Question On BI Total Cost Of Ownership

I need your help. I am conducting research into business intelligence (BI) software prices: averages, differences between license and subscription deals, differences between small and large vendor offerings, etc. In order to help our clients look beyond just the software pricese and consider the fully loaded total cost of ownership, I also want to throw in service and hardware costs (I already have data on annual maintenance and initial training costs). I’ve been in this market long enough to understand that the only correct answer is “It depends” — on the levels of data complexity, data cleanliness, use cases, and many other factors. But, if I could pin you down to a ballpark formula for budgeting and estimation purposes, what would that be? Here are my initial thoughts — based on experience, other relevant research, etc.

  • Initial hardware as a percentage of software cost = 33% to 50%
  • Ongoing hardware maintenance = 20% of the initial hardware cost
  • Initial design, build, implementation of services. Our rule of thumb has always been 300% to 700%, but that obviously varies by deal sizes. So here’s what I came up with:
    • Less than $100,000 in software = 100% in services
    • $100,000 to $500,000 in software = 300% in services
    • $500,000 to $2 million in software = 200% in services
    • $2 million to $10 million in software = 50% in services
    • More than $10 million in software = 25% in services
  • Then 20% of the initial software cost for ongoing maintenance, enhancements, and support

Thoughts? Again, I am  not looking for “it depends” answers, but rather for some numbers and ranges based on your experience.

Comments

BI TCO: using Boris's Approach

Hi Boris:

I always follow your very valuable posts and if I can be useful, I am trying - as you will see below. As you mentioned yourself, absolut numbers depends on many factors, but average %% between different pieces can be (subjectively and in average) estimated. Below are my estimates, using your decomposition of BI into parts. Before I will start. I have to repeat myself and say that BI as we know it is dead and used here only as marketing umbrella for bunch of technologies, including Data Collection, ETL, DW, Data Visualization, Visual and Predictive Analytics, pre-defined reporting, Alerting, Alarming, KPIs, Scorecards, Dashboards etc. So here you go:

1. Initial hardware as % of software cost: 20% for servers, plus some number of client devices (tablets, PCs) which is proportional to number of users.
2. Ongoing hardware maintenance = 30% of the initial hardware cost, because novadays hardware will be obsolete in 3 years.

3. Initial design, build, implement services. This is tricky one: if you using obsolete junk from Oracle, IBM, SAP and SAS you can reach 700% you mentioned. For Microstrategy it is closer to 400%, For Spotfire probably 300%, for Tableau about 200% and for Qlikview even less then that)

with your approach with cost of software I will slightly modify your numbers:
<$100K software = 200% in services
$100K-$500K = 300% in services
$500K-$2M = 150% in services
$2M-$10M = 60% in services
>$10M = 30% in services

Then, 12% of the initial software cost for ongoing maintenance, enhancements and support (20% seems to high, but vendors will attempt to get it)

Andrei
http://apandre.wordpress.com

Very valuable thoughts! Thank

Very valuable thoughts! Thank you!

Size matters, but not much

Boris,

My first impression was with several hundred billion per year invested in enterprise software, surely these questions have been answered on a continual basis, but that thought was quickly erased by remembering what industry we are in-- consulting costs would be lower if these relationships were better understood.

My second impression was that there is no alternative route around the "it depends" issue, although that issue alone I suspect is leveraged for a few tens of millions in consulting services in large projects that could and should be reduced to a fraction, and more automated.

A more uncompressed version of my tweet response:
(relating to services directly caused by software purchase)

1) Organizational type is more important than deal size

2) Product design is more important than organizational type

3) The vendor business model is more important than product design

4) Cloud versus on-premise is most important

Again, all relating to consulting services from software purchases.

I think Andrei's estimates are probably generally accurate on the vendors -- rings true for me, and while the math certainly is in disagreement on the claim that BI is dead-- if so it has a generous growing trust fund for heirs -- I certainly do agree that it's a catch-all phrase that often does do more harm than good for customers who should be focused on value, including from convergence of software industry acronyms. They will undoubtedly spend much less on both software and consulting services if they do.

Let's not forget that buyer beware rules in BI just as it does in most markets, and therefore customer sophistication and internal experience/talent/knowledge is likely the most determining factor (if you don't see a fool sitting around the table.... look in a mirror--that includes enterprise software architects by the way). It's also impossible to relate consulting services to software price tag without drilling down on data silos and integration costs -- still a huge variant there as evidenced by the U.S. Army SAP project among others.

