GTLB – “Software is eating the world”

Good afternoon, everyone -

Tech VC fund owner Marc Andreessen is famous for his quote “Software is eating the world.” If you believe him, then Gitlab (GTLB) may be an investment that interests you.

GTLB has faced downward pressure since its last earnings call, where it guided for 26% growth over the next year (FY25). While this type of growth guidance may cause some of you to yawn, it is important to remember GTLB guided for 25.5% growth in FY24 and ended the year with 37% y/o/y growth!

On April 1, 2024, Bert Hochfeld of Ticker Target wrote a great article on SA about GTLB:

https://seekingalpha.com/article/4681411-gitlab-not-cheap-but-worth-a-premium-valuation

Bert recommend the purchase of GitLab shares in the article. Here is a summary of some of the points from Bert’s article:

  • Growing Industry: GitLab is a “DevSecOps” tool, which is a fancy tech way of saying it helps people develop software. I think software is an industry that is only going to grow as AI takes off.
  • Unique Product: Bert describes GitLab as having a unique software development tool, because of its “security and compliance” features.
  • AI Play: GitLab has just started offering a new product, Duo, which “integrates AI throughout the software development life cycle.” Bert describes: “If Code Suggestions and chat, another key part of Duo offering are successful in real-world environments, than the revenue guesses I have made for this product will prove to be ultra-conservative, and in turn that will mean that growth estimates for GitLab will ultimately be positively revised.”
  • Immediate ROI for Investors: “GitLab suggests that the use of its platform has an ROI of 427% for revenue generating applications.”
  • Consolidation Platform Play: “Gartner believes that GitLab has competitive advantages in term of native security, its open platform approach, and it emphasizes the company’s single platform.”
  • Founder-led and Solid CEO: “But really the company CEO, Sid Sijbrandij is probably the best sales person on the GitLab team. Sid is one of the pioneers in terms of creating a DevOps platform, and then in adding features to make it a DevSecOps platform.”
  • Strong Recent Quarter (absent guidance): “Last quarter was a record bookings quarter, apparently, well above expectations with the largest 1st order, the largest order in the Ultimate tier, the most million dollar deals ever and good linearity in the quarter. Even hyper-scaler demand was at record levels. The cRPO balance-that is backlog to be recognized in the current year- rose by 40% year on year.”

Guidance is the sticking point I see for this stock. Another bear argument is that AI will eliminate the need for software development. In a recent fireside chat, the CEO explained that the opposite is true: https://ir.gitlab.com/news-releases/news-release-details/gitlab-hold-fireside-chat-truist-securities. He made the case that AI means more code, which still needs GTLB to test it, secure it, etc … (whether or not it is written by a computer or a person).

Google has endorsed GTLB buy owning a sizable piece of it. Earlier today, Google gave GTLB the “2024 Google Cloud Technology Partner of the Year Award in the Application Development - DevOps category.” https://ir.gitlab.com/news-releases/news-release-details/gitlab-wins-google-cloud-technology-partner-year-award-devops.

This morning, a Wellsfargo analyst agreed with Bert’s take: “With shares -21% off since 4Q eps (vs. NASDAQ +0%), the FY25 model now more conservatively set, and an AI-led product cycle taking shape, we see an opportunistic entry for GTLB shares. Initiate GTLB shares ‘OW’, $70 PT. We view GTLB as a consolidator in the expansive (+ fragmented) DevSecOps landscape, offering tools spanning the entire software toolchain (vs. most peers w/ point solutions). Further, we see the debate around AI reducing number of devs in the market as largely overstated; expect AI shift grows the number of addressable users over time given proliferation of software/applications, AI’s expansion to citizen devs, GTLB’s product expansion outside core devs. GTLB shares are trading at ~12x EV/S on our NTM ests, still above software averages despite the pullback. Our $70 PT is based on 12.5x EV/S on our Fwd NTM ests, above peers but justified in our view given GitLab’s positioning as a consolidator in the DevSecOps landscape and a likely beneficiary of the significant gen AI-led product cycle forming.”

It seems to be a company that has a lot going for it.

Thanks for reading.

Best,

BTL

@laneylawyer on X (the artist formerly known as twitter) https://twitter.com/Laneylawyer

Long GTLB

40 Likes

Thanks for the write up. I’ve been meaning to look into GTLB but I just haven’t had the time. I appreciate your summary.

