Who is end-user as indicator of long-term growth

Hi all,

Short-time lurker (June?), first time poster. Before anything else, I’ll echo the commonly shared sentiment that this is a haven on the internet and the stuff of dreams to have such a talented, knowledgeable and generous set of folks openly sharing their best ideas. I’m immensely grateful.

I’m fairly new to this style of investing, but having now watched the aftermath of AYX’s slowdown and FSLY’s missed earnings, I’ve had a nagging thought rattling around in my head. I’m intrigued by the interplay of who the end-user for our technology companies are and how the tech-savviness of the end-users affects the moat or durability of growth opportunity for our companies. In short, the thesis I’m exploring is this:

Tech companies that build tools for non-code based employees or consumers enjoy enduring competitive advantages due to their user base’s resistance to learning new software or technology.

Conceptually, this makes sense to me. If I am used to gmail’s user interface, a competitor must not only make a superior product, but it must be so superior as to justify my investment in relearning. This is even more so for a company considering a broad user base. Change management is costly and painful, so a software change must be clearly worthwhile not just on the basis of cost but also usability. If a technology is used primarily in the background by developers, engineers and IT teams, the users may be more nimble in finding cheaper or alternative solutions that eat away at the company’s high margins. Additionally, these companies where developers are the customers may be more prone to surprises in their earnings. As an outside investor, I might be less equipped to identify when decelerating growth is on the horizon.

Put another way entirely, the premise is that intellectual property is much less valuable for these companies than their brand identities and usability achievements with end users. A good tech idea can create tremendous value, but has a limited run of growth before competition heats up or catches up.

Ultimately, much of this overlaps with network effect, but I raise it as I think it builds upon that in cases where the “network” is simply the cost of retraining end-users, even if the tool’s “network effects” are dubious.

To explore this thesis, I’ve explored several companies currently followed here. I’d of course love insight on how past companies line up with this concept if it resonates with others on the board. This is a strictly subjective and qualitative analysis, but I’m curious how it pairs with the performance of past companies that this board has followed as well as present companies.

ZM - In this case, the end user is everyone. For a company paying for a Zoom license, they are considering not just the cost of the license, but also the end users who will interface with it. These end users are likely to be not just their employees, but also their customers and their vendors. Usability is paramount – wasted time, dropped meetings or IT support for downloading the right client and connecting audio peripherals are all much more expensive considerations than any small difference in license pricing. Once a company is using this technology, they are unlikely to change. For a company not using the technology, the network benefit of broad adoption elsewhere is reassurance for choosing to invest in Zoom as a solution. In this framework, one would expect Zoom to build upon its dominance and see limited churn through the years. End user = limited tech capable = good thing for long term growth prospects

DDOG - I’m not a techie, so please correct me if I’m mistaken here. My understanding is that the end user for Data Dog’s services are techies. They are developers and IT folks who have built their careers on adapting to new technology and services as they roll out and provide additional value. In many cases, they also possess the capability to build software to meet their needs should their use cases justify the effort. In this framework, Data Dog’s current dominance and growth strikes me as more perilous. Their tools are innovative and clearly create abundant value in the present moment, but their users are savvy enough to mitigate their value over time. Naturally, my thought is that if the margins are so good, time will provide the opportunity for others to chip away at their tools. I’d expect Data Dog to continue creating new and more powerful tools, but their established business may simultaneously wither away as companies learn from the flexibility of their tool and build (or buy) a cheaper solution that meets their limited needs. (As a paranthetical, DDOG is my second largest holding at present). End user = very tech capable = bad thing for long term growth prospects

