How Do You Upgrade a Live Site Without Closing It?
Quick answer. Plan the outage, not just the install. Find the hours when the site is really empty, then work backward from the time it has to reopen. Write down every appointment in that stretch, including the ones that happen after your crew has gone home. On multi-site programs the repeatable pattern is one overnight: disconnect and set the new equipment the evening the site closes, then open the next morning on the new equipment and verify it with network support. Zero downtime during business hours comes from the schedule, not from working faster.
The problem the IT buyer feels
The branch opens at nine. The cabling is undersized and the switch is at end of life, but nobody is going to close the lobby for a day while a vendor reworks the wiring. So the work gets squeezed into the gap between closing and opening. That gap is where every assumption from the scoping call gets tested.
The install itself is rarely what goes wrong. The trouble is everything around it. The building locks before the work window starts. A badge works on one side of the floor and not the other. A second vendor owns the disconnect and the reconnect, and only one of those appointments got booked. And if the doors open with a lane still dark, your name is on the escalation, not the vendor's.
Why this matters as you scale
One site is a logistics problem you can handle with attention. Forty sites is harder, because every building has its own rules.
Three sites in one program recently had three different ways to get in after hours. At the first, the tech signed in at a loading dock and got a temporary elevator badge that had to be returned before leaving. At the second, the alarm on one side of the floor armed itself at midnight, so the crew had to be out by then whether the work was done or not. At the third, the tech had to be inside the building before 6:00 PM or the elevators locked out, even though the work itself started later.
That last one is common. The time you have to arrive is often earlier than the time the work starts, and almost nobody writes it on the work order.
These differences affect the price. After-hours work costs more because it squeezes the schedule. On one quote, the same site and the same scope came out about 31% higher at night than it did during the day. So it's worth checking, site by site, which jobs can move back into the normal workday.
The other cost is easier to miss: sites slowly stop matching each other. When a tech is working alone at 2:00 AM and something doesn't fit, he fixes it his own way. Each fix makes sense that night. After enough nights, no two sites are built quite the same.
Here is a real example. The drawing called for a cable to run through a specific pass-through. Another piece of equipment next to it was pinching the cable, so the tech ran it a different way. It worked that night, but it wasn't what the drawing called for. We had to go back and fix it, and then we found the same problem at another site.
What changes when one partner owns the site
The question that matters is who answers for the morning. If the same company handles the cabling, the hardware, the disconnect and the reconnect, then the people checking that the site is live are the ones still standing in it. Nothing gets handed to someone else at the moment it's most tempting to leave.
Every site also comes back with the same documentation format. The next cutover in that market starts from a record of what's already in the walls, not from a fresh survey.
What we see in the field
The clearest example in our own work is a national modernization program. It covered more than 630 locations, averaging about 5,000 square feet each, all rebuilt around equipment that was still running. There was zero downtime during business hours, and every site reopened on schedule to the same documented standard.
What made it work was not speed. We turned the outage into a repeatable routine, built around one overnight:
- Day 1, 4:00 PM: the site closes. We disconnect the old equipment and set the new equipment that evening.
- Day 2, 9:00 AM: the site opens with the new equipment live, and we verify it with network support. That is about 17 hours after the disconnect.
That is the whole outage: one evening and one night, with the site never open on dead equipment. You staff for that cycle instead of for a generic "install day."
The full job runs longer than the overnight, typically three to five calendar days. Part of the reason is that a separate security vendor often handles the disconnect and the reconnect of the security equipment, and those are two separate appointments, usually about two days apart. When that vendor is involved, the job stretches to five calendar days. The reconnect is the appointment everyone forgets, because it happens after the equipment crew has already left. If it isn't booked, the site can't go live even when the cabling and hardware are done.
We also learned two things the hard way.
The badge. A tech got into one suite before 5:00 PM, stepped out, and couldn't get back in. His badge only worked on one side of the floor, and nobody from security was there after hours to let him in. Half the floor got done. The rest needed a second trip. Now we test that a badge works for getting back in, not just for getting in.
The job that looked finished. On a device refresh, the hourly log looked fine until 3:00 PM. Then a third-party network vendor left, saying he would finish the setup remotely. By 4:00 PM there were no link lights on the equipment. By 5:00 PM our crew had left, and the site sat without a working connection all night. Now nobody leaves until the network team confirms the link lights are on.
Where to start
Here are three checks you can run on your own sites this month.
Pull your last five after-hours work orders. See whether each one lists the time the building locks, how the alarm behaves on that floor, and whether badges have to be returned. If those fields aren't on the form, add them.
Take one upcoming cutover and list every appointment in it, including the reconnect and the final check that the site is live. Make sure each one has a date and a named owner. The gap is usually the reconnect.
Look for after-hours jobs that could run earlier. An 8:00 AM start, before the site is occupied, can move some of that work out of the after-hours rate.
Frequently asked questions
Can low voltage work really be done with no customer-facing downtime?
Yes, if you schedule the outage instead of absorbing it. Across more than 630 modernized locations, the work happened while each site was empty, and every site reopened on schedule. The limit is the access window, not the labor.
How long does a live-site cutover actually take?
The outage itself is one overnight: about 17 hours from the after-close disconnect to the next morning's 9:00 AM opening, when the new equipment is live and verified with network support. The full job typically takes three to five calendar days. It takes five when a separate vendor owns the disconnect and reconnect.
Why does after-hours work cost more?
Night work means crews rearranging their schedules to hit a tight window while the site is closed. On one quote, the same scope was about 31% higher at night than during the day, so it's worth asking whether the work really has to happen at night.
Can one partner cover a national footprint on night windows?
That is what Comletric is built for. We have 230 service hubs across all 50 states, do low voltage work only, and provide photo documentation and site sign-off on every job. A cutover in one market comes back in the same format as the last one.
Next step
If you want a clear read on where one of your sites stands before you schedule a window around it, we will assess it at no cost and send a written report with photos and cost estimates within 30 days. You can request one at https://www.comletric.com/assessment.