The Cost of Ownership
We wrote about freeware firms such as YouSendIt, DropBox and others, and clearly end-users are embracing freeware. Times are leaner and the genie is out of the bottle regarding technology costs. It looks like easier times for end-users. For example, IBM recently launched a direct attack on Oracle’s database, hardware, and big data technology and pricing with a ‘have it your way’ pricing scheme for both mid-size and large companies. Some of the options they have are try-before-you-buy free pilots, or payment terms that don’t kick in till 2013, as well as lower item pricing compared to Oracle.1 So even big players are getting into the act with creative pricing.
In Part One we talked about technology cost and pricing structures. But there is more to technology acquisition than just the pure cost of the raw technology (software/hardware). The cost of implementation — the project costs (integration, user adoption and training) — plus ongoing costs all add to the Total Cost of Ownership.
Let’s compare some project environments to understand the pre- and post-implementation costs. (You can see this comparison in Figure 4). In the last few months we have talked to many ERP end-users, especially the SMBs.2 These companies have unique, yet highly versatile, environments in which resources are thin, so it is easier to glean the cost drivers and actual costs for technology from their stories.
There have been many estimates of the ratio between services and software implementation in the on-premise world.3 Most of those were estimates given by the provider companies. However, there is also some decent, fairly accurate research about costs that has been collected from the implementation experiences of end-user companies. These are also cost estimates and are a somewhat incomplete reflection of the Total Cost of Ownership (TCO), since most companies don’t compile their internal costs of ownership.
The main issue though, is understanding the costs so users can make a cost comparison in a specific bidding situation. Certain approaches drive additional costs — or eliminate them. So, we asked end-user companies about pre- and post-implementation experiences. Some patterns emerged which are shown in the table below.
In aggregate, three issues stood out for these companies:
- Cost — Chunking out the initial software purchase into monthly payments over a 4 or 5 year period is somewhat equal to the outright purchase price. But here was the economic factor that tipped the scale for the SaaS — the maintenance fees of supporting your licensed software: year-by-year support fees of 16% to a whopping 24%. (Sometimes, upgrades to the big new releases of the licensed code may not be included from some vendors). You do that math. This counters the argument of “with SaaS, what have I got to show for leasing the software for four years?” By paying four years of support fees, you have paid the initial price of the licensed software again. So by using SaaS over a 6 to 8 year period (as long as anybody wants to think), the price is pretty close.
- People time — time learning, time implementing, time supporting, and especially time upgrading. It seems that these projects are never done! As you can see in the chart, though users are looking forward to the new capabilities in a release, they often decide to delay the upgrade due to a competing need for resources. Over time, these delays can necessitate doing mega upgrade projects that top the scales in cost and disruption.
- Risk — fear and risk are big issues with big systems like ERP. These risks are reduced by training, of course. But that adds to the time and disruption issues.
A Comparison of SaaS ERP vs. On Premise ERP Purchases: Two Cases
SaaS Single Code-base
Core ERP Purchase
Perpetual license fee
2 year agreement/monthly payment
Hardware, Data Center
Other Software Costs
MFT, Middleware, Database and Business Intelligence (BI) additionally purchased
0 (all modules included)
4 systems to integrate from the purchase, plus in-house integration to existing systems
In-house integration to existing systems
IT support personnel: in-depth training required, from several classes to 1 week
IT support personnel: 1 day training
Release and Upgrade
Vendor Upgrade: Various schedules that need to be coordinated between each product. Users decide to do upgrade once a year to coordinate the components from each provider.
Quiet upgrade option: impact analysis on internal changes. Decision to upgrade every 6 months. It took 1 programmer 2 weeks plus 1 day per user training for changes.
User-driven upgrades: compliance changes requiring changes to software. 6 week programmer and user implementation.
User-driven upgrades: compliance changes requiring changes to software. SaaS provider upgrades platform for customers using the same industry standard.
Note: Typically users have an option for ‘plug-ins’/integration of some of their own code. And in most situations across both SaaS and On Premise, end-users have to get these requests prioritized and scheduled by the providers.
Annual Maintenance Fee of 18%-25% of Initial License Fee
Monthly or Quarterly SaaS Fee. Varies tremendously, but is often sized so the end-user pays about the same amount over a ~3 year period as the total cost (SW, HW, data center) would have been over the same period with an On Premise system.
Source: ERP for the SMB Case Studies – ChainLink Research
Beyond these basic price drivers, there were growth considerations. Growth presents challenging issues in both the on-premise and SaaS world. SaaS pricing, as you have read, is not very straightforward; often, predictions about the size of data sets, number of users, or connections play into the price (and the contract). With on premise, you monitor your own growth and then turn to your technology partners — hardware and software — for your expansion. (Those firms may have already sold you discounted per-seat pricing into which you thought you might expand; but these days many firms shy away from doing that since they are carrying a cost and maintenance burden on something they are not actually using.) So growth is not a slam dunk in either case. These are additional costs that were incurred over time in the above cases.
But What about the “Per”s?
When the On-Demand revolution hit ~2006, solution providers were excited about this new way to gain access to the enterprise market by reducing the financial objections to big software purchases. And it has worked for some, such as Salesforce.com.4 However, as we noted in Cloud Economics, risk is transferred from the end-user to the technology provider. OK, that’s fair.
However, too much risk on the part of the technology provider is a ‘going out of business strategy.’ So these newbies have learned that some risk sharing is in order. Setup fees have been added in the last two years from some on-demand providers’ pricing models: Costs for setup can range from zero to hundreds of thousands of dollars. So, it is only fair that this one-time fee be covered by the user. This may bind the user to the solution provider. It means users have to do their homework upfront, since this is financial commitment — not just some freeware experiment. After all, a tech company is just that — a company — not an experiment. There are R & D costs, patent filing fees, payroll, and other business expenses, and the company has to recoup its expenses.
Post script: We will be writing about RFID and serialization and other item-level topics over the next few months and we will return to on demand and the per-connection, per-scan pricing issues in those articles. ERP for the SMB 2012 as well as Cloud WMS 2012 will also come out this spring and pricing and costs will be discussed for these, as well. (You can click here to read ERP for the SMB 2011 for a discussion of cloud vs. on-premise providers.)
On Demand Now – an interesting look at the launch of the On-Demand space
3 SMBs are generally easier to get a cost estimate from. Large companies have multiple instances, Business Intelligence (BI)/reporting appendages, multiple levels of integration, often hundreds of APIs to support, as well as distributed data centers. They have more ‘hidden’ costs in user disruption, training time, etc. — Return to article text above
4 Ratios vary by product, of course, from 1:1 to 12:1, service to software. — Return to article text above
To view other articles from this issue of the brief, click here.