It all started so well
I was employee number three at a Startup with the usual, larger-than-life Founder. He was a man with a vision, a big vision, able to pursuade, cajole and sell his vision to pretty much everyone around him including me.
It was a good “tech-for-good” idea and we were all convinced of our mission to make the world a better place by helping the public complain to big companies.
We started building and had quite some success. Thanks to the endless hard work of our founder we managed to get the contact details of large companies across the UK from help desk through to head of customer care and even the CEO. At the time we were getting about 100,000 visits from users coming to the site, looking for advice and then using our platform to make a complaint (we called them cases). We helped people by guiding them through the process and reminding them when it was time to take it further. I was amazed at just how many UK companies in this day and age still seem to think that if they ignored people most would just go away.
The next investment round
The next investment round was looming and we were looking to improve our numbers. The increasingly early Friday afternoon “all hands before drinks” arrived and our Founder ran through his strategy for growth based on some high level numbers. We were getting about 20,000 cases a month off the back of a few thousand companies. Over the next few months we’d double those companies thereby doubling our cases. There was a round of applause from 20 or so people draped across bean bags, high on the positive energy.
At the time I knew very little about data - I was a web developer; people click on things, my code takes that request then gives them something back in a few milliseconds but I had come from company where a similar “let’s just double everything” assumption had failed spectacularly. Despite my two Friday beers and a schnapps I decided to test my statistical knowledge.
The data told a different story…
Exporting some data to CSV, I created a simple graph in Google Sheets. One axis was each company and on the other was the number of cases from that company. It was a profound “J curve”. Yes we had thousands of companies but 95% of our cases came from less than 80 of those companies. Doubling the number of companies would, most likely, do little but add to that very long tail of organisations for which few people created cases - if any.
The data was telling us a story. The story was that we’d get better results learning how to get more cases from the thousands of underperforming companies we already had than just blindly adding more.
We abandoned the notion of doubling our company base but I’m not sure we ever really got into the data enough. If it were my company I’d double down on doing what it takes to make the next 80 companies as succesful as the first but we never went down that route.
Thats a different story