Part 2: Why don't pre-built applications solve the problem?

Part 2 of “Why is BI so broken?”

In an effort to make BI solutions less painful to deploy, some software providers have brought analytical applications to market that are based solely on specific data sources, like Salesforce or Oracle financials. The fundamental thinking behind this approach is: if it is so painful for customers to manually integrate and develop their own BI applications, wouldn’t it be easier if our organization just pre-built those manual configurations for them?

This was the fundamental value proposition behind the Sales Analytics Application family I developed while back at Siebel. Essentially, I used existing BI tools to create an out-of-the-box BI solution for the generic Siebel data model, including pre-configured ETL, data-warehouse, metadata, reports, and dashboards. The market responded so strongly to the promise of BI solutions that could be turned on in weeks that the product family grew from almost nothing to over $40 million a year in license revenue in just 3 years.

Pre-configuration doesn’t address the actual problem

The unfortunate reality, however, was that few companies (shhh!) actually used generic Siebel. Most organizations had taken advantage of Siebel’s metadata-driven application to highly configure their Siebel application, creating unique data models with custom objects and fields. This meant that the pre-built ETL, data-warehouse, metadata, reports and dashboards all had to be modified to reflect the actual data model employed by the customer.

In addition, most companies wanted to include some non-Siebel data in their data-warehouse (most commonly, this was back office sales data, since the Opportunity information typed in by a sales rep did not necessarily match the actual revenue the company received). This meant that the ETL, data-warehouse, metadata, reports, and dashboards had to be further modified to reflect this broader set of data that better captured the actual performance of the business.

Given the scale and scope of these changes for most companies, the speed-to-market benefit of buying a pre-built application quickly eroded. Most organizations found themselves facing the many months’ deployment timeline typical of most BI solutions. In short, masking the shortcomings of BI solutions by pre-configuring them doesn’t actually make the shortcomings go away.

“Pre-built” SaaS analytical applications don’t solve the problem, either

It is interesting to note that the market success of Siebel’s Analytical Applications has inspired several Software-as-a-Service companies to develop Analytical Applications that are limited to known data sources – most typically Salesforce.com. They are hoping that the promise of quick and easy BI will have similar success in the SaaS space as it did in the on premise.

While the basic idea seems sound and may generate revenue in the short term, I strongly suspect that those vendors (most of whom use existing open source tools like J-Pivot and Mondrian) will have a hard time meeting customer needs in the long run, because they still have to manually integrate and manage a solution across a highly fragmented stack. This means that these providers will struggle to adapt to each customer’s customized Salesforce data model and will likely be unable to cost-effectively integrate data from outside of Salesforce into the solution – thereby limiting the overall value of the system. I, for one, am very interested to see how the “pre-built” SaaS Analytical Application model fares.

Why is BI so broken?

Traditional BI solutions are painful, risk-prone, and slow to implement. They require significant investments of money, time, and people, yet still experience high failure rates. As a result, small and medium sized businesses can’t afford BI and large businesses are wrestling with runaway costs and disappointing results. Far too often, a CIO is left holding the bag of an overly expensive implementation of just a handful of inflexible reports.

How did this happen? BI is an immense industry – with companies already spending over $18B per year for BI solutions. Despite BI’s unimpressive track record, the promise of better decision-making and more effective management ensures that BI remains among the top priorities for CIOs. BI sales continue to grow, even in tough economic times.

High cost and complexity = limited reach and limited benefits

The expense and risk of traditional BI solutions make it hard for companies to deliver BI capabilities to all but the most critical portions of their organization – leaving many departments and groups without the insight they need to effectively manage their business.

Sadly, even those that parts of the organization that are served by traditional solutions find that once the systems are up and running, they tend to be brittle and expensive to modify and maintain. This makes it hard for a BI solution to keep up with the business’ ever-evolving needs for insight – limiting most solutions’ ability to deliver on the grand vision of better decision-making through insight.

Why? How can an industry so large still make it so hard for customers to succeed?

“Best-of-Breed” approach just breeds high cost and failure

The answer is that the fundamental structure of BI solutions is flawed. The best-of-breed approach that dominates the industry separates the different layers of the BI stack from one another. ETL tools are separate from the data mart, OLAP servers are separate from reporting servers, which are often separate from dashboarding products, and so on.