FSLY - Well, this triggered the whole premise. When I invested in Fastly, I mistakenly assumed that a Shopify or Tik Tok built on top of it would struggle to meet their customer’s needs should they choose to switch providers, that the cost of moving away from Fastly would be inferior experience for their customers. Ultimately, I now think that Fastly’s advantage was a superior offering, but that their customers possess many of the skills necessary to replicate the neccesary components of that offering without paying Fastly’s margins. As such, companies flocked to Fastly to use the tech and try it out, but then, through near-comparable services from competitors and in-house capabilities, looked to move off Fastly. It’s not that Fastly doesn’t have a place in the market now – clearly, the infrastructure requirements will cause many to choose to outsource this and their product is excellent-- but that their margins will compress as they enjoy no “end-user” or “brand” benefit. End user = very tech capable = bad thing for long term growth prospects

NET - Honestly, this is a company I don’t know enough about to comment on. But, my non-techie understanding is that they would run the same risk of pressured margins. As their tech matures and is replicated by others, they also enjoy limited change management moat… their customers are plenty capable of adapting to a less friendly interface and/or picking and choosing their services from multiple providers. End user = pretty tech capable = bad thing for long-term growth prospects

CRWD - Crowdstrike is interesting in this premise. In some sense, their tool is built for end-users and runs on every employee’s system, but the end user has very little daily interface with the tool. also as I understand it, they have a plethora of tools in the background that, following the basic thesis here, would be subject to pressured margins EXCEPT that they seem to benefit from a beehive like knowledge network that a single company or competitor would be hardpressed to replicate. End user = more tech capable = bad thing for long-term growth prospects, though I think the network effects here more than negate this!

OKTA – Benefits from end-user being non-tech focused as well as network effects of integrations. End user = limited tech capable = good thing for long-term growth

AYX – I don’t know enough here to comment too fully, but strikes me as an in-between. The end-user here is clearly capable enough to use basic analytics tools, so likely fairly capable of adopting to new user interfaces. Additionally, it seems the size of data that alteryx thrives on is large enough that there might an outsize influence to the developer / techie viewpoint within the company when it comes to data solution selections. End user = tech capable, not code capable = neutral

DOCU – End-user is not tech-focused and usability, familiarity are key for successful adoption / use. End user = limited tech capable = good thing for long term growth

MDB – I chose this because I wanted to look back at companies from earlier years to see how they might have performed and how that might inform this basic premise. I think it’s an interesting example of one where on the surface, this premise would indicate trouble maintaining margins and growth over the longer-term – Mongo DB users are fairly tech-savvy, BUT they consistently build tools on top of mongo db that less tech-savvy users interface with regularly. In contrast to Fastly, Data Dog, or Cloudflare where the tool is a small, interchangeable component, the database structure is a bit harder to change without affecting the end-user’s experience (or appears so to my non-tech brain). End user = tech capable, but embedded permanently into built solutions = mostly good thing for long-term growth

TWLO – that brings me to Twilio. Again, the direct users of Twilio are tech-based, but the API’s tie directly into user experience and thus inform usability. A transition to new tech would again require considering usability, which provides a much stronger moat than cost or the broadness of capability. End user = tech capable, but tool’s secret is vast usability improvement for customer-type end users = good thing for long term growth.

Ultimately, this train of thought has come from thinking about how Fastly caught me by surprise and has me rethinking my conviction in DataDog and also holding off from jumping headfirst into Cloudflare. I’d love other’s thoughts on this as it pertains to DataDog and Cloudflare. Apologies if the premise got tangled up with a broader conception of moat and network effect, but I also felt that it can’t be considered in a vacuum as those play into the importance of usability as an advantage.

Long ZM, CRWD, DDOG, DOCU, PTON, MASI, TEAM, TDOC and SHOP and some odd scout ideas I won’t mention cause I’m not sure I can defend (I guess that should tell me something).

54 Likes

Conordot,

You said:

FSLY - Well, this triggered the whole premise. When I invested in Fastly, I mistakenly assumed that a Shopify or Tik Tok built on top of it would struggle to meet their customer’s needs should they choose to switch providers, that the cost of moving away from Fastly would be inferior experience for their customers. Ultimately, I now think that Fastly’s advantage was a superior offering, but that their customers possess many of the skills necessary to replicate the necessary components of that offering without paying Fastly’s margins. As such, companies flocked to Fastly to use the tech and try it out, but then, through near-comparable services from competitors and in-house capabilities, looked to move off Fastly. It’s not that Fastly doesn’t have a place in the market now – clearly, the infrastructure requirements will cause many to choose to outsource this and their product is excellent-- but that their margins will compress as they enjoy no “end-user” or “brand” benefit. End user = very tech capable = bad thing for long term growth prospect.

