Alteryx (AYX) conviction update

Well, Saul mentioned Alteryx (AYX), and my initial thought was:

I would probably put AYX at 8, just ahead of HDP and TLND, because AYX is growing faster and closer to profitability.

I was giving them an 8 because, like with HPD and TLND, I just can’t really understand their product first-hand. I just think there is just so much competition with data companies and I find it hard to differentiate. Still, at an 8, my conviction in AYX was higher than on TLND (7.5) or HDP (7.5) because of AYX’s great financials. I simply wasn’t buying AYX because of the price. TLND and especially HDP are sooooooo much cheaper. I’m glad I looked further. Saul said:

Bear, The more I think about this… I just don’t know! Sure anything can happen, I’m well aware of that. But with a dollar-based retention rate of $135 (and rising), doesn’t that mean that this year’s customers will provide $135 in revenue next year for every $100 they provided in revenue this year? And presumably at lower acquisition expense. That means revenue will “automatically” grow about 35% next year just from old customers, not counting new customers acquired during the year, barring some very unforeseen event.

That seems to me to be fairly secure growth.

I dug some more, and I’ve finally seen enough. I’m in for a 4% position. I’m raising my conviction on AYX because I do think it’s likely that their customers, who understand their offering and seem to love it, will continue to spend more and more with AYX. So I’m moving them up to 8.5:

ALTERYX (AYX) - 8.5. Almost all revenue is recurring, they’re on the cusp of being profitable, they are growing like crazy and have been for years. Their astounding gross margin means lots of ability to leverage – which they are doing as costs drop as a percentage of sales. Financially speaking, they’re a 10. But because I am unable to evaluate their offering versus their current and potential competitors, I’m giving them an 8.5.

I know it’s arbitrary and these things could always go either way, but 8.5 was enough for me to jump in. Because I’m now that much more confident in AYX than in TLND or HDP, I’m willing to take a small position in AYX despite their much higher valuation. Especially as fast as they’re growing. And they’ll soon be showing some EPS, which could kick things into another gear.

I’m still not as confident in AYX as in my top few positions, but it really would surprise me if this business faltered much.


PS Alteryx reports on their December quarter on 2/21 after markets close.


With the important caveat that while evaluating this type of product used to be my job, since retirement in 2010 I have not been following this field closely. IMO Alteryx’ product offerings currently own the analytical space they are in with respect to ease of use and a comprehensive, integrated tool set.

SAS used to be (maybe still is) the number one s/w tool for the kind of statistical analysis that AYX offers, but it was not easy to use. You pretty much had to be a statistician to start with in order to effectively employ SAS. Also, most all the data preparation and integration had to be done with other tools prior to analysis with SAS. Alteryx has greatly simplified that process as well. In addition, the SAS user interface was a beast, but that may have changed. I don’t know.

For the time being AYX seems to not just be the premier product but maybe the only product that simplifies data science with a comprehensive and easy to use suite of tools. The main thing to watch out for (again, from my somewhat dated perspective) is that I don’t see much of anything, other than R&D spend that inhibits another BI product company from providing similar offerings. Their moat, if any, is going to be their customer base. This is the kind of product that once embedded in a company is not easily discarded in favor of something similar from some other company. Once they become familiar with the Alteryx toolset, user will be loathe to give it up. Satisfied customers beget new customers.

There is also value in being able to share not just analytical results, but an analytical package (sample selection and stratification, filters, etc). This provides somewhat of a network effect. I don’t want to make to big a deal out of that, but I thought it worth mentioning.


Hi, Bear.

I’m a database-data-software person and I spend my time doing the sort of work Alteryx does (and I own some of their stock).

Software like this has been around for a long time. Microsoft has a product bundled in with their database offering (SQL Server has SQL Server Integration Studio), and I used a piece of software 20-odd years ago called Mercator that did the same thing. They aren’t alone in the space.

Alteryx allows non-technical people to bring different data sources together, merge and process the data, and then report on it, including doing big-data type things (see the big-data discussion yesterday). I don’t know how easy it is to use, or how customizable it is. Software like this tends to do 80% of what you need very easily, and then makes the other 20% very difficult (we call it the 80-20 rule in the software biz - 80% of the work takes 20% of the time, and the remaining 20% takes 80% of the time). If you have some weird data that you need to process, it can be very difficult to get a tool like this to do it.

