Snowflake's customers

SNOWFLAKE
Q2 this year: added 458 customers
Q2 last year added 397 customers
15% increase

CROWDSTRIKE
Q2 this year: added 1524 customers
Q2 last year added 830 customers
84% increase

DATADOG
Q2 this year: added 173 customers with ARR over $100k
Q2 last year added 55 customers with ARR over $100k
more than 200% increase

This is clearly not a good increase in customers for Snowflake, so the argument becomes, “Well, they’re targeting larger customers!” This may well be true, but it doesn’t mean these large customers will spend a ton with Snowflake. Hmm how can we know what they’ve committed to spend? RPO!

If these new large customers were committing to spend big dollars with Snowflake, we’d see big RPO increases. Instead we’ve seen low RPO increases – not just compared to the massive percentages we saw last year, but the raw dollar amounts.


F2020  138  221  273  426
F2021  468  688  928 1333
F2022 1432 1529

RPO increase each quarter
F2020   10   83   52  153
F2021   42  220  240  405
F2022   99   97

If these large customers were committing to big spends with Snowflake, why aren’t we seeing larger numbers this year? 97m this quarter was barely higher than Q2 two whole years ago! A company that is growing at 100% should see 4x as much!

I don’t understand why this seems ok to anybody. Again, RPO represents the dollars customers are committing to spend with Snowflake. These dollars aren’t growing as fast as they need to be. Looking into this more has made me more concerned than I was when I sold out yesterday.

Bear

122 Likes

Apologies, those were Q1 numbers for Crowdstrike’s customers. Q2 hasn’t come out yet!

Bear

The explanation is here; it consists of several tweets but the bottom line is that:

“However, this really isn’t a fundamental business issue. It’s really just a quark in how RPO is calculated, driven by a mix shift in multi-year deals due to sales rep compensation.”

https://twitter.com/jaminball

He is with Altimeter Capital which is the second largest SNOW shareholder. Ball better know what he is talking about.

He also has a weekly newsletter about SaaS companies.

As for the comparison to CRWD and DDOG, I don’t know if it is apples and apples considering how SNOW gets paid. I don’t know how long usage based models will last, but I don’t understand the concern given the constantly increasing amounts of data and the fact that we are barely getting started, actually (in terms of IoT, ML). SNOW does not need new modules etc to get more from existing customers. The latter will be using SNOW more just by virtue of how data increases. I guess at some point usage will be too expensive, like with text messages back in the day, so there is that.

12 Likes

I decided to calculate the RPO / revenue ratio since publicly available. An increasing ratio means that customers commit for a longer period of time (higher commitments compared to actual consumption) while decreasing ratio means that customers in general commit for shorter periods (lower commitments compared to actual consumption)

RPO/Revenue:
Q3 2019: 2.89
Q4 2019: 3.49
Q1 2020: 3.16
Q2 2020: 3.67
Q3 2020: 3.74
Q4 2020: 4.86
Q1 2021: 4.3
Q2 2021: 5.17
Q3 2021: 5.81
Q4 2021: 7
Q1 2022:6.26
Q2 2022: 5.62

Average of 4.66 versus 5.62 now implying that customers commit for longer periods compared to historical average. I am not concerned, however it’s a metric worth watching.

28 Likes

Hi,

I do long SNOW(10%), and also appreciate your calculations it’s really something. But still need to compare with one thing is recent RPO, yes they seem no provide, may be we can do new added RPO compare with usage this Q. That seems more like the situation now because the RPO is a accumulation numbers. That’s just my little advice.

I also confused with their qoq RPO growth rate, and believe that’s something not good with their land and expand. But I still that’s just short term elements. Every company happens. I probably will keep my position and wait their next ER see how new sales team improved their business.

Cheers

Rick

I might be out in left field, but…

If these new large customers were committing to spend big dollars with Snowflake, we’d see big RPO increases. Instead we’ve seen low RPO increases – not just compared to the massive percentages we saw last year, but the raw dollar amounts.

If these large customers were committing to big spends with Snowflake, why aren’t we seeing larger numbers this year? 97m this quarter was barely higher than Q2 two whole years ago! A company that is growing at 100% should see 4x as much!

Wondering about the R in RPO, Remaining Performance Obligation. If customers consume their “obligation” ahead of schedule it is no longer Remaining, thus reducing their contribution to RPO. I have no idea it this is a factor, but I thought the possibility worth mentioning.

