Instrumenting Your Enterprise for Maximum Predictive Power

By James Kobielus

Business is all about placing bets and knowing if the odds are in your favor.

As I noted in my most recent Forrester report, business success depends on your company being able to visualize likely futures and take appropriate actions as soon as possible. You must be able to predict future scenarios well enough to prepare plans and deploy resources so that you can seize opportunities, neutralize threats, and mitigate risks.

Clearly, predictive analytics can play a pivotal role in the day-to-day operation of your business. It can help you focus strategy and continually tweak plans based on actual performance and likely future scenarios. And, as I noted in a recent Forrester blog post, the technology can sit at the core of your service-oriented architecture (SOA) strategy as you embed predictive logic deeply into data warehouses, business process management platforms, complex event processing streams, and operational applications.

The grand promise of predictive analytics—still largely unrealized in most companies—is that it will become ubiquitous, guiding all decisions, transactions, and applications. For the technology to rise to that challenge, organizations must move toward a comprehensive advanced analytics strategy that integrates data mining, content analytics, and in-database analytics. Already, we’ve sketched out a vision of “Service-Oriented Analytics,” under which you break down silos among data mining and content analytics initiatives and leverage these pooled resources across all business processes.

You may agree that this is the right vision but have doubt about whether there is a practical, incremental roadmap for taking your company in that direction.  In fact there is, and it starts with re-assessing the core of most companies’ predictive analytics capability: your data mining tools. As you plan your predictive analytics initiatives, you should avoid the traditional approach of focusing on tactical, bottom-up project-specific requirements. You should also try not to shoehorn your requirements into the limited feature set of whatever modeling tool you currently happen to use.

To become a fully predictive enterprise, you will need to take both a top-down and bottom-up approach to your data mining initiatives. From the top-down, it’s all about building and integrating alternate models of how your business environment is likely to evolve internally and externally. In our recent report on advanced analytics, Boris Evelson, Leslie Owens, and I sketched out the many business processes that can be enriched by predictive analytics.

So how do you instrument your company to become more predictive? For starters, assess whether your analytics tools support the following capabilities for developing, validating, and deploying predictive models:

• Model multiple business scenarios: You should be able to build complex models of multiple, linked business scenarios across different business, process, and subject-area domains, using such key features as strategy maps, ensemble modeling , and champion-challenger modeling.
• Incorporate multiple information types into models: You should be able to develop models against multiple information types, including unstructured content and real-time event streams, while leveraging state-of-the-art algorithm in sentiment analysis and social network analysis.
• Leverage multiple statistical algorithms and approaches in models: You should be able to develop models using the widest, most sophisticated range of statistical and mathematical algorithms and approaches, including regression, constraint-based optimization, neural networks, genetic algorithms, and support vector machines.
• Apply multiple metrics of model quality and fitness: You should be able to score and validate model quality using multiple metrics and approaches, including quality scores, lift charts, goodness-of-fit charts, comparative model evaluation, and auto best-model selection.
• Employ multiple variable discovery and assessment approaches: You should be able to build and validate models using various approaches for variable discovery, profiling, and selection, including decision trees, feature selection, clustering, association rules, affinity analysis, and outlier analysis.

How is this different from predictive analytics as usual? Traditionally, most predictive modeling specialists focus on the latter three capabilities: statistical algorithms and approaches, model quality and fitness, and variable discovery assessment. Most models are built in narrowly scoped business or subject domains—such as customer analytics for marketing campaign management—and only against structured data sources (such as relational tables). Traditionally, few predictive analytics projects have entailed modeling of multiple business scenarios across diverse domains--such as sales, marketing, customer service, manufacturing, and supply chain-- though in the real world these business processes are often quite interconnected. Also, many data mining initiatives fail to incorporate information from unstructured sources—such as text in call-center logs—though this content may be as important as what comes relational databases and other structured sources.

