AYX angst

I’ve seen a bunch of different threads. I wasn’t sure which thread to post this in since there are at least 3 going on between here and NPI, so I started a 4th :). Here are the threads

The main Concerns I’ve seen raised about the prelim numbers are .

  1. Customer Growth isn’t accelerating and is slowing down on a percentage basis
  2. Revenue Growth Slowed Down
  3. TAM isn’t as large as we think it is

Lets take each one of those separately.

GauchoChris had a great post https://discussion.fool.com/here-is-the-customer-growth-for-the-… which is worth reviewing. Shows the customer counts quarter by quarter along with calculated % growth rates. Some people rightly said that AYX’s absolute numbers look ok but they aren’t accelerating their customer growth from a percentage standpoint.

I think it is important to point out what AYX considers a customer. Here it is from the from the 10-q "
We define a customer at the end of any particular period as an entity with a subscription agreement that runs through the current or future period as of the measurement date. Organizations with free trials have not entered into a subscription agreement and are not considered customers. A single organization with separate subsidiaries, segments, or divisions that use our platform may represent multiple customers, as we treat each entity that is invoiced separately as a single customer. In cases where customers subscribe to our platform through our channel partners, each end customer is counted separately."

So basically the customer count is relatively unhelpful as it doesn’t tell us how many users or what deal size. They could have “one” customer that had 100 users, or they could have 5 customers that were 500 users. Just depends on how the customer is being invoiced. Honestly I wouldn’t pay too much attention to the customer count other than that it needs to keep increasing.

Revenue Growth
More importantly we should be paying attention to revenue growth. On the face of it the 53% is a bit of a slow down from the 58% last quarter. However we just don’t have enough information because we don’t have deferred revenue. As you all know how revenue is being recognized has changed with the switch from ASC 605 to ASC 606. I suspect they are trying to show us that growth is intact by including billing growth of 60% in that press release.

Here is a tidbit from their last conference call A lot of good one-liners in there. A question for Kevin as a follow-up on – just looking into Q4. Obviously you’ve had a really strong quarter a year ago with calculated billings, the metrics that everyone looks at, growing north of 60%. I’m just curious how you’d be encouraging us to look at that setup heading into Q4, just anything to call out in terms of seasonality. I mean, I know you mentioned there’s a little bit more favorable seasonality from a rev rec perspective in Q3, but just any thoughts on bookings or billings basis heading into Q4.

Kevin Rubin

I mean, we don’t guide to calculated billings. We obviously understand the importance to you guys. What I would just say is go back to Q4 guidance and look at that relative to growth rates and how we think about the year.

So for the first time they gave us calculated Billing. I suspect they have done this because they want to show that growth is intact and in line with the last couple of amazing quarters. We will need to see deferred revenue to see how much business they really booked this quarter but if calculated billing growth is 60% then they did just fine. The 53% revenue growth may just not be an apples to apples comparison for a couple of reasons. 1) switch in account practices, ASC 605 to 606 or 2) Different mix of business booked so the contract has more revenue recognized later instead of up front. 3) Revenue recognition seasonality as they mention in the tidbit above.

The one last twist in this which I’m not sure if they are saying is for q4 2018 or q1 2019 for their ASC 605 to 606 switch. Maybe others can weight in. This is from q3 2018 conference call.

Before we turn to guidance, I’d like to update you on our adoption of ASC 606. As we discussed with you last quarter, we will no longer qualify as an emerging growth company after December 31, 2018, and we’ll adopt the ASC 606 when we file our Form 10-K in early 2019. We continue to evaluate the potential impact of ASC 606 on our financial statements and have not yet reached a final determination

I think this is worth watching because I’ve always been skeptical of their TAM numbers. One doesn’t say they are targeting 30 million disaffected excel users and then price their product at 4-5k per seat. Growth will become harder for them at some point. My gut feeling is that this isn’t that point but we need to see all the numbers for this next quarter, conference call and guidance.

I still feel good about my AYX investment.



I agree with everything you said.

Here is the rev. growth history

2017  55  52  52  55
2018  50  54  58  53

No pattern here, just steady excellent growth in the 50's.

