no one was ready for it yet

gistfile1.txt

For example, I was co-founder of a startup called Neowurks (yeah shit name)
about 10 years ago.

It was an app for the telco industry, in the area of rates exchange and route
provisioning.

At least once per week, all the telcos would exchange their rates for
terminating calls at major cities around the world. Pricing was per million
minutes, at 4 or 5 levels of quality of service and for mobile versus landline.
The medium of exchange was *usually* a spreadsheet. Each one would be unique
- there was no standard interchange format. However, each spreadsheet would be
accompanied by a fax. The fax was what made it all a 'legal document'.  There
was no going back on the published prices.

Each telco would have one or more routing guys (depending on how big they were)
that would take a look at the spreadsheets and find the best deals on traffic
termination and make the corresponding routing decision.  That routing decision
would then be written up and sent off to the switch engineers. There was often
a 2 week lead time from making a new routing decision and it being provisioned
on the network.

Well, when you have hundreds of telcos and each publishing terminations to 500+
destinations at a number of QoS levels for mobile and landline traffic adding up
to millions of data points, it pretty much becomes impossible to spot the best
deals. A single engineer at [redacted] would typically make less than 5 routing
alterations per week.

The app was part ETL tool part arbitrage opportunity spotter. It would bulk
import all of the spreadsheets, and compare them against each other and the
currently provisioned routing setup and make hundreds of new routing decisions
in minutes.  When the routes were confirmed by the user, it was capable of
automatically provisioning them by talking to the switches directly, making the
provisioning time down from weeks to minutes.

The interesting thing is that as well as spotting plain good deals in the
spreadsheets, it could also spot mistakes such as classifying local traffic as
international, typos in prices etc and create arbitrage opportunities.

We did an analysis of how much BT would have saved and it ran to the tune of
low-hundreds of millions a year in the 'just pick the best deals and provision
them within < 1 day' scenario and that was before mistakes etc were taken
advantage of.  Needless to say, having a meeting with top level management and
saying 'we will save you millions of dollars' is so beyond expectations that you
get essentially laughed out of the building. 

We were funded by Sonera who was basically on a deal where they fund us and they
get lifetime use of the software and we got the IP. They got bought out (I
forget by whom) and the project was basically killed when we failed to drum up
investment and ran out of our own cash.

The project wasn't the problem - how it was sold was. I'm not saying it was too
advanced, it's more that it was oversold. That and we were trying to sell to
huge corperations that had entrenched culture and way of doing things.

If we claimed much less just to get our foot in the door the story might have
been different. I'll call it naivety. We were close but no cigar.

Yeah, just read the earlier discussion. That clears things up. Windows ME was ahead of its time. Riiiiiiighht.

Although my problem was not the same as the Windows ME problem. Windows ME was fundamentally terrible. Neowurk's product actually did what it was supposed to. Very well in fact. We just killed it in the sales process, when we should have aimed much lower with our claims.