My big concern is their moat - it won’t be hard for a competitor to improve their offering so that it competes on functionality and undercuts on price. It would take time for competitors to take away existing customers, as people have invested in learning the software and creating solutions with it.

If Alteryx does well, I suspect they’ll be bought by a company like Microsoft or IBM. If they aren’t acquired, I would be concerned about a competing product that offers more functionality and/or under prices them.

It will be interesting to watch. In the sort term, however, I like the analysis you provided - thanks!




Sounds like we both came to the same conclusion (though yours was much more succinct)…


SAS has the Visual Analytics. Still requires SAS (language) knowledge, still a large player, unparalleled depth in high end statistical methods and in domain specificity in some verticals.

Qlik seemed strong, don’t know what’s happened post Thoma Bravo. Tableau is probably the best pure play public company in this space. I have shares of DATA.

Microsoft made a very strong move with the Revolution Analytics (R language) acquisition a bit back, and frankly, you can’t ignore the bottom tier of the market that is just Excel charts. SAP and SAP have competitive offerings, too.

1 Like

Thanks for the update. It does not surprise me that SAS is still the best thing out there for statistical heavy lifting, but it truly caters to mathematicians or at least people who are in the sciences who use statistical tools as an ongoing part of their jobs.

To be honest, I know nothing about Qlik. Somehow that product just never made it on my radar. Same for Revolution Analytics - maybe these guys came along (or established a market presence) after I retired?

When I was still working, I got pretty excited about Tableau. Great for visualization, although they had fewer analytical tools than one would have hoped. While it could be used stand alone, if I recall, they provided an Excel integration such that one would do the analysis with Excel, but the graphic output via Tableau. They provided virtually nothing with respect to data preparation. Of course, they’ve not been sitting on their thumbs since 2010. I don’t really know what their product offerings look like today.

As for the “bottom of the market” I suggest you spend some time at the Alteryx website and take a look at the entire suite of tools they offer, it goes way beyond just the statistical buttons. In order to do a statistical analysis you need to do a ton of data preparation first. Alteryx has tools that greatly facilitate that part of the job. If Excel (or any other similar tools) provide that kind of functionality, it’s news to me, and I’m a current Excel user.

In case anyone missed it, Alteryx has a lot of strengths in their tool suite that are not statistical analytical functions. A statistical analysis generally requires a ton of background work with respect to sourcing, accessing, cleansing, integrating, sampling, stratifying, etcetra-ing the data. Most of these functions are supported by Alteryx with powerful, easy to use, tools. As with the analytical package, all the tools provide a click and drag GUI.

IMO this is one of the reasons (maybe the primary reason?) that their tool suite is gaining market share. In that they’ve nicely partitioned the functionality, I’m confident that this is central to their “land and expand” strategy. Come for the analytics, stay for the data prep.

But, the fact remains, their no secret sauce so far as I can see. Their is little to keep competitors at bay other than grabbing as many customers as they can before the competitors catch up. Customers beget customers.

Let me provide a couple of different examples. Just about everyone is familiar with the Adobe product, Photoshop. It’s even become a verb, as in “that photo has been photoshopped.” Is it the only photo processing s/w out there? Of course not, there’s tons of them. Is it the best? I don’t know, but I’ll wager that there are a lot of comparable products. But, Photoshop was the first that really catered to the professional photographer. A culture of users grew up around the product. If you’re a pro, or even a serious amateur most likely you’ll buy Photoshop to process your photos without even reviewing the field.

I’m a musician. I have a DAW (digital audio workstation) installed on my computer. The industry standard in called Protools. I don’t use Protools because it’s Mac based. I’ve never owned a Mac. I think they’ve got a PC version now, but I’ve used a PC based DAW for years and years, no way am I going to climb that learning curve with a new product. But if I were starting from scratch, I’d buy a Mac and Protools. Again, it’s the product culture. With Protools there’s just no problem you can encounter for which that someone else hasn’t already found a solution/work-around.

If Alteryx has a moat, it’s going to be their customer base. They are in a crowded field with a lot of similar product offerings from the competition, but they do offer (I think) a very comprehensive and well integrated suite of tools that may offer greater functionality and ease of use than any of their competitors (at present). IMO, so long as their data spill was a one time event by a careless employee rather than indicative of the process failures of a cowboy hacker shop they’ve got a chance of long term viability.


