Wednesday, April 29, 2009

The Broken Promise of Software Maintenance Fees

Maintenance fees have come under fire in recent weeks as vendors like SAP and Oracle look to extract 22% annually from customers while providing a questionable return. The implied promise of maintenance fees is that you pay them each year to the vendor in order to fund the ongoing development and enhancement of the product you rely on. The promise is being broken. Badly.

There is plenty of anecdotal evidence of this broken promise, like Lawson's admission that maintenance fees are what is making it profitable right now and its assertion that this is ok because customers need the company to remain healthy, even if it is at their own expense. Considerable debate has occurred among bloggers and independent analysts regarding Oracle's ability to provide any innovation as a return on the billions in maintenance it is collecting. SAP announced earnings today and shared that in response to huge (and justifiable) customer outrage it is holding off on its planned maintenance price hike pending an analysis of the ROI customers are to get from the increased fees. Good luck with that.

In light of this I thought some hard numbers would be in order, and being a math geek I figured I could take a stab at providing them. It occurs to me that as leading vendors become more dependent on maintenance revenue, as it makes up a larger and larger percentage of their revenues, R&D spending should be increasing as well, right? They're getting this money to "maintain" the products. If they get more money they should be better able to "maintain" them. I know R&D as a percentage of total revenue is flat-lining or decreasing at these companies. So how about R&D as a percentage of maintenance revenues? I spent the better part of my morning working on this chart.

returnonmaintenance by you.

Does this surprise you? Here's a group of leading vendors who pretty much all are spending only about a quarter of customer maintenance fees back on R&D for the products they are said to be maintaining. Actually, it is probably worse than that. These are technology companies striving to innovate (well, except for Sage) so they'll be happy to tell you about the new stuff they're working on. Really, so after you spend on that new stuff how much less than one quarter of every dollar is going into the product I'm using?

I've never run a multi-million or multi-billion dollar software company, so I guess I don't know exactly how this is supposed to work. I know of course that R&D does not match maintenance revenues. I expect some profit to be taken. Maintenance is a pretty damn profitable business for vendors. Lawson had 83% gross margins on its maintenance business last quarter. I can also see the point that some of those dollars be used to fund acquisitions that fill functional gaps in my product. That's a way of maintaining and enhancing my product...buying the R&D efforts of others vs. spending it yourself. Acquisition binges the likes of which Oracle and Sage have been on, for all kinds of stuff that's unrelated to my product, are another matter.

Speaking of Sage, you may have noticed that the orange line looks a little bit different than the others. It's low all the way across the chart as being dirt cheap on R&D investments appears to be part of that company's mission statement. It's also lower because Sage does not report maintenance as a separate line item. So this is R&D as a percentage of everything that Sage does not classify as product revenue. In that sense you should pay more attention to its trajectory than its position relative to the other lines. The trouble with Sage is that it cannot make up its mind on how to report revenues. The jumps in the line are because Sage has this habit of changing the way it bundles and reports revenues. It is a shell game focused on obscuring an unrelenting decline in new licenses as a revenue source for Sage.

You might think this all leads to an argument in favor of Saas vendors like NetSuite or It's an argument Marc Benioff would like to make, but I'm not quite sold. They don't call it maintenance, but the recurring fees paid to Saas vendors have a support and maintenance element to them, and depending on your configuration and use of the software Saas can be more expensive than traditional on-premise software. Before you exclusively beat up the on-premise vendors for their R&D spend, look at the fraction of revenues that most Saas vendors put back into R&D.

In the case of highly profitable on-premise vendors like Oracle, Sage, or SAP customers ought to complain that too much of their maintenance fees are simply dropping to the vendor's bottom line. In the case of Saas vendors...instead of fattening the bottom line (there's precious little profit being made by these guys yet) you're funding ridiculously high sales and marketing spending. Pick your poison I guess.