It’s very important to build multi-scenario predictive models against complex information sets, but becoming a fully predictive enterprise demands much more. To instrument your organization for maximum predictive power, you should also tool your advanced analytics to support the following capabilities:

·         DW-integrated data preparation: To speed up and standardize the most time-consuming predictive modeling project tasks, you should be able to leverage your existing data warehouse, extract transform load, data quality, and metadata tools to support a full range of data preparation features. These features include the ability to discover, acquire, capture, profile, sample, collect, collate, aggregate, deduplicate, transform, correct, augment, and load analytical data sets.

·         Deep application and middleware integration: To deliver models deeply into whatever heterogeneous SOA-enabled platform you happen to use, your predictive analytics tool should deploy on and/or integrate with a wide range of enterprise applications, middleware, operating platforms, and hardware substrate. You should be able to deploy models seamlessly into your data warehouse, business intelligence, online analytical processing, data integration, complex event processing, data quality, master data management, and business process management environments. And to play well in your SOA, your predictive modeling tools should support application programming interfaces, languages, tools, and approaches such as Web services, Java, C++, and Visual Studio, as well as emerging languages such as SQL-MapReduce and R.

·         Consistent cross-domain model governance: To avoid fostering an unmanageable glut of myriad models, your predictive analytics solution should support a wide range of tools, features, and interfaces to support life-cycle governance of models created in diverse tools. At the very least, your tools should enable model check in/check-out, change tracking, version control, and collaborative development and validation of models. To realize this promise, it should support a full range of tools, standards, and interfaces for import and embedding of models from other tools, as well as export and sharing of models to other environments.

·         Flexible model deployment: To execute modeling functions--such as data preparation, regression, and scoring—on the widest range of data warehouses and other platforms, your tools should support in-database or embedded analytics. And to scale to the max, your predictive analytics tools should deploy models to massively parallel data warehouses, software-as-a-service environments, and cloud computing fabrics.  Your advanced analytics tools should also support development of application logic in open frameworks—such as MapReduce and Hadoop—to enable convergence of data mining and content analytics in the cloud.

·         Rich interactive visualization: To deliver their precious payload—actionable intelligence—your advanced analytics tools should support interactive visualization of models, data, and results. Ideally, you should be able to visualize all of this in your preferred business intelligence tool, or in the predictive modeling vendor’s integrated visualization layer. Of course, you have every right to expect the full range of visualization techniques--histograms, box plots, heat maps, etc.—regardless of who provides the visualization layer.

As you can see, this goes well beyond data mining as usual. Forrester has a slightly different perspective on the development of the predictive analytics market than you’re likely to get from other sources. We see a robust, flexible, SOA-enabled data mining tools as the centerpiece of advanced analytics for fully predictive enterprises. The competitive stakes are too great for businesses to take the traditional silo-mired approach when implementing this mission-critical technology.

What do you think?

Informatica Evangelizes Its Increased Business Relevance With Latest Release

By Rob Karel

Last week Informatica announced the release of Informatica 9, its data integration/data management platform that continues to evolve its flagship PowerCenter and PowerExchange data integration and access technologies into a much more comprehensive data management platform going well beyond the scope of traditional, batch-oriented ETL that remains Informatica’s bread and butter.  

The three main themes Informatica has pitched for this release include:
- Pervasive Data Quality
- Business-IT Collaboration
- SOA-based Data Services

While these themes and capabilities - reusability, SOA-compatibility, real-time, business engagement - are not necessarily new to the broader data integration or data quality software markets, few organizations have been effectively able to execute on them.   For the purposes of this blog post, I’d like to focus a bit more on the DQ and business/IT collaboration parts of the announcement.

With pervasive data quality, Informatica wants its customers to extend the use of data quality software well beyond the limits of data warehousing.   Informatica 9 provides role-specific tools that can be leveraged by both the IT and business stakeholders involved in defining, analyzing, testing, and implementing data quality logic.   Informatica has also included advanced postal address verification and global matching from its acquisitions of AddressDoctor and Identity Systems, respectively.   