I spent a large part of my career engaged in application development. Though I retired about 14 years ago, there’s certain aspects of going from an idea to a production application that I’m confident are still part of the life cycle.

I have only a high level understanding of the primary functions of GTLB, but the way I understand the product, it serves the vital role of Change Management and Configuration Control (CMCC). It provides a secure environment for the managing the various states of the same software as it moves through the life cycle.

While this function is vitally important, I don’t yet understand what makes GTLB unique. In that it’s virtually impossible to develop and maintain commercial software without CMCC, I must assume that there’s a lot of competitive products. You noted that Bert claims GTLB is unique because of its “security and compliance” features. I find it hard to believe that GTLB is alone in this respect. It seems to me that this would be a near universal feature set for CMCC software.

You briefly touched on additional features beyond CMCC. It’s on me to dive deeper into this company and their products.

8 Likes

Github is another. A secure connection to a server at github, lots of version control functionality based on the GIT version control system. You can clone and merge between branches. Features automated regressions of new/changed models after merges or updates. Has the history functionality of GIT. I use it from time to time. I do not have an opinion on its “better or worse” comparison to gitlab because I have never used gitlab. If you are a modern developer you probably know people who have used both. I doubt github is publicly traded. But I know at least one major tech corp who uses it.

2 Likes

Ugh, why is it always bad when non-techies try to evaluate software companies? Bert is way out of his lane here.

Not quite. From this:

DevSecOps stands for DEVelopment, SECurity, and OPerationS. It’s an approach to culture, automation, and platform design that integrates security as a shared responsibility throughout the entire IT lifecycle.

That said, GitLab is more than DevSecOps, and software development needs are much larger than just security related. When Bert says:

Many of the company’s largest wins this quarter and in recent quarters have been based on the unique security components of the GitLab DevOps platform.

I get scared that GitLab is pigeon-holing itself. And note that GitLab isn’t “unique” in offering this.

Big IF there. Here’s a review of GitLab’s AI offerings, by techies for techies: GitLab Code Suggestions: Why You Should Not Build Your Own Generative AI Product for Code

we love this as a baseline test - write a Python function that topologically sorts the nodes in an input directed graph…OK, something is completely busted. We are getting comments in Korean, not even English, and even bad unicode characters. When we try to soldier on with the for loop, no suggestions are produced at all.
Let us back it up to something even simpler… this example just highlights just how poor GitLab’s model-surrounding logic is. First off [cut out techy stuff]…

Summary:

GitLab’s solution today is many steps behind what developers now expect from such a tool in every respect - IDE and programming language availability, context collection, suggestion quality, integration of suggestions into existing code, and baseline model capabilities** . Even worse, it is artificially restricted to work only on GitLab repositories, creating a massive burden of adoption for the product.

So, yeah, far from being “ultra-conservative,” Bert’s analysis is way off base and overly optimistic.

I’m going to stop here. Bert’s analysis is ultra-flawed and the company’s AI offering apparently falls way behind Microsoft’s CoPilot for one, probably many others.

37 Likes

Smorgasboard,
Thank you for weighing in here…
I see your reply regarding Gitlab as a case study in why it is that my portfolio is sooo concentrated.

I simply lack the skill set to see where the growth in software design is going. As AI writes more and more code, I believe we should all get a little more humble about this. Whose going to be able to roadmap where their company needs to move to sustain market share.

When will the way forward be more predictable?

I’m not even sure Pure Storage has much of a moat and am thinking of getting out. I really respect your opinion in this and should appreciate any advice you could offer.

Best

Jason

11 Likes

Microsoft Devops is a direct competitor no? I bring this up because similar examples exist where a relatively inferior product continues to maintain market share regardless of new features.

Zoom - Teams
Slack - Teams
GTLB - MSdevops

Zoom calls would take over the world. Until Teams and Google Meet established that inferior products, which are free to a subscription base will be chose a large percent of the time (good enough)

Slack (now part of salesforce - CRM) was a new way for developers and users to work in channels. Teams was a warmed over version of sharepoint from Microsoft, “NOW WITH CHAT!”. I believe it’s still inferior to the slack offering, but is what almost everybody is using.

Similarly, jira from Atlassian (TEAM) and devops from MSFT perform in this space.

I don’t see the structural moat that is so revolutionary to GTLB that DEMANDS that users switch - and maintain subscriptions.

