Cloud / SaaS Total Cost of Ownership Framework

A very common question in the world of Cloud / SaaS applications is "How do I make an apples to apples comparison of the total cost of ownership of Cloud / SaaS applications vs. their on-premises software counterparts?" At some point, I hope this will become "settled law" - but until then I think it's important to revisit and update with details from real-world implementations.

Two of the best documents on the Internet I've seen on Cloud / SaaS ROI and TCO come from Ray Wang at Forrester Research and from the Software and Information Industry Association.

You may also recall the recent Nucleus Research case study on an Intacct implementation at nGenera that showed $715,000 in savings, 589% ROI and payback in just two months - below I'll provide some color on where nGenera and Nucleus found such an impressive return by adopting cloud financials.

One trend I'm seeing is that as cloud / SaaS applications get more and more popular, the on-premises software vendors are more aggressively pushing FUD and TCO models that only compare the initial purchase price of on-premises hardware and software to the ongoing subscription price of Cloud / SaaS counterparts. Apples to apples TCO / ROI comparisons must also include ongoing human costs for operating and maintaining the system over 7 to 10 years, which is the typical lifecycle of a major application investment. You can be quite sure that you will find the majority of Cloud / SaaS TCO and ROI in these ongoing operating cost savings, which is why the on-premises software vendors try to exclude them.

Below is my attempt at a framework for making these types of TCO and ROI comparisons, with demonstrative examples from Intacct and the financial applications marketplace.

