With all the talk about Viya... There's no question that the Viya platform brings capabilities to the analytics world that could really differentiatie SAS solutions from other competing offerings. Do we know of any plans to push it more as capability that empowers business solutions as opposed to the research environment? Seems like there is an opportunity there?? Begin to remove the open source comparison?
49 replies (most recent on top)
"Sometimes folks have disordered personalities which prevent them from glimpsing reality."
Sometimes they have so much money and perceived power of choice they live in a different reality than the majority of us.
"...I think a lot of the old guard (Dr. G especially) don't understand, or don't want to acknowledge, the amount of competition that they have now..."
"They can't understand or acknowledge this because they lack the humility necessary for those actions."
"...I think a lot of the old guard (Dr. G especially) don't understand, or don't want to acknowledge, the amount of competition that they have now..."
They can't understand or acknowledge this because they lack the humility necessary for those actions. They've live in a delusional bubble that can't be popped. Sometimes folks have disordered personalities which prevent them from glimpsing reality.
"we were prohibited from selling 9.4 in FY2023 (maybe 2022 too"
I sincerely hope your words above are satire.
When a customer wants to buy something, and SAS refuses to sell it to them, and instead, SAS tries to sell them something else(which they did not want in the first place), don't be shocked when the customer walks away from SAS.
Selling the customer what they want seems to have become a concept of the past at SAS. That is just one more example why SAS will not reclaim greatness.
But hey, let's toss a bit of sarcasm: SAS can now honestly boast that Viya outsold 9.4 in FY2023. Even a pittance is greater than zero. Smh...
Former salesperson here- we were prohibited from selling 9.4 in FY2023 (maybe 2022 too, but I don't recall) in all cases unless there was no Viya path. Even looking at upgrades there were a lot of things that Viya couldn't do, that 9.4 did really well.
When looking to upgrade/migrate customers the LOE for the Viya migration was always very significant, and very expensive. Also, anytime we exposed someone to that path, they (rightfully) would look at alternatives given they were basically starting over anyway.
I think a lot of the old guard (Dr. G especially) don't understand, or don't want to acknowledge, the amount of competition that they have now.
Until SAS is following generally accepted accounting principles(that a public company must follow), the revenue numbers are mostly meaningless.
Gross revenue, even when fudged(we do not know that it is fudged) is a meaningless number. The only meaningful conclusion we can draw about the gross revenue of SAS is that it is not growing.
SAS has a slew of products that exist only because of financial corporate compassion. It is similar to a hoarder keeping a barn full of mostly useless junk. Does anyone know a hoarder who woke up one morning and said "enough of this useless junk"?
WRALTechWire says that their 2016-2020 numbers came from SAS Institute.
It’s possible of course to add numbers in different ways, for example to pull Q1 numbers into the previous Q4 to make a down year look better. Many companies do that sort of fudging.
But to become “IPO-ready”, SAS can’t do much funny stuff. To attract buyers, SAS needs to show profitability, and the SEC requires three years of audited financial statements. Questionable accounting won’t sell, so it won’t happen.
And it’s not necessary. An honest way to ensure profitability is through layoffs. The Majority Owner will do all he can to avoid a mass layoff. But SAS needs profitability, so I believe they are laying off just enough to achieve that.
Those seven years of “flat” revenue around $3B are not actually “flat”. Adjusted for inflation, revenue is down about 15%. Headcount is also down about 15%. Cause and effect.
Not sure where those numbers come from. None of the references you cite include the complete list. At least I didn't see it. Maybe you got them from somewhere else, or maybe I just didn't see it. My fault if I missed it.
It doesn't matter in any case because SAS is privately-held and doesn't have to report sales figures, revenue, expenses, or profit. There's no actual (reported) basis for any financial statements, nor does there need to be. Goodnight can take a loss, pump his own money into the company, call it profitable. He can drain savings into capex, depreciate assets, move money from capex to R&D and back again, do whatever he wants. It's literally his company. All the buckets of money are his money.
Wasn't it 2018 or so when SAS eked out profitability of 0.18% after some funny accounting with JMP revenue so it could report yet another year of consecutive profitability? And one of the reasons the IPO is taking so long is getting SAS's financials GAAP-compliant.
The annual report even states that the numbers that are the basis of any statements are reported on some non-GAAP basis:
"Metrics in this report include amounts and percentages determined on a management reporting basis, which may differ from GAAP reporting."
and who knows what that really means?