NET - Honestly, this is a company I don’t know enough about to comment on. But, my non-techie understanding is that they would run the same risk of pressured margins. As their tech matures and is replicated by others, they also enjoy limited change management moat… their customers are plenty capable of adapting to a less friendly interface and/or picking and choosing their services from multiple providers. End user = pretty tech capable = bad thing for long-term growth prospects.

I’m very much not a tech person, but I’ll share how I look at FSLY and NET.

First, FSLY and NET provide a CDN. This allows companies to use a CDN vs providing their own network at origin on servers maintained by the company. There are numerous disadvantages to maintaining your own delivery server at origin. It is a much better business decision for many companies to outsource this work. So, let’s make a basic assumption that the TAM of providing a CDN for enterprises and users is extremely large and growing.

FSLY and NET are disrupting the CDN offerings. They did this by making them faster. One big difference is that their networks are built on SSD as opposed to hard drives. This differentiates them from companies like Akamai that have older and slower networks. I don’t think many would dispute that NET and FSLY provide a superior, faster CDN product. I also think FSLY is proven to deliver the fastest by far, but in general, the superior speed differentiation between FSLY and NET, as I understand it, is not as noticeable given most current content and uses.

While it is true that a large part of CDN traffic is a commodity. FSLY routinely states that they are not in this lower margin side of the business. They target users who need and want speed because of the companies needs and content. I assume NET also targets this customer, but it looks like they have some lower margin business as well. (Individually, I was a free customer of NET). This allows them to maintain a higher margins. AKAM’s CEO for example recently said their margins are moving up to 32%. FSLY’s pitch is speed, consistency, customization and innovation. (Think ZM compared to other video providers).

More and more companies will outsource their CD to a third party CDN. Once moved out, few companies will move this work back in-house. It will not save them money. Tik Tok is an exception because they were forced by geo political pressure to explore this option. They moved all of their traffic off FSLY. FSLY has no POP in China (they have 1 in Hong Kong). Based on CDN search tools, it appears Tik Tok moved traffic to Akamai (who has a large presence in Mainland China). This is for at least the short term. Others have noted that Tik Tok is exploring, or are in the process of building, their own CDN.

Note, FSLY’s business model is to have fewer, but larger POPs. Some might argue that not having a POP in mainland China was and is a good decision based on geo political issues such as privacy etc (I don’t trust China), but it clearly hurt them with Tik Tok in the short run. NET on the other hand has 17 sites in mainland China. Akamai has exponentially more, but again they are slower.

Tik Tok’s high margin traffic might come back to FSLY, and indeed FSLY is holding open this CDN capacity in the short term in case it does. In the interim, FSLY will use this CDN capacity as overflow capacity. It doesn’t appear FSLY will make long term commitments for this capacity in the near term to other customers. Now, is this because they can’t sell this capacity, or is it because they are being a loyal provider and want the Tik Tok business back? Repeatedly and expressly, Bixby said it was the latter. I believe Tik Tok is likely holding out hope to FSLY, but I think the reality is that they need to move forward.

Another concern with FSLY that many of us had is the early release was not clear and alluded to the fact that Q3 revenue also took a hit from based on less use from other customers. Bixby addressed this on the CC and said it was related to unanticipated delays in new customers coming on to the platform, and that almost all of this delay (at the customer level) has been resolved. So, the express or implied statements suggest that any revenue shortfall was related to other existing customers dropping FSLY. (This would apparently include SHOP as some speculated.) In other words, customers in general are NOT moving to a lower cost provider. This conclusion is supported by FSLY’s DBNER of 147% and NRR, on a trailing 12-month basis of 141%. THis presumably includes Tik Tok.