Currently working data guy,

I have used SAS and i sit in the middle of two groups that still do.

I don’t know anything about AYX and have not seen it at my work place. I just checked our documentation store and it looks like finance is using it for some pre-processing of data.

SAS used to be (maybe still is) the number one s/w tool for the kind of statistical analysis that AYX offers, but it was not easy to use.

SAS is still #1, but R is growing fast. R is an open source solution, it is no where near as feature rich as SAS, but with the recent libraries in python that integrate R, it will not be long before the only people using SAS are legacy systems that would rather pay the licencing fee, instead of porting their code over.

You pretty much had to be a statistician to start with in order to effectively employ SAS.

Still true with R, but no matter how simplified they make the process, you will need to be a statistician in order to do statistics. Yes, linear regression is available on most tools now (I think you can use excel for it), but the really interesting stuff in statistical analysis are also getting more complex. In order to understand the new stuff, you will need an advanced degree and keep reading.

Also, most all the data preparation and integration had to be done with other tools prior to analysis with SAS.

On this point I disagree. I am what is now called a data engineer, and we can prep the data quite richly with SAS. Again I would argue that python is easier, but SAS is just as good a tool as SQL.


Data Guy,
Thanks for the update. The depth of talent and knowledge available on this board is amazing.

Like I said, my experience goes ends in 2010. And to be honest, my experience with stat packages ended earlier than that when I was the lead analyst for QA systems where I worked. Of all my years in IT, probably the most challenging specs I ever wrote involved the statistical sampling for receiving inspection. And to be honest, that was pretty light weight statistics. At that time SAS offered precious little with respect to data prep, but then there really weren’t any tools that did it. Pretty much all custom code.

I probably should have clarified a bit more. The Alteryx tool set provides a really easy to use interface for performing a fairly sophisticated set of statistical analysis routines. You still need to know what you’re doing in order to get meaningful results. This has always been true.

I once reviewed a financial spreadsheet created by a finance guy. At first he didn’t even realise that his numbers didn’t look right. One of his colleagues told him that his data couldn’t be right (I don’t know how long he’d been providing estimates with this spreadsheet). He asked me to review it for him. So I climbed way back into his data feeds and figured out that he was redundantly counting a rather major piece of the product, he had created dollar input for a third engine on a two engine vehicle. In that his spreadsheet was all dollarized there was no way he would have ever discovered it just by looking at the spreadsheet.

@brittlerock: I think you misread me on your first reply? Excel is the bottom of the market, not AYX. An enormous amount of BI gruntwork happens in Excel formulas. I’m not looking down on that, either. Spreadsheet are a powerful tool (see any number of online references to engr. undergrad courses that use it for numeric methods solving a 2-d heat transfer problem). Regardless I think you’re off on your point on data prep/ETL. That is SAS’s bread and butter. They likely as many people maintaining data connectors than work in R&D at Alteryx. I know what I would rather connect to DB2 running on HPUX, at least. Maybe you could explain what you meant if I’m misreading? From what I can tell the Alteryx hook is that it doesn’t require as much technical experience to set up or math knowledge to use. Not to trivialize that, either, I’m sure that’s a potent selling point. It might work against the moat you’re talking about with with Photoshop or Pro Tools, though, since the ease of use reduces the friction between other visual tools.

@hrse: current data R&D guy here (among a couple of other domains of modest competency), working for a private competitor of a public company recently discussed on this board. I think you have the R situation relatively pegged. That being said, my impression of the new SAS CTO is very positive and they’re still a factor. I am a python guy primarily, now, and am familiar with numpy/scipy/pandas and like them. I also think you have the core problem relatively pegged, too: you can’t dumb this stuff down, and to know how to interpret nontrivial regression analysis & modeling you really need something approaching postgrad statistical exposure(as you know, most people have no idea how little they know about probability). I wish I had time to poke around with Alteryx, but that is my biggest concern (how much it’s possible to dumb down the irreducible complexity of the domain).

In general, I think it’s a good market, though, and I don’t mean to be negative on AYX (though, again, DATA is my preference). You have MDB, et al. of the world enabling petabyte and larger scale data sets, enormous streams of data being generated, and most importantly a reluctance (aws snowmobile/snowball notwithstanding) to hand over what are essentially the crown jewels of any company to a third party. They fact that AYX has a cloud and on prem offering is smart (a low capex onramp for accounts that can scale to with account maturity).