Of all the many things I got wrong when I started my company, my original revenue forecast was right at the top of my mistakes. Not only was my forecast off, it was off by a lot.
Instead of the $1 million first year revenue I had forecasted, we did about $100,000. My board of directors was not surprised by my forecast miss, but they weren’t happy about it either.
Most startups miss their initial forecast, and I was no different.
Barry, the chairman of the board of my company had a funny phrase he liked to say about startups and their forecasts. It goes like this, “Most startups miss their forecast, and you’ll find we’re no different.”
Barry’s statement is funny when it’s not your problem. However, it was my problem in my case, and I needed to fix it.
My board, correctly, asked me to explain what was going on.
You own your forecast mistakes, regardless of the cause.
We entrepreneurs are an optimistic bunch, aren’t we. That explains part of the reason I was off on my forecast, but it doesn’t explain the whole reason I was off.
Your initial revenue forecast is likely just a model.
I had been in the analog semiconductor industry for close to 20 years when I started my company. I knew how the typical product’s revenue grew over time, so that’s what I used as the basis for how revenue would grow in my model.
Knowing the revenue stream is only one part of your forecasting model. There are a few other pieces you need. The first piece you need is how quickly does it take to develop a product, feature, or revision to an existing product, and how many engineers will it take per product.
I leaned on my experience and Jeroen’s, my co-founder and VP Engineering, experience to figure out the second piece of the model. We built into the model a success rate of 50% for “working first silicon”, so we were expecting to have rework half our products.
The third piece is your hiring plan for engineers and supporting staff. In our case this would be test engineers, applications engineers, and layout.
These three constraints determine your model for how many products and features you can get to market in any given time frame.
Then you go from your model to an estimation of what will happen.
The next step in our process (you may be able to skip this step, but we couldn’t) was to go from generic products to specific products. We would use the generic model for future products that we hadn’t decided upon yet.
This was the revenue forecast we submitted to our board of directors.
Then you hold your breath and see how wrong your model is.
It’s one thing to have real world experience like I did, and it’s another thing to actually see what will happen in your new reality, with a new team, and in a new company. That’s where the fun begins.
One year into being in revenue, we were significantly off our plan, and our board wanted to understand why. The exercise was very good for us to go through.
You can start by attributing the errors in your model to each of your initial assumptions.
As I said, this is where the fun begins.
Are you familiar with the french phrase “force majeure”? It means, “unforeseeable circumstances that prevent someone from fulfilling a contract.” Well, we had some unforeseen circumstances, to say the least.
We had two product families that would account for all our revenue in our first year. We were actually ahead of schedule with our first product family, and it came back from our fab, TSMC, with an inverted mask layer, caused by a mistake by TSMC .
The mistake delayed the launch by four months, causing a revenue miss of $300,000 in the first year. The second product family was late by two months and this cost us about $200,000, so $500,000 of our $900,000 miss was due to product delays.
The other $400,000 was forecast miss. The question was why did we miss the forecast?
You need to fix your forecast based on reality.
You never know how a product is going to be received by the market until you actually sell it. The good news was the customers that knew, the key word is “knew”, about our products really liked them.
More importantly, the customers were buying our products. So what caused the miss?
I had forgotten that we were fighting a two front war with second front building an audience for our products. I updated our forecast to reflect the reality of where we were.
You can use waterfall charts to show your forecast changes.
If you don’t know about waterfall charts, they are a great way to show the changes in your forecasting. Here’s what a waterfall chart looks like if you’ve never seen one:
The one above is for our operational expenses. The top line is the original budget. The gold boxes going diagonally are the actual numbers. The rows are the monthly plan for the year, and the columns are the month for the plan.
The idea is that you and your board can see how your plan has evolved for the various areas of the company over time. You can see here that we started with a plan of spending $3.233 million for the year, and we are going to spend around $3 million.
Then you rinse, repeat, and do it again.
Accurate forecasting is all about learning from your previous forecasts, adjusting, learning, and doing it again and again. That’s it.
Then, get in the habit of underpromising and overdelivering on your revenue forecasts. Take the forecast that you and your team develop, and maybe by 20% cut it.
Make this new number your board commitment. This gives you a buffer as you push your team to the original number.