Zscaler’s remarkable predictability

In the recent quarterly call the CFO repeatedly pointed analysts to billings as a very important metric for ZS. And the thing that analysts seemed to focus on was that this was lower qoq and yoy than the monster billings growth of the year before.

**Bill	Q1	Q2	Q3	Q4**
2019	65	115	85	126
2020	88	135	131	195
2021	145	232	225	332
2022	248	368		

**QoQ	Q1	Q2	Q3	Q4**
2018		77%	-26%	48%
2019	-30%	53%	-3%	49%
2020	-26%	**60%**	-3%	48%
2021	-25%	**49%**		

**YoY	Q1	Q2	Q3	Q4**
2019	35%	17%	54%	55%
2020	65%	**72%**	72%	70%
2021	71%	**59%**		

The bolded numbers above in the qoq and yoy was what got the attention: billings growth this quarter was up “only” 59% yoy vs 72% the year before. This was very well explained by the CFO (tldr: government business was slow due to budget constraints and will pick up as budget gets made available), and is not what I would like to dwell on in this post.

Trying to unpack why the CFO continually points us to billings is what this post is about.

First, a couple of things relating to billings. The relationship between billings and revenue depends on how the amounts billed - typically billed in advance by our companies - actually get recognised as revenue. And the CFO gave us one important data point: the duration of billings is, and has been relatively constant for ZS at between 12 and 14 months over time (if it changed, the waters will be a bit more muddy). So if billings duration is constant and revenue is recognised ratably over time, we should be able to very accurately estimate revenue in advance by using billings.

So I tested that assumption, and it is indeed the case for ZS: billings almost perfectly predict revenue.

For ZS, the trailing 4 quarter average billings (the average of the prior 4 quarters) has a 99.78% correlation coefficient with the revenue of the current quarter and the best-fit regression line has an r2 of 99.6% over the last 10 quarters which is really quite remarkable.

Here are the numbers (spot the very obvious similarity between the numbers):

Prior 4q average billings:

2020	98	104	109	120
2021	137	152	176	199
2022	234	259		

And Current quarter revenues:

2020	94	101	111	126
2021	143	157	176	197
2022	231	256		

So what?

It means that we can be pretty damn sure of revenue for next quarter given that the CFO also said that not much changed in the duration of billings this quarter.

Applying the best fit regression line to the last 4 quarters, including this one, should give us an estimate of next quarter’s revenue which will be very, very close to where they will end up, barring something very atypical happening during the quarter.

So. Average Billings for the last 4 qaurters is $293m. And the best-fit regression based on the last 10 quarters (0.973*t4q billings+$4.7m) predicts that:

Revenue for next quarter will be $290m vs their guide of $272m.

Revenue of $290m in Q3 will be good for 13% qoq and 64% yoy, which will again be a new record yoy growth rate. Based on the current tailwinds, I suspect they may even end up a million or two above that.


Long ZS


Hi wsm007,

Thanks for digging into the numbers and your insights!

As you may know, ZScaler is my second largest position after Datadog and I’m glad to see some of your research reinforcing my conviction.

ZScaler is a unique company and has a great moat in Network Security. Their cloud native architecture that was envisioned many years back is built on Zero Trust, Scalability and Dependability.

From a technical perspective I don’t really see any other solution coming close to disrupt them. There was some noise around Cloudflare having found a way to take market share from them but that hasn’t proven to be true yet.

So, great to see these numbers and I’m quite confident that ZScaler will do very well!





What is also interesting is my modeling shows 290 as the expected number for next Q as well but using a different method. My method is to look at the past 5-6 beats and come up with a min, avg and max of those. So for ZS, their most recent beat numbers are (starting from recent; 5.62%, 8.73%, 5.4%, 7.56% & 6.08%). Note, I use the highest number in their guidance range to calculate the beat percentage.

For ZS, we get a min of 5.4% beat, average of 6.68% beat and max of 8.73% beat. Applying these numbers to predict revenue for next quarter comes in at min = 287M, average = 290M & max = 295M. I have a charts and graph that show this that I posted on Twitter if anyone wants to see it.



One might assume that with the Russian military actions and their penchant towards cyber attacks, the Government might come up with cybersecurity funding fairly quickly.