  1. FYI, the chart is missing (both in Firefox and IE). Pls fix...really would like to see this!

  2. If the chart is accurate, this is pretty eye opening stuff.

  3. Jason - any feedback on Microsoft? I think I know the reason why they're not on the chart but they seem like they should be in the mix.

  4. Great post, very enlightening. It shows you how these companies are still operating very much as they did in the 90's (over-bloated) even though new sales are no longer really supporting it.

  5. Wayne - Microsoft is a hard one to chart because they just don't share any details. They tell us about the growth in something nebulous called "Dynamics customer billings", but no actual dollar figure (though I estimate MBS to be about a $1.2 billion business right now).

    I can tell you that if you go back and look what MBS might have looked like in 2000, just before this chart starts and Great Plains and Navision get acquired by Microsoft, the line would start out above all of these. Great Plains did not report a maintenance specific number, Navision did. You have to make some assumption on how much of Great Plains' "Service Revenue" line comes from maintenance. Suppose half of it was from maintenance. That would mean a combined GP/NAV would have spent 91% of maintenance on R&D during the last quarter in which both were independent.

    What happened after both were acquired? Microsoft deliberately lost money in the MBS group because it SIGNIFICANTLY increased R&D spending levels. I don't know where those levels have trended since I left the company just over three years ago, but while I was there I saw an R&D as a percentage of total revenue metric that was routinely 2 or 3 times what the nearest direct competitor was spending.

    Bottom line: I can’t tell you what the slope of the MBS line would be (probably slightly downward as well), but I’m highly confident it would be well above all of these vendors.

  6. This is a very interesting article but I think it misses some important points.

    First, each ERP vendor offers and includes different things under the term "Maintenance." At Epicor, the term maintenance is inclusive of all multi-channel support as well as upgrades and major releases to existing product versions and investments in next-generation technology and functionality for customers to move to when ready. It is this value-added combination that ensures an industry leading customer retention percentage rate in the mid 90's. It is also an important reason why Epicor gains hundreds of new customers every year.

    Second, it does not take into account the amount of ERP product lines each vendor develops and maintains, which is something customers need to and should understand if they are to be clear on investment scenarios.

    Third, it does not explain in enough detail that certain vendors' Partners (such as the Microsoft Dynamics Ecosystem) actually overlay an additional level of "maintenance" on top of what Microsoft levies for software upgrades. This is or should be common knowldege and can see overall maintenance rates in excess of 22%. Neither Epicor or our Partners do this.

    Also, I think the argument that Microsoft does not break out their figures is weak and bogus, and as the self-proclaimed Microsoft bigot that you claim you are, I must say it seems a little like "competitive marketing" to see an article contrasting the major ERP players but leaving off your favorite. Microsoft could easily make this information available, but have chosen (IMHO) to hide it deep in other areas, because it likely is not, as you suggest "well above the others."

    I believe that all customers and potential customers of Dynamics product lines (just like any other) should demand to know this of their Microsoft Reseller as Microsoft's investment in the Xbox has zero baring on their ERP business. That is, looking at the "Greater Microsoft" balance sheet provides no insight whatsoever into the investments made in R&D for the Dynamics product line. Remember, companies like Epicor are only enterprise business software providers, we do not produce operating systems, programming languages, detabases or entertainment products, and so all re-investment of revenue in R&D is focused on just that: the enterprise business software our customers rely on.

    I would also suggest that readers avail themselves of the Software Licensing and Pricing Bill of Rights as well, available from Forrester Research, which helps potential ERP buyers to understand what they should ask and know around Maintenance services.

  7. Very interesting chart! Out of interest, wouldn't have a very similar slope, assuming some arbitrary percentage for the "maintenance" part of the subscription fee? If so, isn't the downward trend at least partially due to economies of scale of R&D (a new feature is a new feature to me, no matter how many other people also use it) and steadily growing customer bases? One could argue that maintenance fees could/should go down over time because of this, but I don't really see that happening...

