I first learned about the power of simplicity in basic training. I was a private. We weren’t encouraged to think.
The drill sergeants taught us simple, repeatable processes for everything. They called it the KISS principle: Keep It Simple Stupid. How to make a bed. How to do a push-up. How to clear a jam in your rifle. You did it the same way, every time, until you didn’t have to think about it. Then you did it some more. And almost everything was timed. Faster was better.
The phrase was “Train like you fight, fight like you train.” Repetition was the point. Speed was the other point. If you ran the drill enough times, your body did it under stress, fast, without your brain having to weigh in. And the soldier next to you did the same, because he’d run the same drill. You could trust each other to move without checking.
The same logic holds in business and tech. Complexity comes in many forms: tools, headcount, processes, integrations. Every moving part is another place to fail, another bill, another hour spent keeping the parts working together instead of doing the work. Eli Goldratt put it plainly in The Goal: the goal of a business is to make money. Complexity steers you away from the goal.
The drills weren’t invented overnight. They came out of decades of warfighting and millions in research. Simple isn’t the same as easy.
When I was at a relatively small cybersecurity company, we bought the full enterprise stack. Salesforce. HubSpot. ServiceNow. The pitch was that we’d be empowered to compete with the big boys. I mean, what successful company doesn’t run Salesforce? The reality was endless debates on whether data should live in Salesforce or HubSpot, how it should flow back to ServiceNow, what the schemas should look like, who owned what record. Then came the chore of navigating ultra configurable data models to integrate everything together. The license fees were absurd for a company our size, but they weren’t the thing costing us the most. The expensive part was opportunity cost. The human time we spent bending these tools to behave instead of improving the services we provided to customers.
The licensing cost is the small number. The drag on your team is the big one.
The engineering side had the same disease. At an HRTech company earlier in my career, our first answer to any technical problem was to deploy the hottest technology of the day. The young engineers want to work on the new, cool stuff. What we didn’t price in was the cost: knowledge ramp-up, troubleshooting time, fragility. I was just as guilty as anyone. I shipped Elasticsearch instead of fully leveraging search functionality we already had in our existing tools, and learned the hard way. Same disease at the cybersecurity company, different decade. We were heavily invested in serverless, stringing together Lambda functions with message queues and constantly fighting timeouts. A monolithic server on a VM would have done the job and freed up months of engineering attention. And we weren’t the anomaly. Almost every company falls into the same traps.
What if money isn’t the constraint? If you’re at a company flush with capital, you can fight complexity by just hiring more people, right? Wrong. People bring more complexity than most tools: management overhead, training, personal drama, HR, payroll. Every head you add multiplies the coordination cost, and most teams don’t factor in that price either.
A couple of years ago I co-founded a company called Mobility Places. Our team is intentionally small and speed is our key competitive advantage. I wrote recently that AI’s real win is shrinking team size, not making individuals faster. What I didn’t say there is the other half: small teams stay small only if the stack stays small too. Compression on one side without the other gets you a two-person company drowning in a fifty-person company’s toolkit. Taking everything I’ve learned, I deliberately skipped the latest and greatest. We run what some would call an antiquated MVC architecture. It’s been around for years and just works. No Kubernetes. Very few discrete services. No document databases. No over-engineered data pipelines. No heavyweight JavaScript frameworks.
Even with just a couple of engineers, the setup coupled with AI lets us move fast. Really fast.
We don’t run Salesforce. Instead of buying a big-company CRM, we built our own and integrated it into the platform. It does exactly what we need and nothing else. If we want a new feature or integration, it takes hours not months to implement. No license fees. No consulting engagements. We only build the things genuinely useful to the company.
A few weeks ago we noticed our customers going back and forth with us over email to tell us where they needed new parking. They’d describe a neighborhood. We’d ask follow-up questions. They’d send a screenshot. We’d go look. An afternoon of work later, they had a feature that lets them specify the exact geo and requirements themselves. Customers love it. They can tell us exactly what they need, and we can start sourcing immediately.
Try shipping that on top of Salesforce in an afternoon.
Our non-sexy application stack is easy to extend, easy to troubleshoot, and it gets the job done. We can deliver a new feature in hours. If something breaks, we can fix it just as fast. The team stays small because the stack complexity stays small. When we do add headcount, training will be straightforward, because there’s no labyrinth of integrations and bespoke configurations to onboard somebody into. Same idea I picked up in basic training. Simplicity is what keeps you fast under pressure.
The honest objection is that I’m conflating necessary complexity with chosen complexity. Some businesses do need the full stack from day one. Regulated industries. Multi-region operations. Global enterprise organizations. Compliance-heavy verticals. Real constraints, not vanity. Fair. But that’s not most teams, and most teams still buy as if it were. When they evaluate a new tool, they weigh the features. They weigh the benefits. They don’t weigh the drag. Sales pitches help, but the real reason is the one I raised earlier: they want the company to look like a real company. They want Salesforce on the bill. Price in the real cost before you sign.
Novelty is fun and engineers love to tinker. Using the same tools as the big guys feels good. Neither is the job. The job is to solve customer problems and build a profitable company.
My drill sergeants didn’t give out points for the most creative way of handling an M16 rifle. They gave them to the soldiers who could successfully clear a jam by the book in the fastest time possible.
Keep it simple, stupid.