Particularly telling are statements like: "third year of double-digit SAS Cloud growth". That can mean:
2020: $5000
2021: $5500 (+10%)
2022: $6050 (+10%)
2023: $6655 (+10%)
so statements like that are meaningless in the absence of real numbers. Revenue of $6655 is not impressive, but the statement that there have been three consecutive years of double-digit growth is true and sure makes it sound like it's something.
Same for Viya "sales growth". What does that even mean? 2022 sales of $200 million, 2023 $246 million? Who cares? The company has spent billions on Viya. It is literally robbing Peter (SAS9) to pay Paul (Viya), and Viya revenue is not increasing fast enough to offset loss of SAS9 revenue. What's the ROI on Viya? Hundreds of millions on billions? Now that's a conversation starter.
When is the inflection point where Viya revenue replaces SAS9 revenue as the majority of SAS revenue? Another five years? Ten? Twenty? When SAS9 revenue is $800 million? The company will be a 1.6 billion dollar company then. 1.2 billion dollars? The company will be a 2.4 billion dollar company then. But I'll bet it will be nowhere near the 1.6 billion a 3.2 billion dollar company needs, and certainly nowhere near the "continued profitability" number without significant and painful cuts to reduce expenses.
The annual report doesn't contain a single mention of "revenue" or "expenses", and only mentions "profitability" twice, once in a quote from a customer. SAS really likes the word "growth", though.
So why do this at all? Because Goodnight is trying to stem the tide of defections from SAS's customers. He's trying to convince them that the company is still relevant, and profitable, and growing, which is exactly what a CEO should be doing.
I am really looking forward to the required disclosures on IPO and I'm betting that one of the reasons the IPO was delayed (yet again) is that the company just doesn't have the numbers to back its own public statements about profitability and growth.
My manager did not communicate on the remote work issue. As I said, a bad manager; but in fairness I believe SAS policy was not determined at that time. My new position was guaranteed remote, with a better manager and a better salary.
The problem with these pandemic stimulus jobs is that they exist in good times but may not exist in bad times. Many startups currently need venture capital, and can't get it, so are failing. Depending on the stage of one's career, even with all its problems, SAS may be a better place to work.
"SAS reacted slowly to the rise of remote work" -
" huh? Did your manager require return to office? HR certainly did not. We were told to work wherever we were most productive. I haven't worked a single full day in the office since March 12, 2020."
Prior to March 2020 (before the pandemic), remote work was not encouraged and only special circumstances were allowed and approvals must be obtained from managers and their managers, possibly all the way up to the division head.
Since March 2020 (Covid-19), remote work is allowed for folks to work from home and a year or so later SAS asked folks to work a hybrid (come in to the office at least 2-3 days a week) schedule but most people don't even bother coming in and SAS is letting them. They prefer to work from home all the time. Some jobs required onsite presence but most white-collar jobs can be worked remotely. SAS is quite nice and lenient in this regard, letting people to continue to work from home since the pandemic while some companies are forcing employees to come back in to the office again or be fired.
@3dbp+1pYqO6eY
"SAS reacted slowly to the rise of remote work" - huh? Did your manager require return to office? HR certainly did not. We were told to work wherever we were most productive. I haven't worked a single full day in the office since March 12, 2020.
In forums discussing tools for data science out on the internet. I saw some comments in quotes below that people made about SAS that are concerning to me especially the parts about you can't rerun an analysis years later and be certain of the same results in SAS.
" Our department got rid of SAS due to infrastructure issues and high costs. We converted all the SAS workflows into Python and saved a lot of money. "
" SAS is expensive and closed source. To my knowledge, people praise it for data management when you run out of memory. However, SAS has fallen behind in the data science game.
I've also been told that this means you can't rerun an analysis years later and be certain of the same results: it's closed source with versions that you don't own and which update yearly. One of my stats professors refused to use it for that reason.
