Every operator of distributed infrastructure runs the same calculation, usually without writing it down. A water utility with booster stations across a county, a pipeline company with valve sites along hundreds of miles, a telecom operator with equipment shelters on hilltops, an electric utility with substations in every town: each has to decide how often to send someone to look at assets that are mostly fine.
For most of the last century, the only way to know the condition of a remote asset was to go there. Continuous sensing changes the options, but not automatically the economics. This article builds the comparison from the ground up: what a site visit really costs, what continuous monitoring has to replace or avoid to pay for itself, and why the answer depends on distance, consequence and how much you trust your alerts.
How much is out there
The amount of distributed equipment is hard to picture. The U.S. Environmental Protection Agency counts more than 148,000 public water systems in the United States, from large city utilities to small systems serving a few dozen connections. The Pipeline and Hazardous Materials Safety Administration describes the nation's pipeline network as more than 2.6 million miles long. Each of those systems contains pumps, motors, valves, compressors and electrical equipment, much of it unattended, and most of it visited on a schedule.
What a visit actually costs
The cost of a routine visit is easy to underestimate because most of it is not spent on the asset. Consider an illustrative visit to a site 60 miles from the depot: two technicians, a 120-mile round trip, two and a half hours of driving and an hour and a half on site.
| Item | Assumption | Cost |
|---|---|---|
| Driving time | 2 people, 2.5 hours, $75 per hour loaded labor rate | $375 |
| On-site time | 2 people, 1.5 hours | $225 |
| Mileage | 120 miles at 76 cents | $91 |
| Scheduling, records, preparation | Office and planning time | $60 |
| Total | about $750 |
The mileage figure uses the IRS standard business mileage rate of 76 cents per mile for the second half of 2026, a reasonable proxy for the full cost of running a work vehicle. The labor rate is an assumption. Many organizations will have higher loaded rates once benefits, overhead and specialist skills are counted, and two-person crews are common where safety rules require them.
There is also a cost that does not appear on any invoice. In the Bureau of Labor Statistics' census of fatal occupational injuries for 2024, transportation incidents were the most frequent type of fatal event, accounting for 38.2 percent of all occupational fatalities. Every unnecessary mile driven carries a small amount of that risk.
Two ways of knowing
A scheduled inspection program buys information by the visit. Its strength is that a trained person sees, hears and smells things no single sensor captures. Its weakness is coverage in time: the technician sees the asset for an hour and a half every month or quarter, and nothing in between. Most visits find nothing. That is not wasted effort, since confirming health has value, but it is an expensive way to confirm it.
Continuous monitoring buys information by the site. Its strength is coverage in time: it is present on the night the bearing starts to fail. Its weakness is that it only sees what its sensors measure, and it can raise alerts that turn out to be nothing. Both strengths and weaknesses matter for the economics.
A worked comparison
Take an illustrative operator with 60 remote pump sites, each 60 miles from the depot on average, and compare two programs over a year. All numbers are assumptions chosen to show the structure of the calculation.
Program A: monthly visits. Every site is visited every month. That is 720 visits at about $750, or roughly $540,000 a year.
Program B: quarterly visits plus continuous monitoring. Every site is visited every quarter for the checks that need a person: 240 visits, or about $180,000. Monitoring raises 40 real findings a year that justify an extra visit, and, at 70 percent alert precision, 17 more that turn out to be nothing. Those 57 dispatches cost about $43,000. Assume monitoring costs $3,000 per site per year, all in, for $180,000. The total is about $403,000.
| Program A | Program B | |
|---|---|---|
| Scheduled visits | 720 | 240 |
| Dispatches on findings | 0 | 57 |
| Visit cost | $540,000 | $223,000 |
| Monitoring cost | $0 | $180,000 |
| Total | $540,000 | $403,000 |
| Miles driven | about 86,000 | about 36,000 |
On these assumptions Program B costs about a quarter less, cuts driving by more than half, and observes every site continuously instead of twelve times a year. It also arrives at the 40 real findings with an idea of what is wrong, which shortens the repair visit and reduces the chance of a second trip for parts.
The monitoring cost assumption matters as much as any other. At $5,000 per site per year instead of $3,000, Program B rises to about $523,000, almost level with Program A on visit and monitoring costs alone. Anyone running this comparison should use a real quote, all in, including connectivity, installation and the staff time to review findings.
When the numbers do not work
The same model shows where continuous monitoring does not pay on travel savings alone. Move the sites to 20 miles from the depot and a visit costs about $435. Program A then costs about $313,000 and Program B about $309,000, roughly break-even.
Precision matters too. If alert precision falls from 70 to 20 percent, the 40 real findings bring 160 false ones, and dispatches rise from 57 to 200. At $750 each that adds more than $100,000 and wipes out most of the savings. This is why we argued in our piece on false positives that alert precision is an economic quantity, not only a technical one. A cheap remote check helps a great deal here: if an analyst can review the data for twenty minutes and rule out most false alerts before anyone gets in a truck, the cost of a false alert falls from $750 to perhaps $30.
The value that travel savings miss
Travel is the easiest benefit to count and usually not the largest. The bigger value is the failures that do not happen, or happen on a planned day instead of at night. As we showed in the hidden cost of a failed pump, a single unplanned failure at a critical site can cost tens of thousands of dollars in direct and consequential costs, and carry a much larger exposure if a standby unit fails too.
In the worked comparison, if continuous monitoring turns just two unplanned failures a year into planned repairs, and each avoided failure saves $40,000, the case for Program B improves by $80,000. At 20 miles from the depot, where travel savings alone broke even, that is what makes the decision.
There are also benefits that are real but hard to price: less driving risk for crews, better information for capital planning, records that show regulators and insurers how assets are maintained, and the ability to run more sites with the same team as experienced staff retire.
What data cannot replace
It would be a mistake to read the comparison as an argument for fewer people. Some tasks need hands and eyes. Valves have to be exercised. Filters and strainers have to be cleaned. Security, vegetation, corrosion, leaks and vandalism are often obvious to a person standing on site and invisible to a sensor bolted to a pump. Many regulated assets have inspection requirements that no monitoring system changes.
The sensible division of labor gives each method the work it does best. Monitoring watches the slowly changing condition of rotating machinery and electrical equipment, around the clock. People handle the tasks that need judgment, dexterity and authority, and the repairs that monitoring points them to. In Program B above, the 240 scheduled visits remain. What disappears are the 480 visits whose main purpose was to find out whether anything had changed.
The crews also change what they carry. A technician dispatched on a finding that says the outboard bearing on pump 2 is showing an outer race defect can bring the bearing, the puller and the right person. A technician on a routine round brings a clipboard and hopes.
How to run the numbers for your own sites
- Calculate the true cost of a visit, including travel time, mileage, crew size and administration.
- Count the visits that exist only to confirm health, as opposed to tasks that need hands on equipment.
- Estimate the cost of an unplanned failure at each class of site, including standby exposure.
- Ask any monitoring vendor for alert precision on comparable assets, and for how alerts can be checked remotely.
- Run the comparison at different distances and precisions to see where the answer changes.
The result is rarely all or nothing. Most operators end up with a hybrid: people for the work only people can do, and data for the hours in between. That balance is the core of remote infrastructure monitoring done well.
