Here is a scene I have watched play out at more small businesses than I can count. The internet drops for the third time this quarter. Someone shrugs and says “it’s probably just the router,” reboots it, and moves on with their day. Nobody writes it down. Nobody asks why a “router problem” keeps happening to the same router at the same time every month. This is not resilience. This is a company quietly agreeing to be surprised by its own infrastructure on a recurring basis, and calling that normal. If you have started Googling MSP Calgary options during your lunch break instead of during a calm strategic review, that itself is a symptom worth noticing.
Growth does strange things to IT setups. The scrappy arrangement that worked fine at twelve employees, a shared drive, one person who “handles computers” on the side, a password manager nobody quite trusts, starts groaning at thirty employees and can genuinely endanger the business at sixty. The tricky part is that the warning signs rarely look like a crisis. They look like mild annoyance. That is exactly why they get ignored.
Take ownership, or the lack of it. Ask five people at a mid-sized company who is responsible for cybersecurity, and you will often get five different answers, ranging from “IT, I guess” to a long pause followed by “that’s a good question.” A good question is not a good answer. If a phishing email lands in twenty inboxes tomorrow morning, someone needs to already know they are the one who acts, not the one who forwards it to someone else who forwards it again. Diffuse responsibility is how a fifteen-minute problem becomes a three-day one.
Then there is the repeat-outage phenomenon, which deserves its own case study somewhere. The same server hiccup. The same VPN that drops every Tuesday afternoon for reasons nobody has bothered to trace. Businesses will tolerate a recurring fault with more patience than they would tolerate a single new one, which is backwards. A problem that happens once is bad luck. A problem that happens three times is a policy, whether anyone meant to write it or not.
Software sprawl is another tell. Walk into the finance team’s world and you find one invoicing tool. Walk into sales and there is a CRM nobody in finance can see. Marketing has its own project board that syncs with nothing. Each tool was a reasonable purchase in isolation, made by a reasonable person trying to solve a real problem quickly. Collectively, they add up to a business that cannot get a straight answer out of its own data without someone manually stitching spreadsheets together on a Friday afternoon.
And then, the quieter one: the new hire who takes two weeks to get proper access to the systems they need. Not because anyone is being difficult, but because provisioning a new employee is still a manual scavenger hunt through old email threads and someone’s memory of who has admin rights to what. A company that has outgrown its setup does not usually announce it with a dramatic failure. It announces it with friction, repeated, low-grade, and easy to explain away one incident at a time.
There is also a subtler sign, one that rarely gets discussed because it does not look like an IT problem at all: decisions slowing down. A leadership team wants to open a second location, or add a shift, and the first honest answer from whoever handles technology is “let me check if we can actually support that.” Not because the idea is bad, but because nobody has confidence in what the current setup can absorb. Infrastructure that was never planned, only accumulated, cannot tell you its own limits, and that uncertainty shows up as quiet hesitation in rooms where the real constraint never gets named out loud.
The honest test is simple. If your last three IT problems were each blamed on something different, when they were actually the same root cause wearing a different hat, you have outgrown whatever informal arrangement got you this far. That is not an indictment of anyone’s competence. It is just what happens when a business scales past the setup that was built for a smaller, simpler version of itself.
Fix it before the fourth outage, not after. The fourth one is rarely as forgiving as the first three were.