8 Likes

Thanks everyone for the input. I did want to clarify that this was not a quote from Bert’s article. It was my layman, non-techie summary of what this company does. Bert’s article has much more detail on what is a DevSecOps company and what Gitlab does.

1 Like

As with anything AI, the best performing product will depend on having high quality data on which to train. But, I don’t know how to quantify that, which companies have agreements with their customers that enable them to analyze the code being stored, nor which companies have the AI expertise and server access to construct training models and implement inference.

Even outside of AI, I’ve grown to avoid investing in software development companies as a rule. It’s an industry where true features & performance are crucial, but where customers are actually skilled enough to write their own should they have the time/money/inclination, and normally have the ability to pretty easily move their project from one offering to another, not to mention new projects are starting up all the time and can normally be started up on anything so there’s little loyalty.

9 Likes

My best understanding is $GTLB is more relevant as these things increase:

  1. Number of Developers
  2. Disparate/different groups of Developers across multiple geographies, chains-of-command etc.
  3. Number of requirements related to Rapid-Application-Development (“RAD”), Continuous Integration / Continuous Delivery (“CICD”).
  • (NOTE: CICD requires implementation of, and orchestration with multiple adjacent technologies, each with their own scheduling, revision control etc. )
  1. Number of integrations with processes that are adjacent to software development; and again there are probably lots of stakeholders that vary widely in terms of priorities, chain-of-command etc. So: things non-Developer stakeholders are interested in, like Performance, Security, Compliance, Documentation, Analytics etc.

If it’s just two guys working weekends on a side-project, that need a repository, github is fine.

However as the number of requirements related to 1…4), above increase, the more you need a comprehensive framework that “just works” and the more worthwhile it is to pay for it.

imo this in theory means the target audience for $GTLB is limited to large organizations who have complicated requirements with lots of integration necessary across multiple organizations, teams and technologies.

6 Likes

We (cross functional team of 16 people in a company of 100 people) decided to go with gitlab compared to github mainly due to pricing. This was 2021.

For a team that needs agile boards, git integration, cicd and other basic stuff, gitlab was totally sufficient.

I believe most people of the team would have decided to go with github if it wasnt for pricing. The reason is that most devs already have a github account for various reasons ( open source libs are hosted there, its free for private use and the most common option). But we asked the team and all agreed that its no big deal using gitlab instead. Back then they were so similar that switching wouldnt cost much of our time. Software people often can switch tools and libs easily and quickly because they “think software”.

Though switching current projects to other SCM (source code management systems, though technically speaking, gitlab and github are more than a simple SCM as the underlyinh git is the actual SCM) is a thing that needs to be avoided, as all history of that project are stored in the one system. Software projects exist for a long time. And for bug fixing future bugs, you want to rely on the history of the development of that project.

And using another SCM for new projects makes it tedious in one company, as you have to constantly switch between systems.

With micro services every team couls use its own tech stack. But as this approach matured, more and more companies realised that you had to maintain so many different tools and pay so many different vendors and people cant just simply switch from one team to another. Therefore companies started initiatives to consolidate the core tech stack for more common micro services.

My summary would be that I think gitlab is suited for not only large and complex enterprises but also smaller teams due to similarities to github and due to pricing advantages (at least back in 2021).

Also, there is some strong stickiness with SCMs, as long as there are no easiy migration processes integrated in one or the other SCM.

I hesitated to invest in gitlab myself, as I didnt see any strong moat and any strong differentation to other SCMs like github.

(This is my first post on this board. I hope I could add a little to this topic from my own experience)

31 Likes

Great add Rndsheep.

A few more data points since my original post.

Best,
BTL
@laneylawyer

Long gtlb

4 Likes

Jamin Ball’s top performing cloud-based tickers are not limited to just 3 from this tread but include most all of these tickers held by Saul’s Community. Here is today’s list of list of tickers with the Top 10 EV / NTM Revenue Multiples.

Ball’s free Clouded Judgement offers a significant contribution to any analytical assessment of cloud investment opportunities, and I encourage members to take a look, if they have not done so already.

Graydrake


Long - Samsara, GTLB, Snowflake,

12 Likes

I owned GTLB quite a while back. I last wrote about the stock in October 2022 and exited in November 2022.

I decided to take a position again today.

