss percentages are not going away. The paperwork documenting them is only growing.
Where the pattern breaks
- A written number is not a guaranteed purchase. Some utilities disclose a percentage and defer the fix for budget reasons. The number is a qualification signal, not a closed deal.
- Targets and measurements mean different things. Santa Clara's 8 percent is a goal and Warren's 10 percent is a ceiling to maintain, which can mean the metering work is already done. A measured 26.06 percent, like Oreana's, is the hotter signal. Figure 1 marks them separately for exactly this reason.
- Two records carry sourcing caveats. Tekoa School District, WA's 21 percent comes from minutes with no public URL available in this run, and it is a school district discussing its system, not a water utility. Mesquite, NV's roughly 10 percent sits on a vendor meeting-archive platform. Both are noted rather than hidden.
- The keyword still catches the endgame. A bid feed will surface the North Chicago solicitation eventually. The cost of waiting for it is every month of relationship-building between the first disclosed percentage and the posted bid, the same gap we found in how governments buy license plate cameras.
The verdict
Track the percentage, not the phrase.
"Water loss" as public messaging is evaporation timing and pool covers, noise by volume. "Water loss" as a number in a budget or set of minutes is a utility telling you, in writing, that it has a problem, a size, and usually a target. The meter, AMI, or leak-detection solicitation lands months to years later, and in some markets the purchase order never posts at all.
Bid feeds catch the RFP. The number came first.
See the deal before it becomes a bid. Tell us what you sell, and we will show you where demand is forming in your market, in budgets, council actions, and grant awards, before a solicitation is posted. Book a demo.
Methodology: this report draws on 46 public records from 44 agencies across 21 states, May 2010 through July 2026, from NationGraph's index of budgets, board minutes, capital plans, press releases, and solicitations; several records are forward-dated budget or CIP documents carrying future fiscal-year labels. Counts are conservative floors from a single run: 15 agencies with a quantified figure and 10 tying loss to metering reflect only records surfaced here, and the true population is larger. The noise-versus-signal characterization describes how the phrase "water loss" is used in the sampled records and is not a census of all public records. Tekoa School District, WA is cited from minutes without a public URL in this run, and Mesquite, NV from a vendor-hosted meeting archive; both are flagged where used. National figures come from Bluefield Research and industry sources, linked where cited. Every other claim links to a primary public record.
Find your first opportunity in minutes