A couple of other points:

The gross margin has been improving by about 6% each quarter, so gross profit is going up faster than revenue.

The net expansion rate dipped slightly to 129, but still excellent. 

I like that they are getting good growth from both existing and new customers.

On TAM, as companies do more and more data mining it should be expanding.

I will say I was wanting to see the rev. growth at 58% plus, showing acceleration, but how can you complain about steady growth in the 50's.


I think this is worth watching because I’ve always been skeptical of their TAM numbers. One doesn’t say they are targeting 30 million disaffected excel users and then price their product at 4-5k per seat. Growth will become harder for them at some point. My gut feeling is that this isn’t that point but we need to see all the numbers for this next quarter, conference call and guidance.

I think you hit the nail on the head. Perhaps they are overestimating the number of folks who would switch.

BTW, they were asked this question in the last conference call by Michael Turits from Raymond James.

Michael Turits
Well, it’s hard to say you really needed to be doing anything better nevertheless. Is there any sense that you need to increase penetration by lowering price points? Obviously, the – so Tableau successful with that strategy recently.

Dean Stoecker
Well, we play, obviously, in two very different markets than Tableau, and I think they had to do it because there has been tremendous amount of price compression in their space. Our primary competition is SaaS, and we’re actually probably relatively cheap, maybe even too cheap compared to SaaS. So we don’t think that there’s a pricing scenario where we drop prices. That said, if we get to that point of 40% or 50% penetration of the 30 million disenfranchised analysts around the world or the 1 million quants who aren’t productive at being able to deploy machine learning algorithms without a tremendous amount of work, at that point, there might be some reason to move downstream to get verticalized solutions that are – or more inexpensively priced. But today, we’re in the – even in our lowest penetrated accounts, we’re in the low single digits in penetration rates. So we don’t see a scenario where there’s a drastic price change either way.

Kevin Rubin
Yes. Michael, I think just maybe…

Michael Turits
I’m sorry, Kevin.

Kevin Rubin
And Michael, just to add to that, I think we’ve talked previously. We do have flexible pricing arrangements, especially in our larger enterprise accounts, that do allow us to take advantage of pricing when you get to hundreds and thousands of users. So pricing has not really been an issue for us from an adoption and penetration perspective.


So yes, there is some concerns that $4-5K per seat would not cause some companies to easily replace the current offering – which reduce their TAM. Your primary competitor is SAAS offerings but your TAM includes low price Excel users. We have to follow their customer growth, and customer acquisition cost, to see if the lower pickings are over.


SAS is their competitor and was misspelled as SaaS in the transcript. Comparing to SAS, Alteryx is very cheap.


SAS tries to eke out every ounce of consumer surplus from its customers by having very extensive price discrimination. If you are a consultant they charge you more than if you are just a user. If you are an online service provider, they will charge you massively more than if you were using their product for just internal uses. And they charge you a big up front cost followed by about 28% of that amount as an ongoing license fee every subsequent year (it used to be 50%).

Their basic Windows Analytics package costs $8,700 for the first year. It includes BASE, STAT, and GRAPH products for basic data processing, advanced statistics, and automated production graphics. You will need to pay a chunk extra for modules to access databases directly or using ODBC. See Order SAS® Software

For more than one user or versions that run on Linux, Unix (or mainframes) expect to pay more. License fees can easily reach past $100,000/year. Even a single user license for tools like Enterprise Miner can cost something like $140,000 for the first year.


SAS is the granddaddy of data analysis. It’s been around for years And has gone through numerous enhancement cycles. The drawback is you have to be a hacker to use it. For the average user, you may as well tell them to just program up what they want to do in FORTRAN. AYX brings high end statistical functionality together with ease of use.

Are there analytical capabilities available with SAS that are not available with AYX. I don’t know, but I’d venture there are, the question would be how frequently are those routines invoked. I’d venture seldom.