I’m a huge fan of Informatica’s goal of centralizing the definition and management of data quality rules that can then be delivered as real-time data services to any part of the enterprise.  But as I discussed in my “It’s Time To Invest In Upstream Data Quality” research last year and most recently with Forrester BPM analyst Clay Richardson in “Warning: Don’t Assume Your Business Processes Use Master Data”, the biggest inhibitor to enabling this is not technical.  Business process owners responsible for the capture and update of critical enterprise information are too often more concerned with process efficiency than data quality issues that may impact downstream processes. 

Similarly, I think Informatica is on the right track with its Business-IT Collaboration theme, but has a lot more work to do from evolving this from vision to reality.   Informatica 9 provides some useful business-oriented capabilities including browser-based tools designed for business analysts and data stewards to participate more actively in the data profiling and DQ rule definition phase of a data quality initiative.   And from a collaboration standpoint, the shared metadata environment ensures both business and IT have access to the same information, and the ability to bookmark and share views of analysis and business rule output, along with comments, is a great feature that eliminates the old method of exporting results to a spreadsheet and attaching to an email.   My primary concern is that Informatica has not yet created a pre-packaged interactive collaborative environment.  For example, there are no predefined approval, escalation, and validation workflows.  

Interestingly, almost every major data quality and master data management vendor I speak with is investing quite heavily in what some are trying to define as the Data Governance Technology market. Or in other words, tools used to support both business and IT stewards in the data governance process.   Of course data governance is not a technology market, it is a business responsibility that must manage the people, process and organizational challenges required to get business value from data assets.   But there are enabling technologies that can support data stewards from both the business and IT side during certain steps of the data governance lifecycle, and the jury is still out on what approach to this technology will be the most effective. 

But the biggest challenge for enterprises trying to embrace data governance has nothing to do with the tools.  It’s getting the business to participate in the first place!   In a recent Forrester survey asking MDM and DQ professionals to rate their organization’s current level of data governance maturity, 79% said they did not have a data governance organization that included active participation from all the necessary business stakeholders. Unfortunately, there is very little that a software vendor can do to change this culture, so I’m not expecting widespread adoption from the business of Informatica’s or anyone else’s data governance tooling until that maturity improves.

What Makes BI SaaS Architecture Multitenant?

By Boris Evelson

I'd like to drill into some more details on my BI SaaS blog from September 2009. A key critical point to "what differentiates one BI SaaS vendor from another" discussion is what really constitutes multi-tenant architecture. Here are some initiall thoughts to stimulate the discussion:

• DBMS. There's got to be back end, DBMS architecture that allows for one of the following:
• • Automatically generate a separate DBMS instance for each client  
  • Use same DBMS instance for multiple clients, but automatically generate a set of unique tables for each client  
  • Use same DBMS instance and tables for multiple clients, but automatically assign unique keys to to each client so that they can only update and retrieve their own rows  
• Application. Similar functionality has to exist in the application tier:
• • Automatically connect to the appropriate, client specific DBMS instance, or 
  • Automatically use views that only point to client specific tables, or
  • Append "where" clause to each SQL statement to only retrieve client specific rows  
• Provisioning. All of the above functionality has to be automatically executed at a push of a button, a swipe of a credit card, with zero involvement from vendor support staff or internal IT organization.

I believe that these features truly differentiate self service BI SaaS from hosting, virtualization, private clouds, etc, that some vendors are also trying to position as BI SaaS. All these other services and architectures are important, relevant and applicable to multiple situations and requirements, but they are not BI SaaS IMHO. I'd love to hear everyone's thoughts on what else is relevant here.

