Abandon Ship: How a startup went under
Using data, signs the ship was sinking became clear.
I feel like this topic is easier to discuss now that the company went under. I previously worked at Digital Signal Corp (DSC) for 4 years after a four month internship. I really did enjoy the technology and the people, and when I started, there were just 23 people. When I left, there were 137 people. This is a story about why I chose to leave when I did and the data I used to make that decision, specially the data I had collected to approximate turnover rate.
A Brief Company History
The company saw tremendous growth in my four years there, but there were some warning signs that the company wasn’t going to do well for long. The product was very cutting edge: a long-range 3D face scanner that could scan your face from 15 to 25 meters away while you’re walking. They had been getting by on grants and small investors, and after I started, they got some serious investment.
I thought we were on track to hit the big time because the Saudi Arabian government invested $100 million. We grew quickly from there, but it seemed to be growth to fill out an organizational chart, not to push the product to market. Most of the leadership positions even in engineering were brought in from outside, and few promotions were given internally. This didn’t concern me initially, but as the months turned into years, I wasn’t any closer to a promotion or increased responsibility, and the yearly compensation increases were getting old as they just tracked inflation.
This is a photo of the first shipment of systems (I’m not in it, FYI). This should have happened a year or two earlier, but there were always delays. It turned out to be one of the rare shipments of product. The lack of products moving out the door was not a good sign either.
Turnover
The main sign of health to me was the turnover rate. This was not a published statistic. In fact, unless someone went around to say they were leaving, I usually found out after someone left. This bothered me, and I began to see some warning signs. I noticed in manufacturing, there was a high turnover. They started cutting some benefits, and people started to grumble.
For software engineering, it was different for a time. In other departments, it seemed the turnover rate was high, but for us, there was no turnover in three years. This too came to an end slowly, and then quickly, everyone tried to get out as soon as they could.
One engineering manager left after nine months, which seemed a little odd but the reasoning we heard sounded okay. Then performance reviews came, and even though we had a lot of funding, all the raises were minimal. This seemed especially unfair given that the execs were flying first class to Saudi Arabia all the time. It didn’t quite add up.
By that time, I had already been interviewing, but I wasn’t in a big rush to leave until a friend left four weeks after his performance review. One month later, I gave notice. Within the next 11 months, half of software engineering had left, and the first of three rounds of layoffs started.
The Data
Unbeknownst to anyone, I had discovered a year prior to leaving, how to measure turnover. I was too curious not to track it. The method was simple: track the phone list.
A new phone list was sent out every time someone new started, and eventually, once a month. I wrote a script to check who was on it and who was no longer on it. Then I tabulated a rolling turnover rate for the past year. This evidence only confirmed my gut feeling that turnover was very high (20%), and this was masked by so many new hires. Only someone who was there for a few years would notice, and even this population was dwindling.
I took this data, and I’ve done some data cleaning. First, an employee died of a health ailment, so he shouldn’t be counted as turnover. Second, a small group of 11 people split off to form DSCg, so they were also not counted either. I then determined the department people were in using memory and LinkedIn.
Analysis
The first sign of difficulty was seeing a high turnover in Manufacturing. I didn’t plot by department originally, but when I look at it now, clearly, there was a problem. These plots are 1 year moving averages at one month intervals, and I have also included years turnover. Years turnover means how many years of experience at DSC (or the specific department in DSC) was lost due to turnover in that part year of the moving average.
The trend is remarkable in that manufacturing essentially turns over in one year even though they kept hiring, people kept leaving. Engineering on the other hand was very stable near the 11% industry average. The trouble comes when the business experience starts leaving mid-2013.
I also looked at 1st year turnover and greater than 1 year turnover. Less than one year shows people who figure out something is wrong and just eject. More than one year shows people really coming to their senses and getting out. Again, the trend is really bad on the business side for 1st year turnover. Had I seen this, I may have been inclined to leave sooner. The business folks are in it for the money, and they have a pulse on whether money is being made.
Post-leaving: I looked at just Engineering of which Software Engineering is a sub-group. I used LinkedIn to give a clue on when people left. It looks like right before I left, turnover for Software Engineering was non-existent. People were happy. Over the next year though, quite a few people left and then layoffs started.
Warning Signs
Here are some warning signs that have been derived from this experience ordered by priority:
- Lack of products shipping
- High turnover
- High 1st year turnover (especially sales)
- Senior engineer turnover
- Leadership positions not filled from internal promotions
- Lack of promotions in general
If you like, follow me on Twitter and YouTube where I post videos espresso shots on different machines and espresso related stuff. You can also find me on LinkedIn.
Abandon Ship: How a startup went under