How FP&A Forecasts the Unexpected

Building a financial forecast can be difficult but building a financial forecast when the world Is facing an unprecedented global pandemic is a completely different story. To learn from the best, I sat down with Doug Brewster, Sr. Director of FP&A at Confluent, to learn more about how he and his team forecasted when Corona first hit the United States. He walked me through the framework that they used at Confluent to build their scenario planning.

Here's what Doug had to say:

When Coronavirus really started coming to the states, in February of 2020. Sequoia Capital had sent out this memo, and some of these other investment firms had sent out similar memos, warning founders and management teams to plan accordingly and to start building out scenario analysis.

So, we kicked off a lot of scenario planning to look at what the impact would be on our top line. We looked at different booking scenarios, taking it as a percentage of what our original plan was and then also looking at what the impact would be on our new ARR bookings growth.

Obviously, no one knew what was going to happen, and at the time we knew less than we do now. But once we had a handle on the different scenarios that we created, then we decided on what our base case was and then we figured out what the impact would be to our cash. For example, because bookings, we assumed, would come down, we assumed billings and cash collections would come down.

For a venture-backed company, cash is king. And it was no different for us. So, we figured out: "okay, what's the delta in our cash burn for the year and how do we solve to get back to close to where we were for the plan? Then we started looking at our expenses and figuring out: what are expenses that are not necessary and what we can just stop altogether? What are some expenses that are necessary, but we can slow down or at some point and reverse the spend if we need to?

This was everything, from meals in the office. No one's in the office, we're not spending on snacks or meals, so we can stop those for now. That's a reversible item, that when people go back in the office, if that happens, then we can talk about that later. We could always turn that back on, so to speak.

Then there are things that we just absolutely can't not pay: we have salaries and we have rent, and we have hosting costs and utilities, and all those sorts of things. Thinking about the expenses in that way, then we figured out: "okay, what do we need to do to solve for the cash burn number that we're trying to target?"

For Confluent, we haven't had to do layoffs. Which we feel very, very fortunate about. We have slowed down our hiring, but if you look at our job board, I mean we still have probably over 100 job requisitions posted. We're still growing very quickly; it's just slowed down a bit.

After we did all that scenario planning then we basically figured out how to operationalize it. It was very much like a tops-down on the bookings and revenue and billings and collections portion. Then it was a little bit bigger buildup on the expense side. When we got to our Q2 outlook at the end of March and early April, we did a more thorough bottoms-up on the expense side of things to make sure that the numbers we had in our scenario planning were realistic.

If you want to learn more about how Doug's team used recession data to improve the accuracy of their forecasts, check out our blog post, "Using data in worst-case scenario planning."