FSLY long term is still growing its revenues at 40%, even with the elimination of its 10%+ customer revenues. (Short term might take a hit from Tik Tok). Keep in mind, FSLY’s network (like NET’s and Akamai’s) does not have unlimited capacity. Capacity must be built. (If my existing customers are using my network as fast as I can build it out can I really add significant new enterprise customers until existing customer use stabilizes?) This next Qtr it appears they will drop their capex slightly. Is this overly conservative if you have a great product? For Q3, this suggests to me that they will sell the old Tik Tok capacity and are putting capex into the edge@compute. Inany case, I expect that they will have a number of large enterprise adds in the QTR. Note, Bixby said (in an admittedly awkward way) that because of the way they measure their enterprise customers that the “net” increase of enterprise customer isn’t really indicative and that the “add” number is much more important, and they are happy with it. (Someone posted on this earlier). So, what is FSLY’s add number? As far as I can tell, they don’t disclose this (that is some BS - they need more transparency here). I do note though that they estimate 70% of signal science customers (~40) will likely be new FSLY “enterprise” costumers.

One last word on Tik Tok. I believe the Tik Tok scenario is an aberration. Tik Tok revenue likely caused the surge in revenue in Q2 (and a huge increase in the stock price) and also a more dramatic loss in Q3 (with an even bigger drop in stock price).

At the end of the day, FSLY (and NET) are high margin CDN providers for a reason. They are both growing nicely. They are disrupting the industry. The have SSD servers, and those would be hard to replicate in the short term by others. I would not invest in AKAM and it’s old infrastructure.

As for the need for speed and better long term prospects, the thought is that the IOT (internet of things) and 5G will increase the need for high quality CDNs. FSLY and NET stand to gain here. While current speed differences might be acceptable, they may become unacceptable increasing the demand for FSLY’s and NET’s CDNs.

FSLY, and I believe to a lesser extent NET, are developer friendly. Enterprises are able to customize their use of FSLY’s and NET’s network. As technology advances, the ability to customize grows. FSLY’s add of the Mozilla group shoes their commitment to development and innovation.

Now, the strength of both FSLY and NET over and above the high margin CDN traffic is that they both are expanding their other offerings. They are both offering security services and edge network services. NET is likely ahead of FSLY here with its secure offering and its edge network. FSLY’s acquisition of Signal Science’s is incorporating its security offerings already. As for edge, FSLY confirmed that it has moved its compute@edge out of Beta. FSLY’s offering appears very promising, but has yet to deliver any results. These additional services should certainly make them innovative “full-service” disruptors and allow them to maintain or improve their margins into the future.

One difference in NET and FSLY is NET appears to be going after all customers (I was a free customer with their mobile app) where as FSLY is focused on their enterprise customers (and in particular developers). In any event, I like both stocks. Because of my belief that FSLY had better tech, I hold far more FSLY than NET. Right now, that appears to be a mistake. As Saul noted, better tech doesn’t always win. Nevertheless, I maintained my FSLY position after the Conf Call (by closing my short against the box) because I believe that (a) FSLY is not losing enterprise customers en masse, (b) Tik Tok was an aberration - albeit a big one, (c) FSLY is growing and building its revenues at 40%+ over the long term, (d) secure@edge and compute@edge will be revenue and MOAT drivers, and (e) IOTs and 5G will boost longer term future revenues and margins.

If I was investing today, I’d likely buy equal shares in both FSLY or NET. Admittedly, if you had to pick one, I think NET is currently a safer pick than FSLY because NET has recently delivered superior enterprise customer growth and has already rolled out security and edge capabilities.