Oh that's pretty damning. Inconsistency over time with anything but absolute cutting edge stuff is inexcusable. "
" I've been gone from the SAS world for almost 8 years now, and I really do not miss it. Their answer to everything was another expensive add-on module for stuff that you can trivially do for free with high-quality and well-supported open-source libraries in R and Python (specifically, I'm thinking of the time when they tried to sell us bare-bones natural language processing support for something like $60K a year, when NLTK is right there ) "
"SAS is old school and a little behind. It can do powerful data analysis, but it's not good for general programming. Honestly I wouldn't tell anyone to learn SAS at this point unless it was required of them. I'd stick with Python/R and SQL. "
SAS Institute Annual Sales:
2016: $3.2 billion
2017: $3.24 billion
2018: $3.27 billion
2019: $3.1 billion
2020: $3 billion
2021: $3.2 billion
2022: $3+ billion
Sources:
https://wraltechwire.com/2021/05/18/the-state-of-sas-after-5-years-of-flat-revenue-growth-may-be-returning-in-2021/
https://en.wikipedia.org/wiki/SAS_Institute
https://www.sas.com/content/dam/SAS/documents/corporate-collateral/annual-report/company-overview-annual-report.pdf
"simple : The version 9 revenue stream is declining faster than Viya is growing"
That sums it up nicely. Thank you.
Unfortunately, what you shared is an indication that more V9 customers are abandoning SAS instead of staying with SAS by doing a migration to Viya.
… Many of the 500 after-mentioned “hard-core” folks who departed, would have been the same people to, over time (by now) make Viya compatible with V9 under a much broader set of circumstances than it is today. That said, absolute backward compatibility was not an initial mandate when the CAS team formed. This has been discussed in detail on threads going back several months ago.
Political infighting, distrust and lack of respect among the R&D chieftains was a likely reason why said compatibility was not a mandate. Oliver had already successfully presided over technical development of several smaller “architectural extensions” to version 9 including In-DB, HPA and LASR — by operating a small team and steering clear of his much of R&D as he possibly could. His academic achievement/bearing, sheer physical size and command presents caught JG’s attention early on. While Oliver was still an analytics developer, he recognized significant mediocrity and laziness within many pockets of R&D. As CAS came to light and it was clear he would be in charge of that effort, my guess is Oliver did not want to be encumbered by having to deal with what would very likely have been sluggish R&B involvement in attempting to design backward compatibility into the initial effort — effort given SAS history with even lesser version migrations would likely be disastrous anyway.
A brilliant woman (no longer at SAS and who does not get enough credit) was also behind the design of CAS. She was in one of the very successful product groups and apparently chose not to leave to join Oliver’s initial team. This to say that there was input as to what several of the Fintech and ML related products needed to really scale next level.
Putting all of this and much more than it’s been shared on copious posts and other threads, it became clear that CAS should initially focus on building a elastic scalable, multithreaded/multiuser compute engine With a relatively simple action parameterization grammar and the ability to stream chunks of input data response structures between client users of the CAS server. The raw I/O horsepower SAS users had grown to love was designed into CAS early on.
I was part of that exodus in 2021. I had a bad manager, and SAS reacted slowly to the rise of remote work, and was not paying market rates. So they gave me multiple reasons to leave. I lost my commute and gained in salary, as did others.
Now we are on the backside of the pandemic stimulus. If lowering rates to 1% caused a bo-m, raising them to 5% must have an opposite effect. If layoffs total only a few percent, SAS may be again what it was in the past, a relatively safe port in an economic storm.
I don’t know why Viya was not designed to be SAS-compatible. Whenever I think about that, words like “bonehead” enter my mind. And people who post about SAS-compatibility in this forum describe it as a Hard Problem.
So Viya looks to me now like a salvage operation. It should be possible to position it as a financial services platform, for Fraud, Risk, Anti-Money-Laundering, etc. Even in a recession, banks will pay for products that save them money, especially if they have fast performance.
…. certainly quantity
@3gsp+1pYqO6eY …simple : The version 9 revenue stream is declining faster than Viya is growing to compensate due to customers consolidating to fewer machines/SAS licenses, downright canceling, moving to open source etc.
If the top ~200 Viya/ R&D and ~300 presales/tech support folks had not left between 2 and 4 years ago, Viya would likely be doing much better. Collectively those approximately 500 “hardcore” people were a force multiplier. Every one of them was doing 2-5X work in terms of innovation/quality and certainly quality of your average mediocre employee.
Current employees must be white knuckling it through each work day wondering which information is real and which is bs. Sales are flat, but Viya is landing huge deals?
Select experienced people are being cut and replaced with interns (the recipe for fealty over competence)?
Stingy Jim won't pay for competence and skill?
Must be terrifying to hang on there.
"Nearly all sales teams exclusively push Viya and its various solutions. 9.4 is only brought in in cases where Viya doesn’t make sense, usually due to budget or regulatory requirements. As mentioned in the previous post, the deals can be quite large."