Latest results: https://ir.gitlab.com/financial-information/quarterly-results

Revenue

Yoy revenue was up 33% in Q4 and 36.7% for the year:

Revenue Q1 Q2 Q3 Q4
2021 42.2 46.2
2022 49.9 58.1 66.8 77.8
2023 87.4 101.0 113 122.9
2024 126.9 139.6 149.7 163.8

The absolute $ increase in revenue:

Incr Rev Q1 Q2 Q3 Q4
2021 4
2022 3.7 8.2 8.7 11.0
2023 9.6 13.6 12.0 9.9
2024 4.0 12.7 10.1 14.1

→ That is a new record for new quarterly revenue. Compare that with Monday, which is at a higher revenue scale but can’t seem to accelerate that number.

The key other reasons for me re-entering this name are:

  1. NRR: they calculate NRR on ARR, which makes it a forward-looking metric. This has now shown two q’s of improvement and ended on a very healthy 130%. Last 5 q’s: 133% → 128% → 124% ->128% ->130%

  2. Margins: GM% up to a record 92%. Second q of op margin positive: from 3% last q to 8% in Q4, up 19%pts yoy. There is definitely leverage kicking in:

Op Margin% Q1 Q2 Q3 Q4
2021 -103% -78% -53% -48%
2022 -45% -42% -36% -35%
2023 -28% -27% -19% -11%
2024 -12% -3% 3% 8%
  1. Large customers. They have 50% of the F100 as customers, and their >$100k ARR customers grew sequentially by 7%, 8% and 9% the last 3 q’s, for 37% growth in all. They now have 96 >$1m ARR customers, up 52% yoy.

  2. RPO & cRPO: up 55% and 39% respectively, both pointing to acceleration, or at least bigger commitments from customers. This is a similar trend to what we saw with Snowflake’s results.

  3. Competitive position - it’s basically them, MS and Atlassian, and they seem to be taking share from both as shown in these two graphs from Gartner and Forrester.

  1. Guidance of 26% for the coming year is also now the consensus analyst view for next year. I don’t buy it…the NRR alone, if maintained, will result in a ±30% revenue growth, all else being equal. All the analysts also questioned this, but still went with the guide for their forecasts.

  2. Product development. They have greatly improved their product in the last years, and have smartly incorporated AI in just about everything. I thought this was a great slide from their most recent ER presentation:

→ All in all this seems like a company that has a lot of momentum in the actual numbers, vs a slowdown in the guide. I took a position as I think the numbers tell the real story.

Any other takes??

-wsm

22 Likes

@wsm007 - a couple of points to factor into your analysis:

Keep in mind that part of that surge was due to the price increase they introduced for their Premium Tier. Management was asked twice to break-down how much the price increase contributed and they refrained from answering. They did mention that they anticipate the contribution from this price increase to be in the $10M-$20M range for this FY. Not saying its not a valid lever to pull, but worth noting that its not all just a ramp-up in customer momentum.

Without the price increase, I doubt NRR would’ve ticked upwards, as an entire tier of customers had to either churn or accept the higher price. Also makes me wonder the about the proportion of customer adds and RPO that were attributed to this.

Secondly, though Q4 was strong, guide for Q1 points to ~$7M sequential increase. Stronger than Q1’23 but not nearly close to 2H’23.

I don’t think anyone buys it, but I don’t think its flagrantly off. For what it’s worth, I’m penciling in ~33%. Like Samsara, management lowered the expectations regarding expected beats this year too.

The lack of leverage was the biggest story in their latest quarter. The market was expecting a ramp-up in FCF and OpMargin as it had been implied throughout the year; but it seems that management decided to delay profitability by a year. I don’t doubt that they’ll exceed their guide, but this is where I’d like to see some progress.

Headcount actually decreased YoY, and there does seem to be incremental improvements in S&M/Revenue (56% to 45% in a year).

Almost any developer I talk to is using Github copilot, so it will be interesting to see the momentum that Gitlab duo starts picking up…

Tempering my expectations because I’ve been burned by this one before. That said, I definitely think the setup is much better now (at ~10x EV/ntmS with reset expectations) than before their Q4 (at ~15x EV/ntmS with expected profitability).

-RMTZP

17 Likes

Thanks very much for the response.

Your push-back got me to look a bit deeper into the progress of that NRR of theirs, and there is indeed reason to curb my enthusiasm.

