Friday, 24 September 2010

Larry's magic numbers - bite him in the butt

At Oracle Open World, aka #oow10 on Twitter, a lot happened. Fair to say that, whatever you think of it, at least it stirred up things. "Cloud in a box" (bringing back to me fond memories of my first AS400, and later my first ICL mainframe), putting down as not-Cloud, Larry sure got things going this week

A tweet on Oracle R&D from Michael Fauscette, however, was too bold:

#oow10 R&D budget over 6 yrs grew: IBM 6% total, SAP 60%, Oracle 192%

It was his tweet, but it's fair to assume those figures were copied from the Oracle-stage or -screen. By the way you might want to read this related post by Vinnie Mirchandani on the special treatment of analysts by Oracle
Frank Scavo contested the absolute truth of that saying that might be true in real dollars, but certainly not as relative part of revenue - being the figure fetisjist I am I couldn't resist scavenging through another few sets of annual reports and get the real stats. And the result is more than stunning actually, with Larry Ellison overplaying his hand here - by far

Update 24th September 17:50 CET:
In a previous blog post I pointed out the different make-up of these companies, let me repeat it here:
SAP: 23% services, 77% software
Oracle: 20% services, 80% software
IBM: 40% services, 40% software, 10% hardware, 10% financing

Oracle and SAP can be compared perfectly in this way, can't they?

First, as it turns out, these figures are indeed absolute dollars. Statistically, a comparison like this would only make sense if the companies can be compared. For that, it would mean that the relative growth of Oracle, SAP and IBM over the last 6 years was similar. Surprise, it's not - not even close
  • IBM revenue grew 2.3% from 2004 to 2009
  • SAP revenue grew 42.7% from 2004 to 2009
  • Oracle revenue grew 128.4% from 2004 to 2009
So, absolutely incomparable: IBM hardly moved, SAP stayed above inflation level, and Oracle more than doubled its revenue. It is absolutely useless for Larry to compare Oracle to SAP or IBM given these figures, or any other company for that matter because there are few companies that have such growth in such a small time. Sure, it's not organic growth but delivered by take-overs, but still - it's growth in capital letters

Second, it's highly usual to take R&D as a percentage of revenue, for comparison purposes. After all, over the years that's easy to compare within a company, and across companies. So, let's take the relative R&D percentage of each company over those 6 years, and to make matters simple, the average of that. Doing so, here are the figures, showing the percentage of total revenue that is spent on R&D:
  • IBM has an average R&D% of 6.3% over 2004 to 2009
  • SAP has an average R&D% of 13.9% over 2004 to 2009
  • Oracle has an average R&D% of 12.5% over 2004 to 2009
Interestingly, Oracle doesn't have the highest R&D% of all three companies! SAP wins here. That's a surprise (again). Let's take a closer look and see how that yearly percentage evolves over the years:
  • IBM has an R&D% increase of 0.0% over 2004 to 2009
  • SAP has an R&D% increase of 12.1% over 2004 to 2009
  • Oracle has an R&D% decrease of 5.7% over 2004 to 2009
Oracle has a relative decrease in R&D spending? IBM breaks even, and SAP increases it even? That's an entirely different view than pictured by Larry...

One final try here to give Larry the benefit of the doubt, which is getting increasingly harder, but hey: it might just be that R&D increased as a percentage of operating profit! That would justify these figures that so sharply contrast with Ellisons's message - wouldn't it? So, let's first look at the profit as a percentage of revenue:

  • IBM has an average profit% of 15.0% over 2004 to 2009
  • SAP has an average profit% of 25.7% over 2004 to 2009
  • Oracle has an average profit% of 34.8% over 2004 to 2009
Wow. That's impressive right there, Oracle makes way more money out of revenue than IBM or SAP. The bad news is, SAP and Oracle have a profit% decrease of 8.9% resp. 6.8% over 2004 to 2009, where IBM has an increase of 65.3%. Still, Oracle has a huge amount of money to spend on R&D, compared to IBM and SAP, so that would then explain that 192% mentioned above, wouldn't it?