8 Likes

Wondering about the R in RPO, Remaining Performance Obligation. If customers consume their “obligation” ahead of schedule it is no longer Remaining, thus reducing their contribution to RPO

This is an important point. In my mind, looking at RPO in isolation is not looking at the complete picture. We should look at RPO in conjunction with product revenues (since a significant part of revenues will be coming from RPO).

So what happened with product revenues ? It grew 19% QoQ ! (254.6 in Q2 vs 213.8 in Q1)

Also, think about this. New customers would want to sign a lower contracted usage, knowing they can buy additional usage anytime they wish. But because these customers are new, they don’t know how valuable Snowflake really is. But once customers start using the product, they know how good it is. How do we know that is true ? 169% NRR.

In fact, the CFO said as much on the call that they think NRR would be above 130% for a very long time to come. Think about how many companies can say that!

17 Likes

The explanation is here; it consists of several tweets but the bottom line is that:

“However, this really isn’t a fundamental business issue. It’s really just a quark in how RPO is calculated, driven by a mix shift in multi-year deals due to sales rep compensation.”

https://twitter.com/jaminball

He is with Altimeter Capital which is the second largest SNOW shareholder. Ball better know what he is talking about.

This is now the third time this Twitter thread has been posted as justification for the low RPO number (so could we please stop posting it?). That may indeed be the “bottom line” definition, but it is also accompanied by a $1.8B expectation using that same definition. The actual was $1.53B. While I’ve seen plenty of great breakdowns and explanations for interpreting RPO since, none seems to conclude this quarter’s result was actually a good one for SNOW.

Being honest, the more people we have rallying around defining or calculating RPO, the less we are focusing on the fact this wasn’t a great number in one of the two key metrics identified by management itself as key indicators for the health of the company.

We often take managements to task for attempting to move goalposts. While we should always be willing to test and adapt our theses, let’s make sure we don’t drift too far and end up moving the goalposts on ourselves.

68 Likes

To really understand how much was added to RPO, don’t we need to subtract the current product revenue from the previous quarter’s RPO?

Q4 21 RPO 1333
Q1 22 Rev 178.3

Q1 22 RPO 1432
RPO left from Q4 21 is 1154.7

RPO added in Q122 is 277.3

Q2 22 Rev 213.8
RPO left from Q1 is 1218.2
Q2 RPO 1529

RPO add in Q2 is 310.8

3 Likes

Building on Stocknovice’s point, I thought to look specifically at what management said in the past about RPO and what they said now, to see if their justification or my point about larger customer acquisition likely flies; I also calculated current RPO to see if a different picture emerged.

Is RPO important to management?

Q3 CFO:
“we do not focus on the same metrics that the SaaS business would. We focus on product revenue and remaining performance obligation.”

→ Yes.

What does RPO mean to them?

Q4 CEO:
“Remaining performance obligations of $1.3 billion grew 213% year-on-year, reflecting strengthened sales across the Board.”

→ It is a gauge of sales strength.

What is the impact of large customers on RPO?

Q3 CFO:
“As we’re moving more into large enterprises, large enterprises really want to do multi-year deals. They are not interested in having to go to procurement every year and so we saw a number of large enterprises commit to multi-year deals with us and we think that is going to be a trend that will continue.”

→ Large customers tend to do big multiple year deals.

What drove the high growth in prior Q’s RPO?

Q4: “Our strong RPO results continue to be driven by more multi-million dollar deals, as well as our customers’ willingness to engage in multiyear contracts.”
Q1: “Our strong RPO results reflect more multimillion-dollar relationships with particular strength in the telecom and technology sectors.”

→ More big deals and more multiple year deals drove last Q’s increases.

So what happened now?

Q2: “As a reminder, in Q2 last year, we sold our largest multi-year contract ever, a three-year $100 million deal. While the multi-year component of new booking sets up a difficult comparison, we saw a net - we saw new annualized contract value accelerate compared to the year ago period.”

→ They are saying it is because of a tough compare in Q2 last year, but that doesn’t really wash if you compare qoq…

And what does current RPO look like?