This menagerie of software products, each requiring different skill sets and expertise, needs to be hand stitched together by IT every time. The communication and coordination costs alone of this approach – never mind the hardware, software and development costs – are staggering. The cost of failure is high, since a mistake in any one layer can compromise the entire system. Given this structure, it is not surprising that many implementations fail. Those that do launch successfully have a hard time keeping up with the ever-changing needs of the business, since each change has to be reflected over and over again in all of the effected layers.

An integrated price list does not mean an integrated solution

Lately, larger players have tried to address customer demands for an integrated solution by buying up companies to fill out the other layers in the stack. Unfortunately for the customers, this approach just produces an integrated price list; the products are all listed together on one page, but they are not actually integrated technically. The cost and complexity of integration is just as much of a problem as it was before, and the customer’s desire for an integrated solution is not met.

Next: Why "BI Applications" don't necessarily solve the problem.

A new paradigm for customer value, Part 3 of 3

Part 3: Business Intelligence in a true SaaS model

In the last installment, this blog covered how, in general,  the SaaS model provides superior customer benefits and aligns the vendor with the customer.

It also showed how providers of SaaS-based operations software, such as CRM or ERP, quickly discovered the difficulty of incorporating analytics using traditional vendors.  It is clear that these operational SaaS vendors require a true SaaS business intelligence partner in order to be successful and meet customer needs.

It is harder to create a SaaS BI solution than a SaaS operational solution like CRM

Providing business intelligence in a true SaaS model is a fundamentally more difficult problem than developing an operational application like CRM in a SaaS model.   This increase in difficulty is due to the fact that business intelligence inherently requires the integration of data from other applications.  This integration can be very costly when performed using traditional business intelligence tools.

Early SaaS BI solutions try to sidestep the issue

In an effort to minimize this expense, most on-demand BI vendors have focused on creating packaged BI applications built on a few set tables within a known SaaS application, typically Salesforce.com.  However, because most SaaS BI applications are built using existing toolkits, typically including J-Pivot and Mondrian, the out-of-the-box applications are expensive and painful to customize as they fall victim to the same economics of the existing behind-the-firewall solutions. This short-coming is particularly painful for SaaS solutions since one of the greatest value propositions for SaaS software is its ease of configuration, making a rigid BI application completely unsuitable.

What *true* SaaS BI has to deliver for success

For business intelligence to be successful in the SaaS model, it must not only deliver more value than the traditional behind-the-firewall approach, it must also change the economics of BI delivery.  It must be a full featured solution that is quick to deploy, scalable, easy to use, and lower in cost.  Achieving this requires the following:

• 
  
  Providing automation for the most painful and laborious aspects of BI configuration – data preparation, integration, and metadata management


• 
  
  Automating the operational elements associated with BI deployments


• 
  
  New user provisioning


• 
  
  Addition of new data sources


• 
  
  Modification of existing data sources


• 
  
  New customer provisioning


• 
  
  Application replication (taking an existing configuration and re-using and deploying for another customer/deployment)


• 
  
  Upgrade management

Also, a SaaS BI solution should provide the key functional components that most BI consumers expect:

• 
  
  Ability to consolidate data from multiple sources


• 
  
  Ability to provide the full range of business intelligence end-user interfaces (operational reporting/banded reporting, dashboards, OLAP/slice-dice/pivot, alerting, etc.)


• 
  
  Ability to grow with an organization in terms of scale and complexity (number of data sources, complexity of calculations, diversity of end-user scenarios)

Vendors claiming to be SaaS BI have been around for many years now.  But true SaaS BI vendors are in fact quite rare.  You’re looking at one of them.

A new paradigm for customer value, Part 2 of 3

Part 2: The SaaS approach emerges, bringing economic benefits through the restructuring of software

As the flaws in the ASP model were gradually being exposed, a second generation of vendors began to emerge. These vendors did not take traditional enterprise software and try to force it into a hosted world.  Instead, they created new software platforms fundamentally designed for on-demand delivery that had the following characteristics:

· One, single version (with automated upgrades)

· A single, consolidated platform (the entire software stack)

· One target environment

· One deployment methodology

· Automated customer provisioning (new customers were simply turned on)

· Less configuration, more automation.  Users were offered simpler versions of the applications that satisfied the 80/20 rule, but radically reduced the configuration burden.