No, once again. It's all one huge suggestimation

For the final blow: I will compare profit % to R&D %: after all it says a lot whether you're making 10% profit and spending 20% in R&D, or vice versa! In the figures below, the R&D percentage (of total revenue) is compared to the profit percentage (of total revenue)

  • IBM has an average R&D-to-profit ratio of 43% over 2004 to 2009
  • SAP has an average R&D-to-profit ratio of 54% over 2004 to 2009
  • Oracle has an average R&D-to-profit ratio of 36% over 2004 to 2009

Say what?! Oracle has the lowest relative investment in Research and Development of these three companies? 20% less than IBM, and 50% less than SAP?!
Now compare all that knowledge with the initial info:  "R&D budget over 6 yrs grew: IBM 6% total, SAP 60%, Oracle 192%"

Here's the entire overview of all figures for the real cracks out there:
Update 24th July 1:16 AM:  SAP figures are in euros, IBM and Oracle in $ of course. Comparing percentages only, so doesn't hurt

Larry Ellison, are you sure you wanted this 6-year comparison on R&D between IBM, SAP and Oracle?

9 reacties:

Djones said...

I dont share your opinion of the importance of % of revenue. What's important is the absolute $ spend, the breadth of product range on which its spent, and how much results in free-on-upgrade enhancements vs new products. Scale rules in software. A $10m software company can spend 20%, 50% on R&D but its not going to move its product forward as fast as a company 10, 100 times as big. Oracle customers benefit from $2.8bn of R&D investment whether they are paying $100k pa or $10m pa.

When you compare Oracle with SAP, for example, $2.8bn is better than $1.2bn, unless its spread across more products and more verticals, or if the result is new products rather than enhancements. For example, SAP's spend on BbyD doesnt benefit existing customers at all.

Martijn Linssen said...

Thanks Duncan! Glad you comment and disagree on my view of the #oow10 R&D-fest - impressive bio btw, on a side note I'd like to discuss (B2B) Integration with you some day

The (breadth of the) product range is important indeed, that shows how much attention you're giving your total range of products. Usually you favour some and some not, that's only to be expected
I also agree on the importance to distinguish between free-on-upgrade versus new products: that's largely about old versus new customers / markets

Sure it matters whether you're spending a buck or a billion on something, but it was Larry who started the comparison between IBM, SAP and Oracle - not me.
And if you're comparing companies, you should at least also take into account relative comparison: if that billion is less than 1% for company A whereas that buck is half the revenue of company B, company A is not investing in its future at all - period

So while I do share your view on importance of absolute comparison, your disregard of relative comparison puzzles me very highly. You are talking about moving products forward, Larry and I are talking R&D; a much wider scope

Then again, if you do compare absolutely, please also take into account IBM, which spends more than twice as much as Oracle on R&D

Frank Scavo said...

Hi Duncan, thanks for jumping into the discussion. I do understand your points, and agree to some extent. However, as you point out, it is important to consider how a software company's R&D spend is "spread across more products and more verticals." I think by that measure, in fact, Oracle's R&D spending also comes up short.

As you well know, Oracle has made an enormous number (70+?) acquisitions over the past six years. The customers of those acquisitions (e.g. JDE) are still paying the same or greater amount of maintenance dollars. But now, those maintenance dollars are not going for "free-on-upgrade" enhancements (your term) but largely for additional acquisitions (e.g. Sun, Agile, Demantra, etc.) most of do not result in "free on upgrade" enhancements for JDE.

Yes, I know, JDE customers are supposed to get "like-for-like" access to Fusion modules. We'll have to wait to see how much Fusion functionality winds up "free" to JDE customers and how much requires additional customer spending for database and middleware licenses, not to mention Fusion licenses.

If you've read my blog ( you'll know that I've been equally harsh on SAP. But in this case, I think it could be argued that SAP has made many fewer such acquisitions over the past six years. But again, I'm not here arguing that SAP is the desired model. Only that relative to Oracle, it can be argued that it is spending a greater amount on R&D.