As demand partially suggests (sales tactics notwithstanding), there is a lot of value in technologies related to BI, and as we see in the vendor list mentioned above quite a lot of differentiation in how to structure, collect, analyze, and present data for decision making. In the vast majority of cases I've studied, particularly in crisis prevention, it only requires one accurate decision to pay for the entire investment many times over -- it would be irresponsible to forget that rather dominant fact.

That said, from my perspective as someone looking for opportunity to deliver more value at lower costs in the market, even with completely new designs (perhaps especially given designs intended to lower costs and enhance value), I don't see any reason why consulting and maintenance costs should not be reduced significantly -- customers are paying for a great deal of unnecessary overhead and old business models, not to mention large numbers of mid-managers, enormous marketing budgets, and almost always at higher margins than they see in their own industries.

One negative about headlines in hoarding cash and success in IT is that everyone on the planet is aware that they are paying too much for IT, obviously. Much of this overhead is invested not on product improvement, but rather sustaining outdated business models and high maintenance fees. CEOs and boards should be reviewing their own internal influence on that issue as I am seeing otherwise mature adults complaining for years about high ongoing fees, while simultaneously requesting more hand-holding by very expensive suits living pretty well from client revenue. No question the changing global landscape from wealth transference from one industry to another is a key driver in more proactive roles in IT investment by business oriented CIOs as well as CFOs, CEOs and boards.

As of today with a key patent issuance date of later this month and head down on designing a new ecosystem, one for one (software sales to service) looks like a sustainable, mutually beneficial balance with better alignment, and maintenance costs on the south side of 10%, but that's based on the assumption that maintaining 1960s era legacy systems are not included, employing far more intelligently structured data based on standards, and assuming a slightly faster adoption to the cloud and mobile than incumbents.

Mark Montgomery
Founder & CEO
Kyield

Hi Boris, As you know, one of

Hi Boris,

As you know, one of the strong points of our offering is the low TOC of our platform Quiterian DDWeb: light hardware, quick set up and deployment, quick training, no need to create models, cubes or metadata, etc. As vendors we are and based in our direct sales experience (since 2004) these are the percentages:

- Initial hardware as % of software cost = 5% to 15%
- Ongoing hardware maintenance = 15% of the initial hardware cost

- <$100K software = 90% in services
- $100K-$500K = 70% in services
- $500K-$2M= 40% in services

- 15% to 20% of the initial software cost for ongoing maintenance, enhancements and support

We think that the the market trends are changing in the line of:

- Lower hardware costs based in the intensive use of new technologies (in memory, columnar, hybrid, cloud, web applications, etc.)
- Lower sofwtare costs based in new contract models, ex: flat rates, SaaS, pay per use...
- Lower costs of services due the irruption of agile and selfservice applications for the business users that will reduct the IT and process workload.

Hope this information will be useful for your research.

Regards,

Jacinto Barrio
http://www.linkedin.com/pub/jacinto-barrio/4/133/64

Where are Services Dollars Spent?

Hi Boris,

With such high Services numbers, an interesting question to ask is where these dollars are being spent. Is a high percentage of this spend happening in the "data" stage? In other words, building a data warehouse and complex metadata layer. If this is true, I would expect to see quite a divergence between vendors depending on their requirements in this area.

For Actuate, the Services percentages you give are certainly much, much higher than we typically see in real world deployments (ranging from small to very large scale deployments). For most of our deployments we see Services spend at 30-50% of license -- but this may be an artifact of the fact that our technology does not require a data warehouse or complex metadata to be built.

All: Where do the Services dollars get spent?

Paul.

something that keeps most of us BI folks employable

Metadata may account for part of the services cost, but if managed well data structure reduces costs elsewhere like where I think much of the services costs are targeted -- integration. Pretty difficult to achieve good business intelligence with incompatible data and software, and dangerous with poor data quality as we see over and over again in the real world, especially in systemic crisis and lack of prevention thereof.

In terms of consulting services, I haven't see much improvement since I left consulting in the early 1990s in terms of alignment, and IT I think it's become much worse generally evidenced by so few independents.

This quote caught my eye -- causal factor in most of the truly terrible diseases I find in the marketplace, whether in public or private sector-- careerism first, sustainability of the ecosystem/economy second, employing unnecessary and often systemically destructive complexity (IT, legal, tax, EDU/training, engineering..)

"something that keeps most of us BI folks employable - the global lack of effective metadata management."
http://dslayer.net/thetoastedlayer

That said, customer actual reports on TCO is almost certainly much higher than any vendor is seeing -- internal costs and consultants to mitigate service fees are two other issues that come to mind- .02--MM

BI Cost

If you would send me an email and I will be happy to share some input.