By rationalizing the problem and providing a stack that automates the provisioning, maintenance, and normal plumbing that goes along with the implementation and maintenance of software, SaaS fundamentally changed the economics of software delivery.  You may not be able to fully tweak every piece of the software, but people have discovered that most often, they don’t really need to. Instead of selling the kit to build the car, these companies sell the entire vehicle pre-assembled.

SaaS model aligns vendor and customer incentives

Moreover, since SaaS providers incur the cost of maintenance, operations and management, they are driven to reduce these costs.  It has been said that only 20-30% of the total cost of a software platform is license cost – the remainder is the cost of implementing, managing, and operating  the software (people, hardware and other resources). By focusing on this previously ignored part of the software value chain, SaaS providers can dramatically shift the value equation.

Operational SaaS is a success – but then there’s the BI problem.  How do you analyze the resulting data?

As mentioned earlier, software companies that provide operational applications (e.g., CRM, ERP) embraced the SaaS model first.  After a few years of growth and customer expansion, customers had accumulated a significant amount of business critical data in their operational solutions and naturally demanded analytical solutions for reporting.  Since analyzing data directly against an operational database can be catastrophic, vendors began searching for analytics partners.

SaaS operational vendors try to partner with traditional analytics solutions – with unimpressive results

Although these operational companies were based on the SaaS model, many of the potential partner vendors that they evaluated were traditional, behind-the-firewall BI solutions that were already established in the analytics industry.  Unfortunately, these traditional analytics vendors did not have true SaaS offerings, or were struggling to have their traditional model wedged into a SaaS-like delivery.  As a result, the operational software SaaS players were partnered with sub-optimal solutions.

Most often, the analytics partners’ setup costs and deployment method limited the SaaS vendor to the custom, one-off software delivery business when it came to analytics; they lost the compelling economics of their mainline businesses.  In a way, the analytics component ended up as a one-off consulting organization, an inefficient adjunct to the more scalable SaaS operational software model.  The takeaway is clear: mixing software models doesn’t work.  A SaaS vendor can’t redeem a traditional BI partner, and a traditional BI partner can’t add much value to a SaaS vendor.

Stay tuned for Part 3: Business Intelligence in a true SaaS model

Election hounds and pollster politicos - our "click around" demos are up

The Birst website now features the ability to preview what it feels like to use Birst - without even registering or using a password.  Just click on the link to the demo and check out dashboards and reports right away.

These demos are located on the Tours and Demos page.  Click here (http://www.birst.com/tour-demo.php)

For you election junkies, we have a demo with some great data on how the 2004 Presidential election broke out by state and congressional district.  We've also included the demographic data for that congressional district from the 2000 Census, so that you can see not only how a district voted, but how it breaks down by ethnicity, gender, income, etc.  Slice and dice the data any way you like, filter reports to see exactly the data you want, change the way you look at the world.  To go straight to this demo, click here.

If you're more of the armchair CEO type, we have a demo showing results for an imaginary company.  Or if you're a baseball fan, we have a number of charts and statistics on baseball history.

And as always, let us know what you think.  Email us at feedback@birst.com

A new paradigm for customer value: how Software-as-a-Service Business Intelligence offers more for less, faster

This three part series explores how software has moved from traditional solutions to SaaS and how the customer is finally getting the value that they need from BI delivered on-demand.

Part 1: The ASP model offers high hopes, but ultimately low value

The Software-as-a-Service (SaaS) business model is attracting significant attention from both customers and investors, since SaaS promises to fundamentally change the economics and customer value of software.  As a result, customers are increasingly turning to SaaS vendors for software solutions and publicly traded SaaS companies are receiving valuations far above those of traditional software vendors.

To understand why a SaaS model is so compelling, it’s first important to understand the limitations of another business model – that of application service provider (ASP).

The ASP Model – placing traditional software on the web

In the late 1990s and very early 2000s, the application service provider model was very popular. The theory was: take software traditionally delivered via CD-ROM and deliver it in a pre-configured, hosted model to create cost benefits for the vendor and customer. It was thought that by packaging the applications in this manner that the following would happen:

• 
  
  Economies of scale in operations could be achieved


• 
  
  Customer configuration demands would be limited


• 
  
  Deployments of new customers could be streamlined

If these things were accomplished, one could alter the economics of software delivery.

However, this model suffered from some critical flaws.