How To Differentiate Advanced Data Visualisation Solutions

By Boris Evelson

I get many inquiries from clients on how to select a data visualization vendor / solution. The criteria that my clients often site are

• Thick and thin client
• Dynamic visualizations, not just static charts 
• Ability to pull data from multiple sources
• OLAP-like functionality

All these criteria are pretty much a commodity these days. The real differentiation will come once you start looking at advanced (key word "advanced") visualization features such as

• If it’s a thin client does it have Web2.0 RIA (Rich Internet Application) functionality (Flash, Flex, Silverlight, etc)?
• In additon to standard bar, column, line, pie charts,etc how many other chart types does the vendor offer? Some advanced examples include heat maps, bubble charts, funnel graphs, histograms, pareto chats, spider / radar diagrams, financial market specific charts (candlesticks, etc), project management specific charts (gantt, etc), scatter plots, relationship diagrams, overlapping sets, cluster analysis, and others?
• Can the data be visualized via gadgets/widgets like temperature gauges, clocks, meters, street lights, etc?
• Can you mash up your data with geospatial data and perform analysis based on visualisation of maps, routes, architectural layouts, etc?
• Can you have multiple dynamically linked visualization panels? It’s close to impossible to analyze more than 3 dimensions (xyz) on a single panel. So when you need to analyze >3 dimensions you need multiple panels, each with 1-3 dimensions, all dynamically linked so that you can see how changing one affects another
• Animations. Clicking through 100s of time periods to perform time series analysis may be impractical. So can you animate/automate that time period journey / analysis?
• Multi dimensional charts. Can you have a 3rd, 4th, 5th, etc dimensions, such as a size of a bubble, color of a buble on an XY axis?
• Subsets ands supersets. Can you select a point, rectangle, or a "lasso" (irregular region) and drill down into the subset or roll up into a superset (if selecting across several charts)?
• Can you have microcharts (aka trellis or sparklines) – a two dimensional chart embedded in each row or cell on a grid?
• Can you do contextual or gestural (not instrumented, not pushing buttons, or clicking on tabs) manipulation of visualization objects, as in video games or iPhone like interface?
• Is the data that is being analyzed
  a) Pulled on demand from source applications?
  b) Stored in an intermediary DBMS
  c) Stored in memory? This last one has a distinct advantage of being much more flexible. For example, you can instantaneously reuse element as a fact or a dimension, or you can build aggregates or hierarchies on the fly.
• Is there a BAM-like operational monitoring functionality where data can be fed in into the visualization in real time?
• In addition to historical analysis, does visualization incorporate predictive analytics components?
• Portal integration. If you have to deliver these visualizations via a portal (SharePoint, etc) do these tools have out of the box portal integration or do you need to customize.
• Can you reuse metadata, such as KPIs, KPMs, that was already setup in your BI or ETL or DW models?

What did I miss?

Next Gen BI Is Here Today

By Boris Evelson

As I promised in my last year “BI Crystal Ball” blog, BI will look very different in the near future. Indeed most of the 20 next generation BI trends that I list in this blog are actually here today! And, as my colleague Alex Cullen wrote in his recent research document on 15 Technologies That Matter(where of course BI is front and center), these are the technologies that indeed matter. I picked these 20 trends not because the market labels them as trendy, but because I have tons of evidence that the buy side of the BI market can use these technologies today. These technologies can address the real pain points, gaps and challenges that most knowledge workers are faced with today.

 

So what has really changed? The “what-has-to-be-done” of BI has not changed, and will probably not change in the foreseeable future. One would still need to discover, ingest, persist, measure, analyze, report, deliver and act on information. The real trick is “how-it-will-be-done”. I categorize all major next gen BI trends under 4 categories: Automation, Unification, Pervasiveness and No Borders / No Limitations. Here’s my take on what they are:

 

Automation

• Automated Information Discovery. Automatically discover source data based on metadata, content and rules.
• Actionable and Collaborative. Automate actions on results gleaned from BI reports and analysis.
• Contextual (process, desktop, visual). 1) Process: BI is aware of process context; 2) Desktop: BI is aware of desktop (open documents, open emails) context; 3) Visual: gestural, not instrumented visualization manipulation (akin to video games).
• Built-in expertise. Sensing what the user is doing and providing proactive hints.
• End-to-end lifecycle management. Automatically changing and updating all interdependent BI components.
• Decision Management. Automatically documenting and automating all business decisions.
• Self learning and adaptive. Automatically modifying data model based on reporting and analysis usage.

