You create a good design, test it, and put it into production.
Granted I've been doing this since Bill Gates was a kid, but seriously, muddling through it is the wrong approach.
Reading this case study I see two issues:
1) if you don't start with a plan with solid requirements, you can really waste time trying to make a Toyota Corolla ready to compete in the Indy 500.
2) Often businesses and business owners have really no clue about technology. Assuming you can run your business on a $20 router with a 50-cent power supply is naive, at best.
You get what you pay for.
If, at day one, you said, look, here's a Cisco router and a Cisco WLAN system. It will cost you $1500, but when I come back in a year, it will still be running, and it can be configured or re-configured to do just about anything.
If you cannot afford it, here is a lower-cost Sonicwall, Aerohive, or Aruba solution.
Trying to use consumer-grade equipment for a business is a huge mistake. Most of it does not have the feature-set you need, and most does not have the stability and reliability/redundancy features that are also needed.
Been there, done that.
To race in the Indy 500 you start with a race car and refine it, don't try to muddle through to try to make your Honda Civic look like a Honda Indy car
Keep Up with TechRepublic