To Be (To Cloud) Or Not To Be (Not To Cloud) BI

My colleagues and I have just completed yet another engagement with a large client — one of dozens recently — who was facing a to be or not to be decision: whether to move its BI platform and applications to the cloud. It’s a very typical question that our clients are asking these days, mainly for the following two reasons:

  1. In many cases, their current on-premises BI solutions are too inflexible to support the business now, much less in the future.
  2. The relative success of cloud-based CRM (SFDC and others) solutions may indicate that cloud offers a better alternative.

These clients put these two statements together and make the reasonable assumption that cloud BI will solve many of the current BI challenges that cloud-based CRM solved. Reasonable? Yes. Correct? Not so fast — the only correct answer is “It depends.”

Let’s take a couple of steps back. First, let’s define applications or packaged solutions vs. platforms (because BI requires both).

Packaged solutions

  • Subscribe to a solution-like CRM
  • Provide standard business functions to all customers (which makes it different from “hosting;” see below)
  • Difficult to tailor to specific needs
  • Usually are used synonymously (but incorrectly, see below) with software-as-a-service (SaaS)

 Platforms for building solutions

  • Subscribe to tools and resources to build solutions like CRM
  • Provide standard technical functions to developers
  • Contain limited, if any, business application functionality
  • Usually labeled either as platform-as-a-service (PaaS) or infrastructure-as-a-service (IaaS).

Now, let’s define “cloud.” First of all, the term is very often used synonymously with SaaS. But SaaS is just a business model; many on-premises solutions today (open source BI, financed/leased BI platforms, etc.) already use that. So let’s define what cloud-based SaaS is:

  • Application hosting. There’s nothing new here. BI vendors (IBM, Oracle, SAP, MicroStrategy, SAS, Information Builders, and others) have provided hosting options for their single-client BI on-premises BI platforms and apps for many years.
  • Managed service/application hosting. Again, nothing radically new. Leading BI consultancies (IBM, HP, Accenture, Capgemini, CSC, TCS, Wipro, Infosys, and others) also have been hosting and managing BI platforms and applications.
  • IaaS. This is highly standardized selective computing functionality — such as computing power, storage, archiving, or other basic infrastructure components — provided over the public Internet through a utility pricing and delivery scheme. The underlying computing resources are shared among a large number of users and hosted by the IaaS provider. There’s nothing BI-specific in these IaaS providers (like Amazon EC2, Google App Engine, and Microsoft Windows Azure), but anyone can take their on-premises BI platform and application and host it on an IaaS platform. Some partners of the leading BI vendors are already doing so and some native cloud BI platforms, such as GoodData, are based on such public IaaS platforms.
  • PaaS. This represents a complete preintegrated platform offering for the development and operation of general-purpose business applications. This is the least mature segment, with only a few emerging vendors. Open source vendors like Jaspersoft are forming alliances with PaaS vendors like RedHat to add BI PaaS functionality to RedHat’s OpenShip PaaS. Or native cloud BI vendors like GoodData are in the business of offering prebuilt analytical applications to business users or a BI development platform to professional developers who don’t want to deal with their own BI infrastructure.
  • SaaS. This pertains to standard software application functionality delivered over the public Internet and provided through a usage-based pricing model. The underlying application resources are shared among a large number of users. There are hundreds of domain- and industry-specific apps, such as SFDC for CRM and ADP for payroll.
  • BPaaS. Business process-as-a-service involves the provisioning of highly standardized end-to-end business processes delivered via dynamic, pay-per-use, and self-service consumption models. Today, this is mostly limited to specific domains such as customer data cleansing provided by vendors like D&B, Axciom, and HarteHanks.

So . . . when we at Forrester talk about cloud BI, we usually mean a combination of cloud BI SaaS and PaaS. Now that we have defined what it is, let’s drill into the “it depends” statement we made at the beginning of this blog.

As demonstrated in numerous research documents, BI is unlike any other enterprise software or application (like CRM or ERP), because:

  • It is difficult to define requirements up front. Business users may only know 20% or less of the reporting and analytical needs they will have in a few months.
  • It is difficult to put down BI requirements on paper in a standard requirements definition format. Business users often say, “Show me what it’ll look like and I’ll tell you what I want or don’t want.”
  • The “build it and they will come” baseball field analogy directly applies to BI. Only when business users see a report that they can touch and feel will the real requirements start pouring in.
  • As much as we like to be proactive, BI applications and platforms have to be very reactive, because change is the only constant in the world of BI.

These are the main reasons why one can’t compare BI with CRM (or any other enterprise app) and make a conclusion that cloud BI is a panacea to all BI challenges. Having said that, the move to the cloud is happening; it’s just happening much more slowly than in any other enterprise software segment. Why? In addition to the reasons stated above, most of the enterprise data (from CRM, ERP and all other apps) has to get to the cloud first! Then BI will surely follow. As of this moment, here’s the state of the market:

  • There are hundreds (or probably thousands) of domain- or industry-specific cloud-based BI applications. These are not BI platforms and are mostly used by small businesses or departments within large enterprises which have highly commoditized functions (procurement or HR analytics, for example). We expect these cloud-based applications to continue to pop up all over the place.
  • Leading BI vendors like Actuate, SAP, and MicroStrategy are offering cloud-based versions of their on-premises platforms. While these are indeed BI development platforms, they provide a subset of these vendors’ full on-premises functionality (some more, some less). We expect more leading BI vendors to enter this space and for the BI PaaS functionality to mature in two to three years. Why so late and so long? Re-architecting a platform natively build for single client tenancy for multitenancy that cloud based SaaS and PaaS require is not an easy task. The best example demonstrating this are all admin functions. In a single tenancy platform an adimistrator "rules" the entire environment, while in the multi tenant environment a platform has to have an instance administrator and an "uber" administrator. This significantly affects not just administration, but security, provisioning and many other capabilities
  • Startups like GoodData, Birst, Bime, and Alteryx will continue to enhance their functionality and narrow the gap between their offerings and the offerings of the leading BI vendors. This is a perilous road, though, and only a few will survive — we’ve seen plenty of failures in this segment already.