Unification

 

• Logically unified sources. Logical views independent of physical data location and model
• Data and content. Seamless integration of structured data and unstructured content
• Disk and streaming. Seamless integration of data stored in disk based DBMS (historical, batch) and in streaming, in memory DBMS (real time)
• Historical, current and predictive.Seamless integration of historical, current, and predictive reporting and analysis
• Metadata. Unified ETL, data quality, BI, portal, content, process, rules metadata

Pervasiveness

•  Within processes.BI in processes, automatically comes up when a decision needs to be made by a human
• Within Information Workspace. Integration with all IW components: search, ECM, email
• Self Service.Casual user and power user self service, including, but not limited to BI SaaS.
• Offline, disconnected. Seamless operation in offline / disconnected mode
• Mobile. Delivery and analysis on mobile devices

No borders

 

• On-demand data models.Reporting and analysis without limitations of underlying data models, aka “post discovery”
• Unlimited dimensionality. Ability to visually analyze many dimensions at the same time
• Exploration + analysis.Seamless integration of search-like technology to find the data and BI-like technology to report and analyze the results of the search

Drop me a note if you’d like more details on each of these trends, including top vendors that even today have capabilities in these areas.

Ten Strong Hints Your Enterprise May Not Have A BI Strategy

By Boris Evelson

You know that you don't have an enterprise BI strategy if:

1. Your end users keep pointing to IT as the source of most BI problems
2. Your business executives view BI as another cost center
3. IT staff keep asking end users for report requirements
4. Your BI is supported by IT help desk
5. You can’t tell the difference between BI and Performance Management
6. You can’t measure your BI usage
7. You can’t measure your BI ROI
8. You think your BI strategy is the same as your DW strategy
9. You don’t have a plan to develop, hire, retain and grow BI staff
10. (My personal favorite) You actually don’t know if your enterprise has a BI strategy!

Want more hints and best practices? Take our BI maturity self assessment test. Then if you actually want to compare your BI strategy and BI maturity against other Forrester clients, take our BI maturity survey. Good luck!

Process-centric DQ Services and MDM among top technology trends to watch

By Rob Karel

For the past few years, master data management (MDM) has been a hot topic among enterprise architects and data management professionals. But many of these evangelists have felt like rebels without a cause because their MDM initiatives have been primarily IT driven with minimal business sponsorship and participation.   In other words, the business expects high quality data, but hasn’t taken much accountability in delivering it. 

Data quality (DQ) management, a more mature technology and competency than MDM - and a required core capability for any MDM solution - has also struggled to be embraced by the business as a top priority.  More often than not, DQ has been relegated as a supporting downstream batch application for ETL, data warehousing and business intelligence applications, but not for a broader cross-enterprise data architecture including upstream transactional systems and processes.

But Forrester now sees this trend changing for the better, as highlighted by my colleague, Alex Cullen, who recently published important research on “The Top 15 Technology Trends EA Should Watch.”  A key theme Alex identifies is around Process-Centric Data and Intelligence. Alex believes - and I agree with him - that both MDM and real time data quality services will become critical enablers to support real-time operational and decision-making processes.  

Organizational competency in the definition and implementation of DQ rules, processes and technologies that can effectively cleanse, standardize, enrich, match, and merge critical data is a foundational step to moving towards true MDM:  the real-time delivery and synchronization of a single, trusted view of enterprise reference data to both operational and analytical environments.  Why? Because you can’t bi-directionally synchronize master data if you haven’t implemented the DQ rules required to derive a master record.

Unfortunately in many instances the after-the-fact batch cleansing of data in a data warehouse is often too little too late in ensuring high quality data can be delivered via MDM or otherwise.   This is why we see real-time DQ Services as a critical evolution for data architects to embrace and evangelize: embed DQ validation and cleansing logic within the upstream transactional processes that capture and maintain this data in the first place.   Not only will this of course improve the trustworthiness of the data used for analysis later, but even more transformational – it will improve the efficiency and quality of the upstream transactional processes themselves. 