Last three quarters current RPO:
$733m → $773m (5.5% qoq) → $856m (10.7% qoq)

→ Not shooting the lights out either

I’m starting to move over to Bear’s camp here…

46 Likes

SNOWFLAKE
Q2 this year: added 458 customers
Q2 last year added 397 customers
15% increase

CROWDSTRIKE
Q2 this year: added 1524 customers
Q2 last year added 830 customers
84% increase

DATADOG
Q2 this year: added 173 customers with ARR over $100k
Q2 last year added 55 customers with ARR over $100k
more than 200% increase

This is clearly not a good increase in customers for Snowflake, so the argument becomes, “Well, they’re targeting larger customers!” This may well be true, but it doesn’t mean these large customers will spend a ton with Snowflake. Hmm how can we know what they’ve committed to spend? RPO!

If these new large customers were committing to spend big dollars with Snowflake, we’d see big RPO increases. Instead we’ve seen low RPO increases – not just compared to the massive percentages we saw last year, but the raw dollar amounts… [Saul here: the dollar amounts this quarter were less than half of what they were a year ago, $97 million compared to $220 million, 44% in fact).

…If these large customers were committing to big spends with Snowflake, why aren’t we seeing larger numbers this year? 97m this quarter was barely higher than Q2 two whole years ago! A company that is growing at 100% should see 4x as much!

I don’t understand why this seems ok to anybody. Again, RPO represents the dollars customers are committing to spend with Snowflake. These dollars aren’t growing as fast as they need to be. Looking into this more has made me more concerned than I was when I sold out yesterday.

I’m with Bear on this. I sold out too. To be clear, I’m not accusing Snowflake of being a Fastly. I’m sure it will continue to grow. But I have better places for my money.

Saul

107 Likes

If these new large customers were committing to spend big dollars with Snowflake, we’d see big RPO increases. Instead we’ve seen low RPO increases – not just compared to the massive percentages we saw last year, but the raw dollar amounts.

As management as consistently stated, new customers typically sign up for 1-year contracts, but then at renewal time sign up for multi-year contracts. Therefore, one might expect that it’ll take a year from initial sign-up to see RPO increase by an increasing amount. And the emphasis on larger customers may mean that such renewals will be much larger than have been with smaller customers.

If these large customers were committing to big spends with Snowflake, why aren’t we seeing larger numbers this year?

A large customer won’t necessarily start out with a larger try-out, but when successful, word will spread within the organization and the renewals will be larger.

Again, RPO represents the dollars customers are committing to spend with Snowflake.

One of the nice things about SNOW’s business model is that it’s not SaaS, which contracts both sides. With a SNOW contract the customer is committed to a minimum spend, but since the charges are usage-based, customers can easily spend over the contracted amount. These contracts motivate the customer to follow through on the migration or development (for old or new applications, respectively), at which point Snowflake is confident that they’ll see the benefits of their product, as evidenced by their continued high NRR.

22 Likes

I’m with Bear on this. I sold out too.


I applaud your consistent ability to unemotionally walk away from stocks when the investment reasoning you have for them departs from “the story”.

Too many of us stick around for the story a bit too long, myself included.

CRWD doesn’t report until Tuesday, but care to comment on where your highest convictions in your port are now, with SNOW out of the picture?

Dreamer

9 Likes

This is my first time posting to Saul’s board so I hope this is at least somewhat constructive.

Firstly, I would simply like to thank Saul and other long time members of this board for their collective wisdom. I have been visiting this board for most of this year and have been trying to learn as much as can.

I see a previous reply in this thread eluded to the need to take previous RPO in conjunction with current product revenue. So in looking at Q1 RPO (1432 Mil), I would subtract current Qtr Rev (254.6 Mil). This would leave approx 1177 Mil of RPO from Q1.

However, Current Q2 RPO is 1560 Mil meaning there is less than 400 Mil worth of “New” RPO brought in during Q2.

Performing some very quick math to walk this exercise back through the previous year, I get +328 Mil in “New” RPO in Q1, and approx 600 and 400 in both Q4 FY21 and Q3 FY21.

I would expect to see these numbers growing significantly, even dramatically, based on my previous assumption of SNOW’s story. Instead, there appears to be a level of stagnation over the last year.

I’m leaning more towards Bear and Saul’s view and am therefore left considering reducing further or eliminating my position for now.

If I am off base, or beating a dead horse with this thread, I welcome constructive feedback.

Thank you again to Saul and to other long time contributors of this board.

14 Likes

I don’t know that the RPO debate has moved all that further along except for the great tidbits posted earlier–those taken from ECs.