Capital savings: Software and Hardware

  • No need to purchase Infrastructure Stack Software - By definition, Cloud / SaaS applications disintermediate technology infrastructure - Eliminating the need for software purchases like Client Access Licenses, Windows Servers, Application and Database Servers, Middleware, SharePoint, Citrix servers and VPN’s for remote access. Also remember most buyers will need two copies of each of these pieces of infrastructure software, one for development/test and another for production. With Intacct all of these pieces are included in the monthly subscription fee.
  • No expenditures on hardware – Also by definition, Cloud / SaaS applications require no server hardware, network storage, backup systems, disaster recovery systems, power or cooling systems, data centers, utility costs, etc. As above, most buyers will need two sets of systems for development / test and production. With Intacct, our applications run in Tier One IBM data centers, with disaster recovery provided via Sungard and Intacct foots the bill. All you need is Internet access.
  • Reduced need to purchase third party tools, modules and applications - A less obvious place to find ROI / TCO is that many SaaS applications include built-in or bundled functionality that eliminates the need for separate, add-on software tools, modules and applications. This is particularly true for new SaaS applications vs. older on-premises applications. For example, Intacct includes built-in reporting, business intelligence, dashboards, portals, workflow, employee expenses, multi-currency, CRM integration, financial consolidation, customization services, web services and much more, each of which are usually add-on or third party tools, modules and applications in the traditional on-premises software financial applications world.
  • No software or hardware replacement needed in ongoing years - A common mistake I see in TCO models is that only initial purchases are considered for on-premises software and hardware over a multi-year TCO lifecycle. More accurate models should break the TCO lifecycle into three section - initial deployment in year 0, ongoing operation and maintenance ongoing, and upgrade / replacement costs for hardware and software not covered under maintenance somewhere in between year 3 and year 5. Requiring periodic upgrades and purchases of new software on top of maintenance is part of the business model of most on-premises software companies and should be included in any TCO / ROI calculation. Servers, storage and other hardware components wear out and need to be replaced, typically on a three or four year schedule.
IT operations savings
  • Reduced / eliminated spending on IT operations - This is usually one of the top sources of ROI / TCO savings in the Cloud / SaaS world. One of the most common mistakes I see in TCO models is not taking into account the staffing costs in the IT department for deploying, operating and maintaining applications and underlying infrastructure. In Cloud / SaaS applications, the vendor takes on all of the costs associated with installing, running and maintaining the applications, the underlying software infrastructure and the associated hardware. For Intacct clients, this represents saving at least one full time IT professional ongoing and can be much more for larger deployments. As an aside - this doesn't always mean eliminating jobs in the IT department - it can also be seen as removing unnecessary, low value added work from IT, which allows the IT team focus on more strategic, value added services.
  • Reduced support costs - The cost to the customer organization, both in downtime and in human costs of diagnosing and fixing problems is a major ongoing part of software TCO. A major improvement in the SaaS / Cloud world is that through the magic of the Internet both the support people at the vendor or VAR and the client can be logged into the system and looking at the same screen / data / problem at the same time. The customer doesn't have to describe the problem to the support rep anymore - the both can look at it at the same time from the comfort of their own web browser. The ongoing support savings are significant.
  • No Application and Infrastructure Re-implementation costs in years 3 to 5 - Most TCO models miss the need in the on-premises software world to re-implement and upgrade to new versions of either the applications or the underlying infrastructure stack. This is unnecessary in the world of Cloud / SaaS applications like Intacct since the vendor continuously keeps the client on the most current versions.
Departmental / Business operations savings
  • Savings through improved application functionality, process improvement, automation and integration - Cloud and SaaS applications are by definition far more modern than the older on-premises software applications they are replacing - so an important part of TCO is to look at departmental effectiveness before and after implementing new applications. For example, Intacct’s order to cash functionality is particularly efficient, helps our clients reduce or eliminate the need for the order management function in the business. Similarly Intacct eliminates many manual processes and spreadsheets, allowing the finance team to do far more with fewer people. Intacct clients typically identify savings of one or two headcounts in finance as part of their deployment.
  • Savings through higher worker productivity - Cloud and SaaS applications typically offer higher worker productivity because of the always-on, always connected aspects of cloud computing. Because workers can access the system anytime and from anywhere, they typically get more done than if they can only use the system in the office during normal business hours. For example Intacct clients can get their work done anytime and from anywhere, go paperless, and extend web-based portals, dashboards, reports and employee expenses to all employees in all locations – eliminating wasted time and paper processes - and extending productivity improvements beyond the finance department and across the business. These productivity improvements quickly add up, and in an average Intacct implementation can free up one or more full time equivalents across the business.
Additional areas in a complete TCO / ROI model
  • Reduced Risk / Project Failure - Advanced TCO models should reflect the higher risk of project failure when deploying on-premises software applications. The hardware and software that must be procured, install, configure, optimized, integrated and tested in the on-premises world carries a very non-trivial risk of failure - thus on-premises ERP implementations have a widely documented and much higher rate of project failure than SaaS / cloud counterparts. You can account for this in a TCO model by adding in an extra cost line to your model - a "project risk reserve." Based on my experience, I'd suggest a risk reserve of 25% of the project implementation costs for financial applications. Intacct has a 98% project success rate - contrast this with what you find when you Google "ERP failure rate" - you'll see failure rates from 40% to 60% for on-premises software implementations.
  • Faster Time to Value - With SaaS / Cloud applications, there is no hardware or software to procure, install, configure, optimize or test. In your implementation timeline, you skip all of these steps - which offers both cost and time savings. Since time is money, your TCO model should include the acceleration of value you get with SaaS / cloud applications - because you will begin using the new, higher value system much sooner than you could in the on-premises software world. In the Intacct world, implementation times are typically 90 to 180 days faster than for on-premises alternatives - the larger and more complex the implementation the bigger the difference.
  • Easier to Adapt - It is much easier and faster to modify functionality and business processes in SaaS / Cloud applications than in on-premises counterparts. This is because SaaS / Cloud applications are designed to be configured by line of business people, whereas making changes to on-premises software tends to require programming and customization by IT professionals. Because of the ease of configuration, businesses that deploy SaaS / cloud applications make relatively frequent, iterative changes to evolve their processes to match changing market or business conditions - which is rarely possible to do in the on-premises software world. The business benefit to reflect in a TCO model is in the savings or increased revenue you expect from improved operations that result from being able to adapt more quickly to changing business conditions.
  • Superior Operations - For all but very large enterprises, the SaaS / Cloud vendors will include substantially better operating capabilities than the internal IT department is likely to be able to provide cost-effectively as part of the service. This means with SaaS / Cloud applications you get higher availability, better performance, faster problem resolution, better backup and recovery, fewer issues with maintenance, 24x7x365 operations, less chance of data loss etc. than you would get if you were running the applications yourself in your own IT department. A thorough TCO analysis would include these superior operating characteristics as an added expense to on-premises alternatives, or as an added benefit on the SaaS / Cloud side of the equation. Intacct applications run in tier one IBM data centers, with disaster recovery via SunGard, and with extensive service level guarantees that are far superior to what most internal IT departments are able to cost effectively provide.
  • Other areas - To be complete, there are other areas that a complete TCO / ROI model should cover, but my experience is that they come out about the same whether you choose to deploy SaaS / Cloud vs on-premises software.These areas include professional services costs for the initial configuration and roll out of the applications, ongoing changes to the applications to adapt to new business processes and training for both administrators and end-users. You'll likely identify additional areas in your complete TCO / ROI analysis where things come out about the same - my attempt in this post was to highlight the differences that come up over and over again when comparing Cloud / SaaS applications with on-premises software.
The bottom line