One last point, others have noted that Artur Bergman (FSLY co-founder) has sold “a lot” of stock recently. However, he’s 39 and his sales are clearly pursuant to a 10b5-1 prearranged sales plan. He still holds $560 million worth of FSLY stock. Every financial advisor in the world would be pushing him to do this to diversify or create assets he can spend. $157 million of his holdings are in annuity trusts (likely GRATs). These essentially allow him to give away (to family members in the case of GRATs or charity in the case of CRTs) the appreciation on the stock gift tax free. This means he’s placed 27% of his holdings in trusts betting that the value goes up. If the value doesn’t go up, GRATs don’t provide a benefit and CRTs don’t work as well.

Mike
Disclosure, I hold 70+ stocks. Current top 10 holdings: FSLY, BILL, ROKU, ZM, CRWD, SHOP, DDOG, TTD, APPN and MELI. (NET is currently #33)

20 Likes

Unfortunately, I have make one big correction:

So, the express or implied statements suggest that any revenue shortfall was not related to other existing customers dropping FSLY.

Mike

Hi ConorDot.

Excellent post and I appreciate the creative thinking.

One aspect that you’re not including in your analysis (at least that I saw) is who pays for the product.

With a company like Alteryx, desktop tools may come out of the budget of the end-user department, although server tools may be bought by IT (I’m not certain of this, but it makes sense to me, having spent my career in IT).

With a company like Okta, it is almost certainly the IT department making the decision to purchase an identity management solution, and it is coming out of their budget. Of course, IT will include “non-tech user-friendliness” in their decision-making process. But there is a tech-savvy group controlling the purse-strings.

Does that help?

Fool on!
Thanks, and best wishes,
TMFDatabaseBob (long AYX, OKTA)
Maintenance Coverage Fool
See what a “Coverage Fool” does here: http://www.fool.com/community/community-team.aspx
See my holdings here: http://my.fool.com/profile/TMFDatabasebob/info.aspx
Peace on Earth

Please note: I am not a member of any newsletter team. My opinions are my own and do not necessarily reflect those of the TMF advisers. I am not an investment professional, merely an investor.

8 Likes

Excellent post and I appreciate the creative thinking.

One aspect that you’re not including in your analysis (at least that I saw) is who pays for the product.

I agree, interesting out-of-the-box thinking. The user-payor relation can be very simple or quite complex. In a business like Dress for Less, Ross Stores (ROST) the user is the payor. Although not a Saul style investment, the off-price business is interesting because there are many more poor people than rich so the market is huge, the 80%, and we all need clothes. Livongo is totally different, the relatively small number of payors don’t use the service, their members use it for free. The payors benefit because as their members use the service their healthcare costs drop. Members benefit from better health and payors benefit from lower healthcare payments. Fastly is another category, since they charge by usage, the more IoT devices use the service the higher Fastly’s revenue. That’s why I’m excited by the fact that IoT devices exist by the billion. It reminds me of ARM’s story, ARM’s RISK chips are “inside”* all sort of mobile devices like smartphones, printers, disk drives, automobile components, etc. Fastly is the ideal CDN for these devices to connect to the net.

There is no easy way to convert these numbers to “high conviction” that I know of but a mix of utility and large user base is what I like to see in addition to moat and network effect to keep users from defecting. Zoom. Teladoc, and LIvongo seem to have these qualities. I’m betting that Fastly does too. Additionally, this high conviction has to be tied in with the market penetration “S” curve. Even with all these qualities growth can only happen with relatively low market penetration, when there are new users easily to be found. At the top of the “S” curve these businesses are cash cows, value stocks, but not growth stocks any longer. It’s not the Law of Large Numbers as some claim, it’s market saturation that caps growth.

Denny Schlesinger

For those who don’t know or remember the “Intel Inside” blurb, large Intel CISC chips were inside PCs and servers while smaller, cheaper ARM RISC chips were powering mobile and peripheral devices. Users might have an Intel chip on their desktop that they knew and raved about and half a dozen or more ARM chips that they didn’t even know existed. Pundits kept saying that Intel would eat ARM’s lunch. It never happened.

22 Likes

First off, Welcome to the board. I’m a long term member, but have become an infrequent poster. There’s a lot of noise and redundancy, so, not wanting to add to the clutter, I have become very circumspect about posting.