Oracle's financial performance over the past several years has been outstanding. But it has attained this performance by acquiring software companies and their customers, squeezing out operational inefficieny (which there was plenty), and cutting R&D expenditures as a percentage of revenue. That frees up cash to make additional acquisitions, and the cycle repeats. The enormous scale of the operation also allows it to invest in Fusion middleware and Fusion Apps development. But, I argue, much or most of this does little to help the customers of those acquired companies.

David Dobrin said...

Just a question for all you number-crunchers. Are you taking into account the fact that much of IBM's business is in consulting services, where R&D spend is (and should be?) minimal. Factor out consulting spend, and IBM's percentages go up. Also, when talking about Oracle's growth, are you including Sun's revenue (and R&D spend) during those five years? I assume not. Finally, are you factoring in the differences between hardware R&D and software R&D? To compare SAP, IBM, and Oracle, the only thing that really should be compared is software R&D at each. To compare IBM and Oracle, the only thing that really should be compared is hardware plus software R&D at each. Naturally, I don't think Larry is doing this, but are you?

Martijn Linssen said...

Hi David, thanks for making a valid point. I updated the software-service ration as of your comment; a few posts back I figured out what these three do but that was rough, this one here is spot-on for 2009

All I'm doing is going through all the consecutive Annual and financial reports for IBM, SAP and Oracle from 2004 till 2009 (my wife's on holiday, great occassion really)

I'm not splitting R&D, it's total revenue, profit and R&D - but that would be a good idea for the *real* financial analysts out there to do (nudge nudge)

Please don't forget the level of reporting across these companies is very different. SAP states almost all their R&D is personell cost, Oracle doesn't give any hints whatsoever, and IBM spends half on software, other hals on "scientific research" - that's as far as I'll go

Frank, thank you very much for commenting of course - and discussing what the customer is getting back from Oracle's R&D; I admire your knowledge and information on Oracle!

duncanwjones said...

On the whole Oracle buys companies to expand sales from and investment in them. Legacy customers of Agile, Demantra etc benefit from increased $ being spent on those products, whether or not the revenue has gone up faster than the R&D. Thats why the absolute $ spent on that product is more significant than the % of its revenue. The diffusion across products and verticals is even more important. When I was at QAD I believed our customers got a better deal from our $30m or so R&D than customers of much larger companies, because ours was focused on 1 product line laser-targeted at their needs. I still believe that.

So I agree wholly with Frank that there is a danger that JDE, Psoft, Siebel, etc customers end up subsiding other product lines. They should judge that by the quality of their products' roadmaps, not just the raw numbers, and move to 3rd party support or SaaS alternatives if they are dissatisfied.

And Martijn, I apologise. I didnt mean to diminish your main point, that Larry's numbers were misleading.

Martijn Linssen said...

Thanks again Duncan, Larry Ellison using misleading R&D figures at #oow10 was my main point indeed - I have little deep knowledge of how ORCL or SAP or IBM operate - let alone my own Capgemini...

I now really get your point, and that you'd like to see the low-down in money per product invested and returned (I know I'd like to!) - then again I know that sometimes the bonus budget is used to pay expenses or so - never a clean science

Still, it will be next to impossible to make R&D and RO-R&D transparent across these three companies alone, I think - but I invite anyone to give it a shot! It will be very interesting, and I actually think it would be a good way for either of them to make good publicity for themselves

Vinnie Mirchandani said...

Martijn, nice one.

David, good point about IBM (or HP or other services heavy vendors. But within software, the disappointing thing is big fixes and re-platforming (to new devices/os or geographies) and regulatory releases make up bulk of R&D delivered. Not exactly innovation -

on the last I honestly would love to see governments recover royalties from these sw companies - they define new laws, payroll algorithms etc that vendors then code. I would love to see some of my tax money recovered :)

Mike Saxton said...

Does anyone know how much ORACLE spend on Fusion R&D vs Apps, Databases and Hardware?

Post a Comment

Thank you for sharing your thoughts! Copy your comment before signing in...