This is what the CFO said about NRR in the last 4 quarters:

Q1 2024:

The metric: NRR 128%, down from 134% in Q4. His comment about Q1’s NRR contribution vs Q4 contribution:

“And so seats is about 50%. Price increase is about 25%, and [tier change] is 25%. So there really hasn’t been any change whatsoever.”

→ My calc is that without price increase, NRR would have been 121% in Q1

Q2 2024:

The metric: NRR 124%, down from 128% in Q1.

The CFO didn’t break up the split of NRR like in Q1, but he did comment that customers are just buying what they need.

→ Assuming price contribution was 33% like it was in Q3 (see next section), my calc is that without price increase, NRR would have been 115.8% in Q2

Q3 2024:

The metric: NRR 128%, up from 124% in Q2. NB! The NRR pickup was due ENTIRELY to change in calculation method:

“we’ve made a change in the way that we’ve calculated dollar-based net retention rate this quarter to better reflect the business itself. And so, this quarter, if you did the old way, be flat 124% to 124%. But the new way basically takes into the account, the account hierarchies. The old way that we did it actually kept a static view of the parent accounts. And if there was a merger, a new subsidiary, division being shut down or something, that actually showed a churn and then a new business. And so the change that we made, it’s important to note, doesn’t change the business at all. It just shifts between growth and new to give a better reflection of what’s going on in the business. This is reflected in our 10-Q for Q1 and Q2. And so you can see the disclosures there. But under the old method was 124% to 124%. And under the new method, it’s 128%.”

And the components of NRR:

“Historically, seat change has been roughly about 50%. Price change and tier change have been 25%. Because of the price increase, the price change now makes up roughly about a third of the total of the net dollar retention rate. The seat change is a little less than 50%. And then the tier change is a little less than 25%.”

→ Ok, this I missed and this is quite important.. They changed the way they calculate NRR in Q3 and that was the only reason for the uptick in NRR this Q. Sure, the new way is better, but apples to apples, NRR stayed flat between Q2 and Q3, and without the bigger contribution of the price increase, it would have declined.

So on the old calc, the NRR would have been flat at 124% in Q3, and without the price increase, calc is that without price increase, NRR would again have been 115.8% using the “old” way*

Q4 2024:

The metric: NRR 130%, up from 128% in Q3:

“happy to go through what is the buildup in the dollar based net retention rate. And so super happy that the dollar based net retention rate went up this quarter to 130%. We break down between seats, price and tier. So 40% of that was related to seats, 40% was related to price and 20% was related to tier, not too dissimilar from last quarter. It’s just a little bit different last quarter. Churn and contraction also continues to improve. As I stated earlier, it’s even better than rates that we saw six quarters ago.”

→ So, the price increase contributed 40% of the expansion in Q4 vs 25% of the expansion in Q1. Quite a difference, and material in both q’s. And assuming that NRR ticked up by 2%pts using the “old” way too, the NRR would have been 126%. And without the price increase it would have been 115.6%.

So apples to apples using the “old” way, NRR did roughly this in 2024:

128% → 124% → 124% → 126%

And assuming price contribution was 25% in Q1, 33% in Q2 and Q3 and 40% in Q4, the NRR ex the price increase, NRR was:

121% → 115.8% → 115.8% → 115.6%

Essentially flat vs the prior Qs!! @rmtzp is right…what will NRR do once the price increase comes out of the wash? Much lower.

I’m now quite a bit less enthusiastic on this stock.

-wsm

20 Likes

Yes the price incr will not repeat often but its EFFECTS are not one-off. CFO said FY25 will be $10-20M boost from it and hinted FY26 will be more.

I appreciate the apples to apples view on NRR definition change, but think you are taking the breakdown too far trying to back price increase out. It is going to ripple through for 2 more years for renewals plus all new customers.

Know what else is not factored into NRR? The coming uplift from AI add-ons… + $19 for dev code assistant and chat, and soon a fuller $39 Enterprise tier for Ultimate to access a complete AI suite (15 capabilities listed… maybe lots more in a year from now)

We’ll have to see what the uptick ultimately is … but given the success of GitHub Copilot…. Developers are clearly embracing AI tools. If GitLab can get their code assistant anywhere near par I think they take a good share. (That for now is up to the capabilities of Google Codey vs OpenAI GPT-4. And Gitlab has no qualms about switching to better engines … as long as it is not Microsoft - which unf rules out OpenAI.)

I am excited about the extended AI features yet to come and the bigger price tag of Duo Enterprise. They are really pushing to take away a lot of the drudgery of non-coding tasks (merging requests, creating or summarizing issues, test automation, automated vuln remediation).