Also, AYX brings new drag-and-drop capabilities to data prep. Not often discussed, but really data prep is the underlying basis for validity - get this wrong and all those stats are either meaningless, or worse, misleading. In addition, AYX provides drag-and-drop capabilities to the problem of data sourcing. Sourcing the appropriate data can be a programming intensive, time consuming exercise. In fact, certain analysis which may provide very interesting and useful insights is at times forgone because sourcing is so intractable. Not to suggest that AYX does away with the problem, but it greatly expands the ability to access more sources more easily.

So, it’s unlikely that experienced SAS users will decide to use AYX (that would be an interesting study, I’d like to know if a significant number of SAS users are abandoning it in favor of AYX). But, anyone who has the basic knowledge to perform data analysis but lacks SAS skills and doesn’t want to take the time to scale a steep learning curve will readily opt for cheaper and far easier to use AYX.

Careful, much of what you just read is opinion based on prior experience in the field (I was never a SAS user, but I knew several). I can’t back it up with stats.


SAS is their competitor and was misspelled as SaaS in the transcript. Comparing to SAS, Alteryx is very cheap.

Saw an interesting clip vis-a-vis SAS and Alteryx in a head-to-head:


“We had a large file, 250 million e-mail addresses…”

SAS took a week

Alteryx took an hour and a half.


“We had a large file, 250 million e-mail addresses…”

SAS took a week,

Alteryx took an hour and a half.

That says it all, doesn’t it? No more discussion necessary.



SAS took a week, Alteryx took an hour and a half.

That says it all, doesn’t it? No more discussion necessary.

Not really. It’s easy to use tools stupidly, especially as regards performance. There’s never anything in the documentation or training that says “this should take ten minutes or so”. And often there’s no business reason to make things any faster.

It would say a lot more if they said that they had complained to SAS who had sent out people to help and the conclusion was that the week long run time was about right. Often customers are idiots. Often companies are happy to let them be idiots because the solution is for them to buy more licenses and bigger machines and generally throw money at the problem in a stupid fashion.

I would be willing to bet that if SAS put somebody good on the problem it would take them a few hours to get the run time down to less than a day (that’s a WAG of course, but backed by a lifetime of experience). And maybe with a little work they could get it to be even faster than Alteryx.

Anyway, this isn’t to say that Alteryx isn’t better or that SAS isn’t old and slow, but “that says it all” is just wrong, at least from what we know from the video.



While I agree with everything you wrote, I think Saul’s observation remains spot on. Sure if you have a seasoned SAS user with many years of experience they most likely could write the SAS program to run as fast if not faster than AYX. That would be advantageous if the analysis were repetitive in order to compensate from programming time as well as run time.

But, the fact remains that it didn’t take a seasoned AYX user to get the analysis to run quickly, it took someone without a lot of experience (I know this for fact as AYX just hasn’t been available all that long).

So get your results in a week with average IT skills, or get results in a half hour with about the same skill. And, pay less for the product to boot.

As I mentioned before, SAS is the granddaddy, there’s probably no data analysis that you can dream up that SAS can’t handle if you have skill set and experience, so I’d venture there’s probably some hyper-sophisticated stuff that SAS can do that AYX can’t touch.

I knew guys using SAS in relational to computational fluid dynamics (no, I don’t know exactly how) and 4 dimensional space (again, I don’t know how, but they were using it to help solve a complicated geometric problem), possibly out of reach for AYX (today). These guys were PhD mathematicians with years of applied math and SAS experience. Not exactly the AYX target user group.


I use SAS in my job (academia) as do a lot of us dinosaurs who grew up in the pre-R and pre-Stata world. We work with large datasets. I don’t know anybody in my field using AYX to analyze data, in fact when I first heard of AYX six months ago I asked a bunch of people around me and they were also unfamiliar with it. I suspect this is less true of people using SAS in the business world and would be curious to know.
Of course you need to know what you are doing with SAS and R (know the research question and how to go about analyzing the data), it sounds like an advantage of AYX is that it may reduce the need for technical skills and time needed for analysis.


I like reading this post. It means that AYX is not even close to securing its TAM. Worse part of my Master thesis was having to learn to write SAS code. One wrong placement of a word and you would lose 24 hours of data analysis. The dinosaurs may not switch to it, but the next generation most likely will if it saves time.