Skip to content
All insights

The CrowdStrike Lesson: What One Bad Friday Taught Small Business About IT Dependency

Matt Wood · Tech Inc

The short answer

The CrowdStrike outage of July 2024 was not proof that businesses depend on technology too much; it was proof that most businesses depend on technology nobody in the building can fully explain. The real risk exposed that Friday was unexamined complexity: stacked software agents, redundant tools, and single points of failure nobody had mapped. The durable fix is a shorter list: fewer, well-understood components, recovery that has actually been tested, and a provider who can explain every piece of your stack in plain English.

One bad Friday

On Friday, July 19, 2024, a faulty update to a widely deployed security product began crashing Windows machines around the world. Flights were grounded. Broadcasts went dark. Checkout lines stopped. And in plenty of ordinary businesses, the morning began with blue screens and an honest question: what is this software, and why is it on all of our machines?

The recovery made the point sharper than the outage did. There was no instant remote fix; affected machines needed hands-on attention, one at a time. Businesses that had a map of their environment and a practiced recovery routine got back on their feet in hours. Businesses that had neither got a very long weekend.

Tech Inc's founder, Matt Wood, wrote about the outage that same month, in a piece on IT dependency and what the day actually taught. Two years on, this article revisits the theme, because the lesson has only become more relevant as businesses add AI on top of everything else.

The wrong lesson

The tempting conclusion was the one in every headline: we are too dependent on technology. It is also a useless conclusion. A modern business cannot un-depend. Payroll, scheduling, communication, invoicing, production: all of it runs on software the same way all of it runs on electricity. Nobody responds to a power outage by resolving to use less electricity. They respond by understanding the dependency and planning around it.

Dependency was never the failure. Understanding was.

The right lesson: unexamined complexity

What actually failed that Friday, in business after business, was not "technology." It was a specific, nameable structure: deep-access software multiplied across every machine, updating on its own schedule, inside a stack nobody had examined in years. Three patterns deserve the attention.

Stacked agents

Every security and management product you adopt installs an agent: a piece of software with privileged access to the machine, its own update channel, and its own failure modes. Agents are how modern IT works; they are not the villain. But each one is simultaneously a capability and a liability, and the liability compounds quietly with every addition. A stack assembled over years of "let's add one more layer of protection" is also a stack with more ways to fail all at once.

Redundant tools

Two tools doing overlapping jobs feel like double protection. In practice they are double the update channels, double the conflicts, double the spend, and double the surface for exactly the kind of failure the world watched that Friday. Redundancy like this rarely arrives by decision. It arrives by drift: each tool solved a real problem in the year it was added, and nothing was ever retired.

Single points of failure nobody mapped

Every business has single points of failure. That is not the scandal. The scandal is how few businesses can name theirs. The outage posed a question most owners had never been asked: if one vendor's bad update could halt your operation, would you have known in advance which vendor, and what your first hour of response would look like? For most, the answer to both was no, and not because the answer was unknowable. Nobody had been asked to map it.

The fix is the shorter list

The answer to unexamined complexity is not zero technology, and it is not another tool to monitor the tools. It is a shorter, fully understood list.

Fewer, well-understood components

The protections a small business actually requires form a short, known set: endpoint, email, identity, and backup, with monitoring across all of it. Everything required. Nothing piled on. Any component beyond that core set should have to justify itself against your business's real needs, not against a vague sense that more layers mean more safety. The Friday in question is the standing counterexample: a layer of protection was the thing that failed.

Recovery that has been tested

A backup that has never been restored is a hope, not a plan. The businesses that recovered fastest that day were the ones for whom recovery was a routine rather than an improvisation: they knew what ran where, who did what, and in what order. Test the restore. Rehearse the bad morning on your calendar, before it schedules itself.

A provider who can explain every piece

Here is a test you can run this week, with us or with anyone: ask your IT provider to list every agent running on your machines, what each one does in plain English, and what happens when each one fails. A provider with an examined stack answers from documentation, without jargon. If the answer is a pause, a buzzword, or an offer to get back to you, treat it not as an embarrassment but as a finding, and the most useful one you will collect all year.

The quiet compliance dividend

There is a second payoff to the shorter list. Sooner or later, everyone answers to somebody about their technology: an insurance carrier at renewal time, a larger customer pushing security requirements down its supply chain, or a regulator, with the FTC active and Colorado's own AI law arriving. Every one of those conversations reduces to the same demands: document what runs in your environment, explain why it's there, and show how you recover. An examined, well-understood stack answers those questions almost for free. That is why Tech Inc weaves compliance readiness into the core service instead of selling it as a separate, fear-priced product: a business that can explain its own stack is already most of the way ready.

FAQ

Was the CrowdStrike outage a cyberattack?

No. It was a defective update from a security vendor: protective software crashing the very systems it protected. That is exactly why the lesson is about complexity rather than threats. The tools you install to reduce risk carry operational risk of their own, and that risk needs to be examined and planned for like any other.

How many tools does a small business IT stack actually need?

Fewer than most stacks contain. The required protections form a short, known core: endpoint, email, identity, and backup, plus monitoring and support across them. The count matters less than the criteria: every item on the list should be understood in plain English, do a job nothing else on the list does, and have a known recovery path.

How should a small business prepare for the next major IT outage?

Three moves. Map your dependencies, so you know which vendors and systems can halt your operation. Test your recovery, including an actual restore from backup, not a checkbox. And require plain-English answers from whoever runs your IT, so the stack stays examined instead of drifting. None of that requires buying anything new. It requires discipline.

See your own list

If reading this made you want to look at your own stack, that is precisely what the 15-minute walkthrough is for: the entire Tech Inc process applied to your business, the real pricing line by line, and an honest verdict on where your stack stands. No pitch. Or start by asking Ellie, the AI concierge on every page at techinc.co. The plain-English test works on her, too.

Fifteen minutes gets you the whole picture — with your numbers.