The Top 6 On-Demand BI Confusions

I’m a big believer in the future of on-demand BI, and yes, it’s hot.  But a lot of the articles promoting on-demand BI don’t always separate fact from self-interested fiction. Here’s my attempt to redress the balance.

1. On-Demand = Subscription Pricing = Mid-Market

There are three different axes that are often discussed as if they overlapped:

  • on-premise vs. on-demand
  • perpetual license vs. subscription license
  • enterprise vs. mid-market.

On-demand BI is typically subscription-based, aimed primarily at the mid-market, but it’s far from being the only valid combination.  and on-demand BI can’t

Perpetual licenses vs. subscription. On-demand BI isn’t the same thing as subscription pricing. For example, SAS has always charged an annual fee for their on-premise software. Perpetual licenses for a hosted services are rare, but they do exist outside of BI — e.g. lifetime fee offers for hosted web sites.

Enterprise vs mid-market. Mid-market organizations may well be attracted by on-demand options, but the BI vendors are all successfully selling mid-market on-premise packages. In addition, larger organizations are showing interest in hosted BI — although the extra complexity around data sources and customization may mean that it takes a while longer to become popular.

2. On-Demand BI is Cheaper

First, what does “cheap” mean? I haven’t yet found anybody who wanted to pay more for their BI software, but whatever the price there will be some people who can’t afford it (this includes “free”, since it always take effort to set up). According to Gartner, BI is the fastest-increasing investment priority for CIOs in 2007, indicating that BI is considered “cheap” by most organizations — i.e. they see possible benefits that outweigh the price.

On-demand BI is clearly targeting price-sensitive, lower-functionality-requiring customers first. But there’s nothing in the rule book that says that a hosted service has to be cheaper than an on-premise equivalent:

  • A major cost component — running the machines — is provided by the vendor. This means that total cost of ownership could be lower, but it also means that vendors could (“should” in a perfectly competitive market) charge higher prices than for equivalent on-premise software.
  • On-demand BI that charges subscription fees rather than a perpetual license can be cheaper initially — but more expensive over the long run. And again, subscription pricing is available for on-premise BI.
  • Equivalent functionality has to be taken into account — hosted on-demand doesn’t (yet) provide the same level of BI sophistication as on-premise BI, and so doesn’t have the same top-level prices. There’s also a huge market in cheap on-premise BI notably with Crystal Reports (Business Objects) and Microsoft.

3. On-Demand BI is Simpler and Easier to Use

This is a mantra of on-demand offerings, and it has to be true to some extent, or there would be little incentive to implement them, given the cultural misgivings about SAAS in general.

But there’s a certain amount of deliberate confusion between “simple” on-demand solutions (because that’s all that is available, or because the complexity is hidden by the vendor) and “complex” on-premise solutions (nothing says you have to every feature available, or implement it all yourself).

Ease of use: there’s clearly nothing inherently easier about an on-demand interface rather than on-premise interface. In fact, since most of the vendors that currently sell on-premise solutions support both web and full-client access, there’s greater choice (you can do analysis during that transatlantic flight), especially for power users. And two of the vendors that try the hardest to differentiate on ease of use and simplicity provide primarily full-client, on-premise products (Tableau and Qliktech). And Xcelsius, a full-client tool that produces online dashboards, also provides great ease-of-use.

Data integration and data quality: On-demand BI is not a magic wand, sadly. The data isn’t going to integrate itself or spontaneously become error-free. Having an on-demand BI vendor do it for you, on their platform, can certainly make your life easier — but so can a consulting company doing it on your machines. And having pre-built solutions for particular data sources and business uses can help with integration — but that’s also applicable to both on-demand and on-premise

Ease of installation. This is clearly an area where on-demand BI can claim to have an advantage — but not as much as you might think, especially with the introduction of BI appliances and virtual appliances.

4. On-Demand BI is Less Secure

In case I’m accused of being too hard on on-demand BI, here’s a misconception that clearly goes the other way. The perception of security is a cultural and trust issue, not a technological one. Do you keep your money under the mattress — or is it stored “on-demand” in a bank?

There’s no technology differences that accounts for why an organization should trust their funds to the international banking system but not their data to an on-demand BI provider.

Having trust, governance, compliance, or legal concerns is perfectly legitimate, but they are already dealt with on a daily basis with full outsourced IT deployments, so shouldn’t be an issue with on-demand BI.

5. On-Demand BI is Only about Software

A key part of BI 2.0 will be “information on demand” — the ability to easily augment internal data with external benchmarks, such as market share, economic predictions, etc. Note that the information on demand can easily be consumed by either on-demand and on-premise solutions.

6. On-Demand BI is The Future of BI

No, it’s part of the future. It’s not really a choice between on-demand and on-premise — organizations want it all, just like we don’t want to choose between the internet and our PCs (internet appliances were apparently one of the “eight biggest tech flops ever”.)

Despite the silly “no software” slogan (OK, not so silly: it ‘worked’), even has had to provide offline versions of its products for sales people on the road, and companies can spend lots of time connecting it to their internal systems.

Organizations ultimately want a solution that can be installed on-premise, used on-demand, or any combination in between. With service-oriented architectures, the boundaries between on-demand and on-premise solutions will become increasingly vague. Tomorrow’s BI users will be accessing a seamless combination of web services that are executed on systems inside their organization and systems that are outside their firewalls.

Other resources/articles that illustrate or deal with the points above…





2 responses to “The Top 6 On-Demand BI Confusions”

  1. Alan S Michaels Avatar

    How about: On-Demand BI Confusion point #7 ?
    7. Is there any confusion about what will be in BI 3.0; and when will solutions be available?
    BI 3.0 is just around the corner. BI 2.0 information on demand is already starting to merge with disciplines in competitive intelligence (see as well as strategic planning (see
    BI 3.0 will introduce the era of Automated Corporate Planning On Demand.
    It will include holistic corporate and business unit planning (including strategic, financial, and operational planning) which leverages internal and external BI information, software and services.

  2. Ken Rudin Avatar

    Timo, though I agree with some of your individual points, I think you’re down in the weeds and missing the main points behind why on-demand BI is so hot. Your comments about pricing models, deployment costs, target markets, data integration, etc. all miss the main point that traditional on-premise BI solutions are complex to deploy and manage. As you know, they typically include connectors to data sources, data integration technologies, data cleansing engines, a data warehouse, and an analytic engine. The main point of on-demand BI is that it’s ridiculous to ask someone to deploy and manage all that just to get even basic insight into their business. It’s like asking them to manage their own nuclear reactor.
    It makes much more sense to provide BI as a service. For example, it would be crazy to suggest to a company that they manage their own nuclear reactor when all they want is electricity. You could argue that having your own reactor gives you more control over your electrical power, and that you can customize it to generate however many volts you want. But, at the end of the day, no one really wants to manage their own reactor. We all gladly pay someone else do it for us.