The bottom line of this post is that an overall TCO analysis that makes true apples to apples comparison of SaaS / Cloud vs on-premises applications should include a thorough analysis of costs and return across at least three components - Capital expenditure, IT operations and Departmental /Business operations. The TCO model should extend over the total expected lifecycle of the applications - I'd suggest 7 to 9 years in the world of financial applications - and break the overall lifecycle into three sections - initial deployment, ongoing operations and upgrade/replacement in year 3-5.

You may of course choose to include or to not include each of the above categories as you do your own analysis and you may choose your own lifecycle for how long you expect to leverage your new systems. What I think you'll find though is the above framework should be helpful in making a more apples to apples comparison that draws out the true costs and opportunities whether you are looking at Cloud / SaaS or on-premises deployments.


Jan Zawadzki (Cloudbreak) said...

Hi Daniel - this is a great post - added it to our standard CIO reading materials list!

I would add another item to your list: time to market. In many organisations the driver to implement a cloud solution comes from the business, and the difference in implementation timeframes between traditional and cloud solution can have massive financial implications.

Daniel Druker said...

Thanks Jan - I added new sections on time to value and on operations based on your feedback. I also added a new bullet on TCO improvements from improved adaptability based on feedback from Phil Wainewright here

Subraya Mallya said...

This is an excellent comprehensive comparison of TCO of SaaS v On-Premise. I did a similar comparison to help our sales to combat some of these questions.

We need to keep highlighting the virtues so as a industry we move to a more risk-equitable model (i.e, SaaS)

Smartupz Editor said...


Great post and good summary of SaaS TCO.
The only question remains - is lower cost really enough to drive people to adopt SaaS? I can not think of Linux vs. Windows history...
If someone is interested - I've posted something about it here:

Song said...


Very in-depth analysis of the TCO and ROI issues around SaaS solutions. I just posted a blog on the topic that SaaS vendors should use SaaS software to run their own business. One of the issues involved is whether SaaS is in fact cost-effective over the long run.

I linked my article to yours so my readers can perform the cost analysis for themselves. Thanks.


Fred said...

Very comprehensive assessment Dan. I have found that not only do customers bear the cost of IT operations but they frequently cannot meet the same uptime and and SLA's that a SaaS offering can provide them.

Rob Brown said...

This is great stuff, would you mind if I quote you or make reference to this entry in a public venue?

Anonymous said...

Hi Dan,

This is a great article.

Have you or has anyone looked at the "real" impact of cloud computing from A/cing perspective ? Are there any benefits when the transactions hit G/L entries from OpEx perspective ? Thanks in advance,

Ramkumar said...

Hi, Good analysis. Is there a similar analysis possible on the TCO of building and maintaining a SaaS application (from the perspective of the ISV or the IT department of a multinational enterprise). There are so many alternatives of building a SaaS application - such as using some PaaS like Long Jump, Force or Apprenda, going to a SaaS development partner who bundles a rapid development framework along with their service, or buying a SaaS Framework such as cellosaas off the shelf and building the business features on top of it.

Anonymous said...

Nice article. I have few question for you that I think I should have include if you want to calculate true TCO and reason for SAAS adoption.
How SaaS will address customization that most of companies have? How SaaS ERP will be linked to other company applications? Is there certain threshold when this is not practical because of volume of data that need to be send back and forward in certain time window.

Post a Comment