-muji

20 Likes

@wsm007 Willem, I wanted to circle back after you made me dig a lot deeper. I greatly appreciate your consistent attention to detail (here and every post!) and, let’s be clear, you know this company and its nuances better than I. However, I think your conclusions above are incorrect, and beyond that, I’m really not sure how you are confident enough to calc your own NRR based on the add’l breakouts given by CFO (and in my own attempts, I come up with way, way better NRR).

The price increase

First off, let’s start with the specifics of the Premium price increase. As I said before, this isn’t a one-off, but was a staggered increase at time of renewal. From April 2023 on, the first year renewal of Premium tier was offered at a discount of $24. Next year (after April 2024) gets greater uplift at full $29 price. So for a price increase of $19 to $29, the first year (across FY24) was half of the increase (+$5), and second year (across FY25) was the other half (+$5).

From Q124 on, the CFO has repeatedly said the price increase would have less impact on FY24, stronger impact in FY25 (add’l +$10-20M), and into FY26. Yes, this will continue into FY26 a bit, as the price increase doesn’t match their FY. Q124 had 1 month of it, and Q126 will have 2 months of final renewal at full price.

The NRR breakouts

Beyond that, the CFO is providing breakouts are part of his discussion of NRR. This means he is generalizing the contribution to overall expansion revenue TTM in these breakouts - aka the part of NRR over the base 100% from the ongoing subscription renewal. (I assume no seat changes, but obviously that is a factor here too – however they say that contraction has stabilized.)

I think you missed how 25% of this expansion revenue TTM (part of the NRR discussion) was already on “price yield” before Q124 first saw the effects of the price increase.

CFO in Q423 (Mar-23): “So first let me talk about the net dollar retention rate. Seat growth is still the number one contributor to the net dollar retention rate, followed by tier upgrades and then price yield and so that hasn’t changed that much. Seat growth, two or three quarters ago was about two-thirds. That’s down to around 50% now and then the other to make up the remaining 50%.”

Okay, so going into the price change, we had expansion coming from 50% seat increase (from 66% a few Q prior), and 50% between tier upgrades and price yield. [My guess as to why it decreased? Success of Premium->Ultimate tier upgrades bumped up in the mix.]

What is “price yield”? He actually clarified this nicely in a recent Truist fireside chat after Q424 (will get back to those breakdouts in a sec).

CFO at Truist fireside chat after Q424: “As we go through the dollar-based net retention rate, we historically have given sort of what the breakdown is between seats, increased price yield and tier upgrades. And so about 40% of it is related to more seats this 12 months over the prior 12 months. And the other 40% is increased price yield. It’s important. We’ve talked about this since being public. And so when we sell the product, we may give a certain discount for year 1. And so then in year 2, we give less of a discount. Year 3, we give less of a discount. And this is because we put so much feature functionality continually in the platform. So it’s comprised of increased yield per customer as well as the premium price increase, and that was roughly about 40%.”

So price yield has traditionally been about the removal of early discounts in later contract years. Now you can add the price increase to that.

NRR change

CFO in Q3: “We ended our third quarter with a dollar-based net retention rate of 128%. … This quarter we have updated our dollar-based net retention rate calculation methodology to reflect operational changes to our customer account hierarchies. The trend of declining but stabilizing rate has been consistent based on both the previous and the new dollar-based net retention rate calculations.”

… Later in Q&A: “Yeah, I’ll take the dollar-based net retention first. And so, I wanted to call out, we’ve made a change in the way that we’ve calculated dollar-based net retention rate this quarter to better reflect the business itself. And so, this quarter, if you did the old way, be flat 124% to 124%. But the new way basically takes into the account the account hierarchies. The old way that we did it actually kept a static view of the parent accounts. And if there was a merger, a new subsidiary, division being shut down or something, that actually showed a churn and then a new business. And so the change that we made, it’s important to note, doesn’t change the business at all. It just shifts between growth and new to give a better reflection of what’s going on in the business. This is reflected in our 10-Q for Q1 and Q2. And so you can see the disclosures there. But under the old method was 124% to 124%. And under the new method, it’s 128%.”