The deals can be quite large. One would think "quite large" means moving the needle on revenue, yet revenue has been flat for a few years(maybe even down if adjusted for inflation). Please explain why "quite large" Viya deals are not moving revenue upwards.
“I’m confused by the notion that SAS is just giving Viya away to anyone who will listen. I’ve seen this in multiple threads over time.”
They don’t know what they are talking about
A consultant has said this about SAS on the internet:
"SAS used to be about the best tool available for large scale analytics. Now Python is effectively the leader and only choice for deploying the most modern analytic frameworks across dozens or hundreds of gpu or cpu instances on the cloud. SAS has neither the capacity nor the pricing model to support such use. While SAS can often run much faster than Python for raw I/O, that benefit is lost when each instance costs $15,000 or so to license, compared to nothing for Python. That is a big deal when working on the cloud with large deployments that need to scale.
When Google, Facebook, OpenAI, Microsoft, or others introduce new analytics capabilities it is almost completely unheard of for them to be implemented in anything other than Python. And absolutely never in SAS.
I am not biased against SAS. I have spent more than twenty years consulting in this area. And I have an affection for it. But in my software engineering firm we never find a need to use SAS; we use Python, and we work on the most advance machine learning models with it."
I don't know how SAS Viya is charged for the customers, if the license is based on each instances deployed and how expensive it is. If someone is actually in SAS Sales and has the knowledge, could you share and explain how it works?
Does the Viya platform perform operations using vectors as opposed to loops (e.g. R vs. Python)?
Because the personalities at SAS are still arguing over it.
I’m confused by the notion that SAS is just giving Viya away to anyone who will listen. I’ve seen this in multiple threads over time. The only free version I’m aware of is the 14-day trial which leads to a sales conversation. On desktop versions, there is a notion of a 30-day evaluation license for more in depth testing/POC, but again these only come after sales has been engaged and there is traction. Viya architecture doesn’t really lend itself to evaluations. So usually further engagement is paid through PSD for POC’s.
Viya may have originally been given away to key strategic clients, but I have no knowledge of this. At this point in time, it is 100% for sale only. Nearly all sales teams exclusively push Viya and its various solutions. 9.4 is only brought in in cases where Viya doesn’t make sense, usually due to budget or regulatory requirements. As mentioned in the previous post, the deals can be quite large. Sales comp incentivizes Viya deals, why would sales-people cut themselves short by not pushing it? The answer, they don’t.
Performance measurements/comparisons are going to be highly dependent upon specific use cases and scalability requirements. The biggest issue with deployed open source is that Customer then have to hire considerable talent to deploy, extend, make it work internally, write business specific logic, etc. — in many cases, even more so than for SAS Viya.
That does not invalidate open source and going in that direction if that’s what’s good for a company, but honestly in a lot of cases it’s the millennial generation “building their brand” which is often to get a bunch of open source analytic skills and then move companies every 2 to 3 years to jack up your salary. Not running down this strategy just pointing it out! OSS may be free For mini cases, but the talent necessary to make it work in the industry often is very costly.
Also important to point out that enterprise grade open source software, especially that runs in cloud deployments is not always “duty-free”. IOW, you can’t just download it and start using it in every possible environment and business case. The license may require royalties for specific extensions, libraries, etc. Sure, Python is free to download and program in, but that’s not the equivalent of having enterprise grade software that can simply be turned on and used by workers who don’t write code or plug complex components together.
I was part of a POC at one of SAS’ most strategic international customers where Viya/CAS ultimately won over open source in a multi-million $$ deal.
I saw someone (an engineer and data analyst) said this on internet
"Previously SAS was supposed very fast when dealing with large datasets. But currently open source packages like TensorFlow is much faster than SAS."
I am not familiar with open source packages so I can't say if it is true or not but if it is true, it's not good news for SAS.
"Pay them and they'll come!"
We're back to the same problem already discussed in other threads.
SAS won't pay them (the best talents) enough so they will not come or stay! You can't run a successful company with hopes of growth in modern times with mediocre people.
Seriously, does anyone really believe that SAS Viya is actually 30x faster than the competitors by this Futurum Group who was hired by SAS to run the test? If SAS is truly 30X faster, why hasn't SAS been able to capitalize on this potential? Customers are still rejecting Viya despite SAS practically giving it away for free to get them to try it.
"If it is truly 30X faster, or even 10X, the technology has potential. I hope that SAS can use that potential."
If that "potential" has to be FIRST recognized here on this site, that is a BIG problem.