The critical flaw: the same software that was designed for customer installation and configuration was being used for hosted operation. Many of the design trade-offs that software vendors had made favored extreme configurability.   This allowed them to answer “yes” to feature/requirement requests from customers by essentially pushing back onto the customer the burden of installing and configuring the software.

The economics of traditional licensed software created strong incentives to discount the operational burdens of the customer and create half-features that,  in order to become fully functional, required the customer to finish the job with services.  In the traditional software model, the customer didn’t realize their obligations and added expense until well after the software is purchased. This is a critical reason for the disillusionment with enterprise software that has taken place over the last decade.

The ASP model essentially required that the hosting provider finance the cost of customer configurations. Instead of improving the economics for the vendor, it actually made the economics of software worse.  In fact, it has been argued that customers of ASP solutions demanded more from their ASP providers than they would have of their own internal IT had they relied on the internal organization for delivery.  These costs far outweighed any economic gains from standardizing operations and scaling up support staff.

The end-result: the ASP model did not create any economic benefits over a traditional software delivery model. The total cost of software delivery was shifted a bit, but not changed (if anything it got worse). ASP providers ended being sold-off and treated more as a conventional outsourcing option.  Customers did not see a significant reduction in cost or improvement in service.

Business Intelligence walks into the ASP trap

History has shown that every architectural shift in computing begins first in operational applications and then moves to the Business Intelligence market about 5-7 years later. The same seems to be true for the shift to the ASP hosted model.   The first generation of ASP companies provided operational applications.  BI providers followed later -- several attempted to take the same software that they sell via CD-ROM and provide a hosted alternative.  Despite calling themselves “SaaS,” these offerings are essentially the ASP business model.  As a result, they have encountered the same challenges that ASP providers of operational applications experienced; they are having a difficult time providing compelling value for a compelling price.

Stay tuned for Part 2: The SaaS approach emerges for operational software

Now what do we do? Ask Birst.

Like many other people, you’ve probably been watching the markets tumble downwards in the past few weeks and started wondering, what comes next?  And shortly after that, what do I do about all of this?  How should I move forward?

Now is the time to take stock of your business and think carefully about the strategic moves you’ll make in the upcoming months.  In order to make those decisions, though, you need solid information about your customers and business.  That’s where business intelligence comes in.  BI can give you insight into the drivers of revenue and costs, tell you which markets have the largest opportunities, or where to stop investing your efforts and resources.

Historically, BI solutions have been very expensive and difficult to use, requiring large sums of money and small armies of developers to get started.  Now there’s Birst.  Birst is the first automated, on-demand BI solution that’s so easy to use that you can get started using it today.  By yourself.

Birst is free to get started.  If you want fancier features and more application space, you can upgrade to our paid versions.  As you’ll see from our pricing, we’re far more affordable than other solutions.  And we scale to fit your needs as you grow.

So as you watch the market wobble and wonder how you can afford the insight that will tell you how to best move forward, know now that you can relax and stop wondering.  Try Birst.

Birst launched today.  Thanks to all of our beta customers and fellow Birst enthusiasts!

Birst moves to open beta

After a few months in private beta, when we were growing slowly to make sure that we could keep everyone happy with their Birst experience, we're now in open beta.  Anyone can sign up immediately, for free, without any special promo code or invitation.  Welcome!

Please take Birst for a nice data spin and let us know what you think.  Everyone gets free access to Birst Professional during the beta, so you've got all of the fanciest features at your fingertips.  You can post feedback in the forums or send us an email at feedback@birst.com.

Thanks for a few decades of patience. Get ready for the new new thing.

Yes, business intelligence is an old category.  It's been around in one form or another for decades.  Multi-billion dollar companies have successfully been created to try to meet customers' BI needs.  And yet those needs are still not fully met.

Traditional BI is expensive, hard to use, inflexible, and, ultimately, disappointing.  There is still a large number of people out there who are having a hard time getting the timely, relevant data they need to make important decisions.

That's why we created Birst.  It's fast, it's affordable, and it's powerful.  And it finally provides an answer to all of those people frustrated with spending hours using Excel trying to approximate a solution, all of those people who have outgrown Excel but can't afford an enterprise BI solution, and all of those people who have an enterprise BI solution but *still* can't get the answers they need.  That's a lot of people.

So if you're one of those folks, know that you're not alone.  And know that you now have a solution.  Welcome to Birst.  We hope you like it.