To me the summary looks like this:

  1. RPO–TBD, nobody seems to know for sure either way;
  2. The competitive landscape, new to me (as a non-techie) concern since a couple of threads were posted here last week–and then I went and read other materials on SNOW vs Databrics re: ML;
  3. The market cap plus valuation, old concern of mine since SNOW’s case is unprecedented
  4. Saul’s and Bear’s intuition (sorry LifeOfDreamer, there are no clear cut hard facts to go by here, since the RPO numbers’ interpretation remains TBD).

Well, there may be no clear cut facts (ain’t FSLY or AYX) but that’s a whole lotta TBDs!

I am new (1+ year) but it does seem to me that jumping ship when doubt arises is a key takeaway from Saul and others with long term sustained success.

I will see what my options are, thanks for all the input everybody and have a great weekend.

6 Likes

Here is a different take in understanding RPO pertaining to Snowflake. A while ago (more than 5 years ago) I was consulting at a large bank as an independent consultant. One of my recommendations to them was to move a lot of their workloads to the cloud, where compute, network, storage and any other resources are available on demand and it would allow them to handle spikes in workloads a lot easier, also reduce the upfront costs (time and large expenditures) of securing infrastructure ahead of new projects

One of the things about moving to a cloud provider that they did not like was the OPEX cost model over CAPEX. It seems CAPEX allows for some beneficial accounting.
Most customers using AWS and Google Cloud do Opex, ie: Create an account, give out their billing information and start using infrastructure resources. Also once your business starts using the cloud, you don’t go back, your consumption only increases, sometimes at ridiculously large rates and AWS bills you based on your usage of resources. I don’t know the details but AWS also offers all forms of discounts to customers signing longer term contracts. This they post as revenue backlog.

I see Snowflake having a similar model, where people migrate their data to Snowflake and pay based on usage. They might not have to sign long term deals, they just use it by the hour and Snowflake bills them. However it seems to me that Snowflake wants more visibility into customer use of their resources, so they incentivize their salespeople to get customers to sign long term contracts.
I speculate that there is a benefit for Snowflake if a customer agrees to a longer term contract as it might force the customer to using Snowflake during that duration. My take is that once they start using Snowflake there is no going back, just as it is with AWS etc. At this point, there might be no benefit for Snowflake to sign an extension and offer more discounts (Or there might be an advantage, I don’t know). From this thread I can see that there definitely is an advantage as it pertains to attracting investors.

So in short,
Once a customer starts using Snowflake and their many wonderful features such as data sharing, they are not going anywhere else. So this RPO stuff is not very meaningful to me.
New customers are more meaningful, It is not clear to me if they are reporting just larger new customers or total new customers.
Retention rate is going to be high, the two choices are you either start using Snowflake within the contractual period and use if forever after that or don’t ever use it.
Snowflake is going to be around and will be much larger in a few years, how fast it gets there is the current debate that I see here.
Snowflake has no real competition, they are miles ahead on the breadth of features such as multi cloud support and the aforementioned data sharing. From a recent discussion I can see that they are getting into more Data Lake and Data Science use cases as well increasing their addressable market. True for a subset of cases somebody could choose Google Big Query. But if you are a larger organization, Snowflake will handle a lot of your needs best.

I for one will keep my current stake intact or trim if the stock starts failing to hold up, but I will always keep adding on dips.

14 Likes

I’ve heard Jason Moser state multiple times on podcasts that you can make investing as hard or as easy as you want to make it.

There have now been multiple threads on this board with no clear consensus on what to make of SNOW’s RPO. One of the most advantageous aspects of SaaS companies is visibility into future earnings. I’m not sure if usage based business models will have the same visibility.

For me, there are too many other companies that I have more conviction in that I sold my small starter position in SNOW

8 Likes

There have now been multiple threads on this board with no clear consensus on what to make of SNOW’s RPO.

Hi Billski,
It’s never a clear consensus. There are always people selling at the same time you are buying, and vice versa, or we wouldn’t have a market. But even with something as clear as Fastly there were people, at the time, who were hanging on, and some who were really angry at us for getting out and maligning their darling.

For me, there are too many other companies that I have more conviction in that I sold my… position in SNOW.

That’s exactly how I saw it.

Saul

38 Likes

There are a few pieces of information that I think are being missed or that we simply don’t have.

  1. What is SNOW’s historical customer adds? I can’t find any data on this before they went public. With their high NRR, they don’t need as many customer adds to continue growing as an uber-hypergrowth mode.
  2. 1M customers is a TRAILING figure. That’s not the current run rate. That’s like using P/S TTM to evaluate a company instead of P/S NTM. It doesn’t really tell us much. Combine that with the NRR of 169.