  8. Excellent point Timo. In that way the R&D going into new projects like Oracle's Fusion are significant to the JDE or PeopleSoft customer. Oracle should be gaining economies of scale in development of those legacy apps. In theory the "extra" resource there goes toward something that benefits the customer in the future.

    That or, as you say, it's reasonable to think that maintenance fees should go down over time, either for the benefit of the customer, or because the vendor has the financial means to use lower maintenance as a competitive advantage.

  9. jnorwood - I appreciate that "maintenance" encompasses many things, and the ones you listed I suspect every vendor on the chart would argue they are doing. They can all make that case (except Sage with respect to developing new next generation technologies). That's all fine, but it doesn't nothing to explain the downward trend in the line.

    You also raise a fair point regarding the number of product lines a vendor is maintaining. But here again I see very little difference between the vendors included in the chart. To Epicor's credit it is making some efforts here by putting a .NET/Visual studio based wrapper around the disparate technology set that makes up Epicor 9.

    Now, regarding this being "competitive marketing", I did not set out to create a sales tool for anybody. A number of bloggers that I follow were having a conversation about maintenance fees and I realized that from tracking these companies as long as I have I might have something useful/interesting to add to the conversation. Is it useful to somebody selling Dynamics? Marginally. As you point out, the immediate question is where is the Microsoft line.

    I can tell you where I know it would have started on this chart based on the public reporting of Great plains and Navision prior to Microsoft acquiring them. The Return on Maintenance for Dynamics at the end of 2000 and entering this chart would have conservatively been in the 70-90% range, or about twice where Epicor and the other vendors start. From there it would have gone right off the chart as Microsoft significantly increased R&D spending around these products post acquisition.

    xBox aside, the R&D investments in the Dynamics product lines dwarfs that which Epicor puts into its multiple product lines. I don't leave Dynamics off of this chart by choice, but because Microsoft frustratingly doesn't share the details I'd need to plot it out.

  10. Thank you for the interesting chart and observations. I would be interested to add a SaaS provider there as well, so that we could compare the differences in the business models of SaaS versus on-premise solutions.

    I wouldn't straight off go and say that salesforce has a similar curve, actually I'm quite certain it wouldn't. Salesforce's Revenue from subscription and support grew faster (44%) than cost for subscription and support (39%) last year. R&D in its turn, grew faster than revenues for subscription and support (56%).

    Therefore, I would estimate that the R&D spending compared to support revenue for salesforce has increased. Right? The difference might come from the fact that salesforce only has exactly one software to support and maintain. I would rather see this fact as something that contributes to costs than something that should be taken into account in the calculations.

  11. The tough thing on is figuring out how much of that Subscription & Support line to attribute to "maintenance". If you just plot SFDC's R&D against the total Subscription & Support, the line is trending ever so slightly upward right now, but is basically flat. Here are the percentages over the last 10 quarters.

    10%, 10%, 9%, 10%, 9%, 9%, 10%, 10%, 11%, 11%

    That said, a couple of questions/observations...
    First, are SFDC's customer add rates accelerating or decelerating?
    Second, how does the number of net new customer adds during a given year compare to the total base? If you divided each year's new customer add number by the total customer base, that is naturally a smaller percentage increase each year, right?

    That second point may seem a fairly obvious one, but if that is coupled with customer adds that are decelerating....

    1 - Each quarter I'd attribute a larger and larger percentage of the Subscription & Support line to the "maintenance" of existing customers
    2 - R&D as a percentage of that increasing maintenance would have a downward slope like the on-premise vendors.

    I've looked at my data on NetSuite and would conclude the same. I think this supports the point that Timo Elliot makes above. Some portion of this is just attributable to economies of scale. Even so you have to wonder about the vendors that are sloping downward faster than the others? Are they that much more efficient with their R&D or are they gouging the customers more? And secondly, for all the vendors, why doesn't the maintenance rate go down over time?