If that "potential" is real, it should have been acted on several years ago. Yet revenue is still flat and has been for several years. That can only mean that SAS has a problem with the deployment of Viya potential - or the "potential" simply is just ONLY talk.
In the software industry, especially in the software industry, a window of opportunity must be acted upon with haste, otherwise someone else(or something else such as open source) will beat you to the punch. That is a fact and is way more true today than it was just 3 years ago.
The Viya Architecture has lots of possibilities. Build out enterprise grade solutions products attop Viya for the win SAS! To do this correctly you needs more high-end R&D and presales/tech support talent! Pay them and they'll come!
The poster said “healthy”, not “unhealthy”. But the word “disgruntled” can be offensive. It means angry, dissatisfied, with connotations of grumbling and sulking.
I don’t feel disgruntled. I was treated badly at SAS. But I’m not angry or dissatisfied, because I left — just like you would leave a bad relationship. And like a bad relationship, I still care about SAS. I spent most of my career there, and still have friends there.
Most of us are not disgruntled. We have a mix of feelings. The person who said we were “trauma bonded” got it right.
This thread has been most informative regarding Viya. Previously, I thought SAS should cancel it because of poor sales. But if it is truly 30X faster, or even 10X, the technology has potential. I hope that SAS can use that potential.
“Calling folks "disgruntled" is condescending and disingenuous. It invalidates the pain and anger they felt and experienced in their situations. Stating it is unhealthy to feel that way is a judgement call on your behalf, and smacks of more of the above. It denies our reality, and gaining a grasp on reality is part of the reason why many of us are here.”
Disgruntled is a word that is neither condescending or disingenuous. Do you honestly believe that a large chunk of the posts here don’t fit the word?
I was the one who first said disgruntled in this thread and nowhere did I say unhealthy and in some cases being disgruntled is probably warranted.
it doesn’t mean everything you guys post is correct and it clearly is driven by raw emotions in many cases
Calling folks "disgruntled" is condescending and disingenuous. It invalidates the pain and anger they felt and experienced in their situations. Stating it is unhealthy to feel that way is a judgement call on your behalf, and smacks of more of the above. It denies our reality, and gaining a grasp on reality is part of the reason why many of us are here.
The "Oh, I'm just here to give the facts. All my intentions are good, really." tone of some of these posts is really tiring. It's simply more condescension and invalidation. Give us all a break.
SAS has built/is building specific solutions that sit on top of Viya. The most successful ones to date seem to be in the finance space. Fraud, AML, Risk.
The point of both the OP and @1fbx+1pYqO6eY is not that Viya is a market success, but that it has market potential. A 30X performance improvement certainly would.
The OP asks whether SAS is trying to monetize this potential. Is SAS building business solutions on top of Viya?
"Do we know of any plans to push it more as capability that empowers business solutions as opposed to the research environment? "
Wow, sorry OP but you are seriously deluded. Viya is a failed product. One of the best measures of that is the demand for people with Viya skills. Just checked LinkedIn Jobs and found 7 Viya jobs in the entire US....Python 360,000, Spark 18,700....Viya is a joke, a laughing stock, a failed product that nobody wants. If JG wants to continue to flush $ down the toilet that's his perogative but it won't make the slightest bit of difference to Viya's prospects.
Disgruntled! Not simply.
This site shows SAS gross revenue at $3.05B for year 2014.
How has Viya changed that? Numbers usually tell a story of what the market wants - or doesn't want.
I understand past employees being upset about how they left or concerned if they left on their own volition. I also believe it’s healthy the be unhappy or disgruntled depending on each individuals situation. I also think on many of these threads, there are posters who are here just to stir things up not actually provide factual information. Are the past employees or just random trolls? Not sure. Regardless I only try to provide raw information. It’s up to everyone to make an informed decision about it.
30x is impressive, I’ll admit. But, I also would like to see further analysis that is truly independent. That would be a good litmus test. When I first saw the study, that was my main criticism.
Cost is a factor in some cases, but in others it doesn’t matter. If you compare the price of some competitors (true open-source excluded) the price isn’t that far out of range. A quick google of h2o.ai results in base pricing of 300k-500k annually with far less functionality.
Viya may have a premium price in some cases, but more robust functionality makes up the difference. In some cases technical ability limits the customer. Kubernetes is nice, but many business are still in Excel mode. They don’t have the IT staff to support more advanced deployments. Sure hosting can be provided in cases like that, but then the budget must be large enough to accommodate.