From Q324 10-Q:

During the quarter ended October 31, 2023, we implemented operational changes related to our account hierarchy process for tracking customers. These changes resulted in an update to our methodology for calculating Dollar-Based Net Retention Rate. Under this new methodology, Dollar-Based Net Retention Rate would have been reported as above 130% and as 129% in the quarters ended April 30, 2023 (Q1) and July 31, 2023 (Q2), respectively. All prior quarter disclosures would be unchanged.

I really hate changes to KPIs that allow companies to change the goal posts. This doesn’t seem like that. What they are essentially saying is they were undercounting NRR before due to the way they tracked overall company accounts around mergers and separate divisions.

They restated only 2 quarters. Something happened in Q1-Q224 that made them realize their internal tracking of customer accounts (parent accounts) was allocating things wrong. So these mergers and division changes were being allocated to new lands or churn, not renewal. (PS this is something Snowflake encounters a lot, but they have decided to change their entire historical customer counts every year to handle it. I hate it!)

This makes me think their new method is more sound. It seems this change added +4-5pp to NRR.

  • Under old method, Q324 was 124% - flat to Q2 (stable).
  • Under new method, Q124 was >130%, Q224 was 129%, and Q324 was 128%.

Q4 was reported as 130% in the new method, so it moved back up +2pp either way.

Issues in your calcs

Problem 1: You are prematurely attributing a huge NRR adj from price mix in Q124, only 1 month into the first upgrade cycle.

You started pulling from their NRR w/ your own calcs in Q, all of one month into their price increase in a 12mo renewal cycle (as it happens at the next annual renewal of every Premium customer). I’m not sure why you think 128% needs to adjust to 121% here.

CFO in Q124 (one month into price change): “It’s been relatively the same. And so seats is about 50%. Price increase is about 25%, and the last is 25%. So there really hasn’t been any change whatsoever.”

I think his wording was confusing… he used price yield before, and price increase here. He later clarified it is both at that Truist conf. So you essentially have 50/25/25 breakdown of expansion TTM that was unchanged between Q423 and Q124 after the price increase kicked in. Not that surprising as Q1 was likely 1/12 (one month) of a 1/2 price increase (the first +$5).

Problem 2: the starting point for price yield contribution is 25% before price change, which increases to 40% by Q4. Your issue above (adjusting their NRR lower) magnifies as you adjust NRR down to 115.8%

We don’t know what Q2 looked like, but it had to be under Q3 as the price change continues through 3 more months of renewals (4/12 of 1/2 price increase now complete). Let’s say it was 29% from price yield (+4pp).

By Q3, it was 33%, seats <50%, tier< 25%. This means that the price increase took the price yield part of the mix from 25% to 33% (+8pp).

By Q4 (7/12 of 1/2 price increase now complete), it was 40%, taking it from 25% to 40% in contribution to expansion revenue TTM (+15pp) over FY24. That extra bump makes me think a majority of renewals were in Q4.

IMHO you really shouldn’t be recalculating NRR on your own without any of the inputs. But if you wanted to… as I said before, his breakouts above are on expansion revenue overall, aka the part of NRR over 100% (the same subscription renewal). Premium price increase took that part of expand mix from 25% to 40% (+15pp) over last 3Q (again, 7/12 of this first upgrade cycle now complete). For Q4 with NRR of 130%, this is at most 4.5pp of NRR. (.3 times .15). We don’t know the mix of land to expand in net new revenue/ARR so can’t really extrapolate this further.

I don’t like doing this, but it’s merely an example. I have no idea how you lowered all this to 115.8% in your own calcs – but I think you need to revisit.


WHEW that got long. Long story short, there are 2 ways to look NRR from GitLab that I think we should stick with instead of trying to extract our own.

NRR over FY24, old way:
128% → 124% → 124% → 126% (estimate from +2pp)

NRR over FY24, new way:
over 130% → 129% → 128% → 130%

FY25 will bring another EQUAL ROUND of revenue increase from price appreciation in the 2nd wave of this upgrade cycle. FY26 will have 2 months of it in its Q1. Revenue growth and NRR might start muting then after Q126 after we get entirely through the cycle. However, lots of new products in FY25 are coming to boost things before then.

As for the recent Truist fireside chat, I listened to it on the Quartr app as it is no longer available on IR’s event page. There was very helpful commentary as noted above.

-muji

21 Likes

This is what I love about this forum: the ability to break out of my own echo-chamber and get challenged to get to a better understanding of the companies in question. I went digging some more, as understanding the NRR dynamics is fundamental to an investment in this name imo.