But, your post is an interesting perspective. I might have something worthwhile to contribute. Just for background, I spent 30 years in IT at a Fortune 50 company. I retired 10 years ago, so it’s safe to say that my tech knowledge is now nearly 100% worthless. 10 years is an eternity in the tech industry.

But, I think there’s still value in my people knowledge. I think human behavior stays pretty much the same long term. And that’s the point of view I have on your post. There are a lot of techies who follow this board, if I’m wrong, I’m confident I will be corrected.

I think the resistance to change by the non-techie that you based you thesis on is actually shared by the folks in the tech business as well. I was one of the managers of a technology insertion project at my company. In a nutshell, we decided it would better serve the needs of the company to migrate from centralized data centers, predominantly using IBM equipment and related s/w to a distributed UNIX network running on smaller boxes. At the same time we migrated from COBOL to C and from IBM’s hierarchical IMS database and network to relational Oracle database. My role was to manage the team that established the standards and naming conventions for the software components of this migration. Let me assure you, there was enormous resistance inside the IT community.

What actually occurred was that many of the older employees moved to maintenance teams in that many of the older systems were not replaced. Others left the company for different employment where their skills were still in demand and others decided it was time to retire. We maintained the two infrastructures running side by side for quite a long period of time. In fact, when I retired, which was a good 15 years after this migration, some of the older systems were still running despite the fact that they were COBOL IMS applications running on a mainframe. The applications were enormously complex, unique to our business so there were no COTS alternatives and integrated with the legal certification of our highly regulated primary products. Redevelopment on the new technology platform was cost prohibitive.

The new technology became the province of a lot of new hires and younger employees who had not been with the company very long. Of course, I’m generalizing. My observations were largely true, but not universally true. So, I guess what I’m saying is that there’s an age factor to your thesis. Even in the tech industry, you will find a lot of resistance to change but it’s on a curve related to the age of the employee.

17 Likes

Nice thought-inspiring post.

I think your thoughts are just a subset of “stickiness”. At first I was going to leave it there but I actually want to go a step toward the extreme and say this is dangerous if taken as a blanketing thought that would prevent a truer understanding. What I mean by this is that each company (for the most part) has a unique blend of stickiness concepts. The first two are explored in the original post I think:

  1. Switching costs.

  2. Network effects can be sticky if they prevent people from moving away when a solution. For example one that is required for multiparty interaction like Zoom meetings. It means a group has to decide to move together or individuals have to be willing to use multiple solutions or miss out (e.g. switching away from using Facebook with friends).

…there are a number of others. Here are just a couple:

  1. Integration complexity. a) If a company sells a platform which users then build further solutions on top of, switching away can be a big project. b) the solution could integrate with other things, like security systems or workflow automation systems, and all of that might need to be re-created too.

  2. Proprietary data. If a company generates a lot of valuable data within a system then that data either needs to be processed and migrated, if possible, which can be a big undertaking. Think about switching from GoogleDrive to…whatever else people might use. You’d have to go document by document to export it and then import it, and even then the document formatting might not translate perfectly and need fixing. You may even need other tools or custom software to do this.

  3. More…

“Ultimately, this train of thought has come from thinking about how Fastly caught me by surprise and has me rethinking my conviction in DataDog…”

This is where I think things might get dangerous. These are so completely different. Fastly (and Cloudflare) CDN business is not real sticky but it is funding the ongoing development of their underlying edge networks and other services. I think you went wrong by not knowing this detail of their business. The other businesses at both of these companies can be much stickier then CDN services. This has absolutely nothing to do with DataDog’s business however.

My point is this: This thread is a nice thought primer but just like we don’t have a magical stock screener to find good investments I don’t think you can sum up tech in this way. When you dive in to build a thesis on stickiness per company I think, to use a food analogy, you build a palette and the flavors start to taste different from each other. Also, sometimes it doesn’t matter at all. It all depends.

8 Likes