A simple example:
A “Ship To” postal address is typically captured as part of an order management process.  Before real-time DQ Services are implemented to ensure that the addresses captured meet formatting standards and are in fact deliverable, order fulfillment runs a very real risk of being delayed – leading to potential postage penalties and negative customer satisfaction due to the delays.  But DQ Service can significantly reduce this risk by recognizing these DQ issue while the customer is still engaged in the process and best able to respond.

As further evidence of this trend toward transactional data trustworthiness, in my just-published “Trends 2009: Master Data Management” research I share the results of a recent Forrester MDM readership survey where over 64% of respondents said their planned or implemented MDM solution must support real-time subscription requirements, and not just scheduled batch loads.   Therefore it’s not too surprising that almost 79% of those surveyed that have decided upon an architectural deployment style for their MDM solution selected a Hub, SOA, Federated, or Registry approach since all of these are optimized to support real-time, transactional MDM requirements.  

But don’t expect this transformation to real-time DQ and MD to happen overnight. The complexity and invasiveness of these data services to the streamlined transactions that are managed make many EA’s wary, especially since most organizations still have extremely immature data governance programs.   In the same readership survey, nearly 46% of companies rate their data governance maturity as low or very low, while another 29% rate their maturity average, seeing significant participation from just a few key business stakeholders.   Data governance – an organizational and cultural shift – is the first trend EA’s must follow and evangelize in order to deliver the process-centric data and intelligence their organizations are demanding.

BI Saas Vendors Are Not Created Equal

By Boris Evelson

As many of my readers know, for years I’ve been quite skeptical about non-mainstream BI solutions, such as BI SaaS. Security, control, operational risk, data, metadata and application integration are just some of the requirements for enterprise BI that are still on my watch list as potential reasons to be weary about BI SaaS. However, I am also a very pragmatic analyst and truly believe that nothing but supply and demand drive the markets.  And I am now, slowly but surely, beginning to believe there couldn’t be a better case for demand for BI SaaS especially after findings from one of the project that I am currently conducting.

I recently talked to a few dozen non-IT professionals (specifically in front office roles, such as sales and marketing) across multiple industries, regions and company sizes. Guess how many of them fully or partially relied on IT for their day to day operational and strategic information needs? BIG FAT ZERO!!! This finding was a huge surprise to me – yes, I did expect to find something like less then 50% reliance on IT, but I surely did not expect to find 0%.

It is truly amazing that after 30 or so years of BI software, services and solutions vendors and internal IT organizations making a strong push and a case for BI, they are still not even making a dent in the front offices. What are the reasons? Among many others, sales and marketing folks that I interviewed are citing being low on IT priority list (why???), IT not speaking their language, high cost, inflexibility of company standard BI solutions, lag times to roll something out – and the list goes on and on and on.

So what’s a sales or a marketing professional to do? Fortunately, they do have options. I am sure you all know what their option #1 is. Yes, you guessed it – Excel. But while Excel remains and will continue to remain for the foreseeable future #1 BI tool, it does have lots of limitations, and all of the business folks I talked to are looking for something that has more pre-built BI solutions, can handle gigabytes (and eventually terabytes) of data, and is more Web 2.0 and collaborative than Excel. Many BI SaaS vendors indeed provide just that.

Is BI SaaS a panacea to all these issues and challenges? Absolutely not. My concerns cited at the beginning of this blog still need to be addressed. Chief among them is the financial viability and long term survivability of BI SaaS vendors, since most of them are tiny, startup companies. So how do you short list a BI SaaS company and mitigate your risks?

First consider some soft differentiations. While these are hard to verify, and may or may not directly correlate and contribute to ultimate vendor success, these are a good starting point (or ending point, to put the final touches on your evaluation) to begin shortlisting your BI SaaS choices:

• VC backing. Is the firm backed by a VC with good track record in information management space?
• Profitability or loss run rate. Is the business profitable or is the loss rate manageable, predictable and adequately financed till the planned break even / profitability goal?
• Salesforce.com recommendation. Does Salesforce.com (or another major vendor on whose platform, data source, etc analytics are based) provide favorable or unfavorable recommendation?
• Management team. Does the management team have a good track record with successful startups and solid BI experience?