With that said, I think that in broad terms I would agree with your point that my conclusion above is not wholly accurate wrt the “new” vs “old” way of calculating NRR, but I believe my conclusion is still valid.

That is because in my post I was referring to pricing impact in general, not only the one most recent price change to the Premium tier (I was not clear in my post, but that is what I assumed the CFO meant, and you showed above that that is indeed what he meant).

There were several recent pricing/subscription changes that they made that are all still playing out:

  1. Jan 2021 they announced phasing out bronze/starter tier over 3 years: GitLab is moving to a three-tier product subscription model
  2. March 2022 announced limits on free tier effective Oct 2022: Upcoming changes to user limits on Free tier of GitLab SaaS
  3. Mach 2023 announced increased price on Premium: New pricing for GitLab Premium

All of these are currently stil in play due to that discounting dynamic you describe above which is how they implemented those earlier price changes. The positive impact of price increases on NRR should, bar further price increases, fade in time. Perhaps they do manage to squeeze more out of their customers in future, and yes, there are other things that they can do to expand NRR like selling new products, but I wanted to isolate the pricing part only, as it is meaningful and has increased of late.

So first, what I did:

My simplified understanding is that excluding churn NRR for Gitlab is influenced by 3 things: pricing changes, mix changes/upselling (selling a different product/higher tier subscription), seat changes.

An example (skip this if this is obvious):

Let’s assume no churn and only two subs: a $10 one and a $15 one both per year.

So if I have 1 customer with 10 seats on the $10 beginning of the year, then my opening ARR is $100.

During the year I increase the price by $1 on the lower $10 sub to $11, sell two more seats on the lower sub @$11 each and upsell 2 of the original 10 to the $15 plan. Let’s assume no churn.

So at the end of the period I have:

  • 8 original seats @ $11
  • 2 new seats @ $11
  • 2 original seats @ $15

ARR = $140

NRR = 140.

And the contributors to the $40 expansion are:

  1. price: 8 x $1 = $8
  2. seat expansion 2 x $11 = $22
  3. mix: 2 x $5 = $10

Or, of the expansion:

  • price is 8/40 = 20%
  • seat is 22/40 = 55%
  • mix is 10/40 = 25%

So if I want to exclude the price change impact, I need to take out $8 of ARR, for an NRR of 132.

Or alternatively I just use the percentage to get to what I need to take out:

40 x (1-20%) = 40 x 80% = 32, so NRR is 132 ex the price impact.

And now on to Gitlab’s NRR:

The CFO did not give us churn but he did break out the % of expansion (i.e. of the part above 100 in the NRR) that was due to mix, price and seat expansion. So if I wanted to exclude price all I had to do was multiply out the expansion part of NRR with the % he gave us.

I think you are right that I should just have stuck with the “new” NRR calls for this year and not tried to calc new “old” ones without the info. Your argument that the new method did not materially change old numbers means we are in fact comparing apples to apples using those new numbers. So let’s take those NRRs and then work out what I was trying to get to:

Q1 24: >130%
Q2 24: 129%
Q3 24: 128%
Q4 24: 130%

Let’s ignore Q1 and reverse out the price impact in Q2 to Q4 using your % for Q2 and the CFOs for Q3 and Q4.

Q2: price impact was 29% (using your estimate)
Q3: price impact was 33% (per CFO)
Q4: price impact was 40% (per CFO)

So in 2024 ex pricing increases, NRR would have been:

Q2: 100 + (29 x (1-29%) = 120.59
Q3: 100 + (28 x (1-33%) = 118.76
Q4: 100 + (30 x (1-40%) = 118.00

Same conclusion…ex pricing increases (plural, sorry…), NRR was flat.

17 Likes

Microsoft’s competitor to GTLB just reported some very solid numbers: https://x.com/Beth_Kindig/status/1785724907403292736
Since all the companies in this industry seem do being doing well (see my post above about JFROG and TEAM being top performers last Q), I think this a situation of “rising tides lift all boats” (i.e., an industry doing well) instead of a competitor stealing business.

Best,
BTL
@laneylawyer on X
Long GTLB

1 Like

I remain leery as they are competing with Microsoft, who can (and likely will) give much of these things away to entice developers onto their platform.

Does that headwind ever truly go away? Even for developers who will swear off M/S, they will have to justify to Mgmt their spend here.

2 Likes