Back in 1995 I had many Unix Systems, but I got my first version of Microsoft’s New Technology platform: Windows NT. There were no internet applications for it. Apparently the fact it even had internet capability was a last minute addition. Back then Microsoft had yet to discover the internet…
Email was already a key communication method but organisations needed specialist computers from HP, Sun Microsystems or IBM and expert staff to run email packages such as sendmail, postfix and pmdf. They were either expensive and/or difficult to configure (hire an expert!). I needed email and I couldn’t get any of them to work.
So, one week in 1994, I download the specifications for email (RFC821 and RFC1939) and wrote an email server. An engineers’ solution! It was dirty and quick but worked. Version 0.1 of NTMail was less than the Minimally Viable Product (MVP). Three weeks later key coding was added and a fledgling list of options were implemented and the first customer – Falmouth College of Art in the UK – purchased the first copy of version 1.0. I considered the software difficult to manage – it required direct editing of text files to make changes (but it was much easier than the Unix offerings). The software worked efficiently and quickly with minimal system resources. It was the first email server written for Windows NT.
Why did people buy it? It was cheaper and faster than the alternatives but not yet better:
- Cheaper than the competition (no need to pay an expert nor purchase Unix hardware which was typically 2x or 3x the cost of a Windows PC).
- Faster because it used new programming techniques to reduce the resources required to handle multiple events at the same time.
- Not yet better because it was difficult to manage.
The next two months I spent making it better. I hired people, designed a configuration interface, published manuals and wrote the company’s own Customer Relationship Management system. By making it easy for customers to use and buy, I had created a Maximally Sellable Product. I increased our version number to 2.0 which was the traditional way to show product maturity.
The basic architecture was to last three years but during that time thousands of copies were sold allowing the company to rapidly expand and re-engineer the core code. That core code is still in use today, fifteen years later. But that is another story…