I hope this post did not sound too negative about the cloud BI market; it is indeed one of the top current and future BI trends that we predicted and currently track. I only intend to separate fiction from reality and help our clients make more informed “to cloud or not to cloud BI” decisions.

As always, all comments are more than welcome.


Not too negative


Thanks for sharing this. The only thing that comes to mind which is beyond this brief format is that the cloud and mobile individually and combined offer not only new and different technical functionality than on-premise (as well as a few unique challenges), but also the opportunity for new business models. So we are likely to continue to see new and different use cases and value derived from the cloud than on-premise quite apart from just a great deal of new competition in the market.

Interestingly, some of the underlying drivers in technology enabling the changes currently are also bringing new innovation, functionality and at least the possibility of cost efficiency to on-premise as well. I personally think the combined impact will broaden the definition of BI with convergence just simply because it's convenient, efficient, and relatively inexpensive compared to previous generations to do so. Of course none of this will lessen the need to translate increasing volumes of data to decision making which you point out consistently, rather if anything quite the opposite.

There are lots of failures in all segments of tech -- haven't run the numbers but doubt it's any higher in BI than it is in other relatively comparable segments.

One thing I am seeing is a growing gap of needs and budgets reflective of the broader economy. Some budgets are under severe strain and looking primarily to cut costs-- great if they can do so with new and better functionality in the process, but many many many are on trajectories that are not sustainable and cannot afford the big spend traditionally associated with enterprise software and services. I think that trend is under reported and misunderstood even by senior strategists in companies/vendors focused on those markets-- or we are just looking at different numbers and talking to different people. This segment represents a big part of the push for cloud I am seeing in the market.

On the other side of this uneven cauldron are organizations with more capital than they know what to do with, including quite a few of the market leaders in IT. The relationship between the have mores than they know what to do with--including increasing numbers in developing markets--and have not enough to sustain, is perhaps less of an issue in BI than enterprise software more generally, but it's I think probably the most misunderstood dynamic taking place in our economy and not reflected well at all yet by the majority of models in this industry. -- MM ---PS- thanks for the mention in your Future of BI report -- we've invested a great deal of serious R&D over a long period in directly related and overlapping areas. The educational voyage continues....

Mark, as always, thanks for

Mark, as always, thanks for the thoughtful comments and contributions. One of my jobs these days is to convice clients that technology (cloud or not) is a small (significant, but small) part of BI equation. While cloud BI will make a dent in the BI technology and support budgets, it will not address all other BI challenges which hugely contribute to BI costs and complexities: clean, integrated and reconciled data, governance and other processes surrounding BI such as SDLC and PMO, BI ownership, and many, many more.

Great article


Very good summary of the state of the market. Here are my two comments:

- BI is and always was an application data aftermarket. We experienced it in the client/server era and we see it today as well. This means that cloud BI is not happening more slowly. There is simply time lag - people have to move their applications and application data to the cloud first.

- "Cloud-based versions of on-premises platforms" is an oxymoron. Google Docs is not a hosted version of MS Word. The architecture and internal structure of GoodData PaaS BI has nothing in common with the traditional on-premises platforms. Our platform is architected, deployed and operated very differently and the cloud design point gives us the scale, flexibility and innovation velocity that is out of the reach of traditional architectures.

Roman Stanek
Founder, CEO - GoodData

Roman, very valid points,

Roman, very valid points, thanks. Making updates to the blog now.



A couple of questions to ya, if I may -

1. I'm am very curious as to your opinion on what kind of pain cloud BI really solves... Even if adoption does pick up, the issues with BI were never about slicing and dicing an Excel file. The challenges were always around data integration and management. As more and more applications move to the cloud, the integration problem remains the same as there is no one vendor that host all application data in the world. So while you could host Zendesk's data in your data warehouse, how does that help someone who wasn't to mash it up with another cloud application that is on a completely different 'private cloud'?

2. I'm also interested to know how exactly GoodData's architecture scale better than "traditional architectures". In fact, I don't even understand what you're trying to say. What exactly do you mean by "traditional architectures"? I don't claim to know anything about anything, but I do know the incumbent ("traditional") BI players have customers that run deployments on a single site that are probably larger than your entire customer-base combined... I'm not a big fan of old BI, but still, credit must be paid where it is due.


BPM applications share

BPM applications share similar characteristics to BI in that they are difficult to define up front and by their very nature must continuously change. The cloud is a solutions delivery framework and not a platform delivery framework. Thing is however in order to deliver the solution the platform must be cloud enabled first.

Peter, excellent point.

Peter, excellent point. Agility, flexibility, reactivity are indeed common themese that BI and BPM share.

BI and BMP

There is a fundamental difference between implementing BPM and BI, especially the business process analytics. The main obstacle for BPM is defining the process up to the lowest-level details, including all special cases. As a result companies are reluctant to introduce BPM to the existing processes. BPM adoption rate speaks volume to it.

Process analytics in BI only requires defining the key milestones in the process which is much simpler task. Moreover, this task is incremental, i.e. a company may start with analyzing very few obvious milestones and then "drill" into the parts of their processes which require more attention.

Thanks Boris...

First time have come across such a clear definition relating BI to the Cloud...definitely separates fiction from reality

Thanks, we try. Watch for a

Thanks, we try. Watch for a Cloud BI Wave that I'll be kicking off in a few days