  3. RPO concerns. I really thing this piece is not very as important, maybe I’m wrong… Smorg is dead-on right that the money paid up-front will get eaten thru quicker then companies expect. From there, companies are paying for usage as they go. This likely is playing a part in the RPO number not growing as some prefer. Its not like the OKTA RPO where it is pretty close to known how much of that RPO will be turned into income on a steady basis.

Data is like a drug, once companies see the benefits of it and how easy it is, they can’t get enough. In addition, Snowflake makes it SO SO SO easy to push the easy button to get that data faster (and pay more for that). Its so easy to say we’re going to pay double for this query and go back and fix it later… later is like tomorrow, it never comes, its always later. The next big issue comes up or next big project and that query is running fine, providing end users what they want. Then if it still isn’t scaling and slows down, we can click that button again later to ‘temporarily’ fix that problem to the users and of course, promise we’ll come back and fix that later… It happens all the time.

8 Likes

Hmmm… a challenging analysis with limited information for sure. FWIW, I may very well end up being in error over the long run, but can’t escape the following logic I do know right now: Didn’t SNOW just throw up 104% revenue growth (bigger than ANY other company we own short of UPST) and more or less tell us next quarter would be triple digit rev growth again?! I can’t imagine a much more upbeat, optimistic management team as that call.

I personally have come to the conclusion that I have to take RPO with a grain of salt in lieu of the incredible revenue growth and DBNRR actually given and their clear guidance…RPO is just a minimum legal obligation (and not a small one at $1.5B) but I seem to also thought I recall them saying $1.2B of it would occur in the next 12 months!! And another point 56% of it, which would be something less…but either of these results tell me several things…there is a LOT of RPO obligated to be recognized in the next 12 months, but which we know has to be a minimum, as none of these customers are limited to what they can spend, their spend it increasing quickly, and this number is not even close to big enough to hit the undoubted 100% growth again next quarter…which tells us it’s not comprehensive of all the revenue they will get from existing customers, especially if you assume it’s not coming from new ones (as Bear adeptly points out is diminishing in growth. Second (which they also told us) they have a large number of Fortune 500 and other very large companies who may not have ANY obligations reported in that RPO number yet and they’ve told us will not even start generating revenue for 6-9 months from Q4 (so Q3?). And not to overlook their stated Intl revenue growth or the record number of sales people brought on in Q1 that dwarfed the entire prior year and will likely just start gaining traction next quarter! I feel I have to rely on their stated rev guidance and DBNRR and take RPO (harder to define and not the same ingredients as many of our other SaaS companies) with a grain of salt.

SNOW is simply not the stereotypical SaaS company and my gut tells me their existing customers could provide 2x times what their contracted RPO number may indicate…or far more. If they don’t, I can’t fathom how they will maintain the 100%+ revenue growth they just told us they will hit next quarter…in which case I will run, not walk, away. Naturally, I too certainly wish we had more insight, certainty and a trend of some sort to follow around RPO in the form of consistent contract length, utilization, a break out or current vs total RPO, etc, but without it, I have to rely on the mgmt team and the two absolutely critical and clear numbers they are giving us and trust that they will have the RPO (or bookings, billing, future business, increasing usage, or whatever additional and/or recurring revenues likely to materialize from every customer that is part of that 169% DBNRR in order to hit that triple digit growth they are telegraphing. If I didn’t have that faith in this mgmt team or this company (who have yet to mislead us, IMO), I would certainly not choose to be invested.

Would I prefer to be buying more at $200/share, as I did when I loaded up on $250 shares and call leaps? Of course. But I am very happy after a 10% surge last week to continue to hold at my current relatively small 5.4% share position until I discover a more compelling reason to add it or dump it.

Thank you for the great discourse. It certainly makes me think harder about SNOW, but I also find it easier to hold at a smaller level in the supersonic growth, since it’s not a massive 10% or 15% position for me. I love the company direction and the path they are forging… it is only my 7th sized position out of 9 total companies I own, but is my/our second fastest grower by a long ways (only now behind UPSTs plaid speed hyper jump quarter)…and may be the fastest company at scale!!

SNOW and Slootman’s team have bought another quarter of allegiance from me. Cheers! -Pokeeko

35 Likes