Next, proceed to evaluating hard facts. These have to be easily verifiable, and will most probably directly contributes to the ultimate success of a vendor:

Architecture and technology

• Multitentant architecture. Is the architecture truly multitentant or is SaaS offering really an ASP/MSP under the covers ? In other words, can a customer swipe a credit card and get provisioned instantly without a necessity for any manual intervention, setup, etc? This, by the way, is the main reason why I often exclude mainstream BI vendors from BI SaaS category – but I welcome their challenge to this point.
• In the cloud infrastructure. Is the vendor in the business of supporting servers and DBMS, or is all infrastructure hosted by in the cloud vendors like Amazon or Google? Can the vendor concentrate its efforts 100% on functionality and not infrastructure?
• Salesforce.com dependency. Is the software offering 100% dependent on a single data source? Does the vendor have anything else to fall back on should that one single dependency not work out? Some other typical sources include Netsuite, Quickbooks online, clickstream data from Google, payroll data from Paychex, and others.
• Fixed or flexible data model. Does the vendor provide a very industry or functionally specific data mode, vs. can you build your own data model or both?
• Metadata. Does the tool give you capability to import/export metadata so that business and technical metadata can be integrated and reused with other enterprise applications?
• Data integration. Does the vendor provide simple (flat file) import mechanism, SQL or MDX based import or a full blown ETL procedural/scripting language? Does the vendor automatically infer  and build a star scheema or requires you to do the target data modelling?
• APIs, Web Services. Can product functionality be modified, exposed and reused in other applications via APIs or Web Services?
• 11 styles of BI. Does the vendor provide BI functionality to support all 11 styles of BI (8, such as reporting, querying, OLAP, dashboards, etc are listed in my recent BI Wave)?
• Architectural secret sauce. Since any SaaS business is in danger of becoming commoditized does a vendor have a “secret sauce”, a protectable IP that truly differentiates them?

References

• Are customer references available? From customers in production (not POCs or prototypes)? What is the total number of verifiable customer logos in production? How diverse is this verifiable customer base by industry, enterprise size, function?

Support

• Set up services. Does the vendor offer software and application setup services?
• Integration services. Does the vendor offer broader data integration services?
• Management consulting services. Does the vendor offer strategic, management consulting-type advisory services?

Hopefully, this checklist is comprehensive and practical enough to help you shortlist a BI SaaS vendor that fits your business needs and is less then likely to fail in the near future. However! You still need to mitigate the risk by

• Backing up your own data
• Having a BI SaaS vendor Plan B.  Keep track of the other BI SaaS vendors - and be ready to have one of them pick up the business if your current BI SaaS vendor fails.
• Having a commitment from Salesforce.com (or the other appropriate ERP vendor) to support your SaaS analytics in case your BI SaaS vendor fails
• Having your internal IT resources stand by and have a plan and a periodic fire drill on how to take over and migrate your data into the enterprise BI solution

Now that I re-read my blog – I am thinking, why not actually poll all current BI SaaS vendors and have them fill out the above questionnaire and turn it into a research doc? I think I will do just that shortly. The current list that I have includes:

Adaptive Planning
Analytixondemand
Birst
BlinkLogic
Cloud9Analytics
Gooddata
HostAnalytics
LiteBI
Oco
PivotLink
RapidMetrix
SuccessMetrics
TelemeTree
Youcalc

Am I missing anyone?

Process Data Management: Like Your Brain And Your Heart, BPM and MDM Can’t Survive Independently

By Clay Richardson & Rob Karel

 It's an age old question:  Which came first:  the data or the process?  Okay, not an age old question, but an interesting one to ponder nonetheless!  Of course the answer depends on who you ask.  Ask most Business Process pros and the reply:  "Process existed even before computers. In fact I saw on The Discovery Channel where they found a process map in King Tut's tomb. No joke..."  Ask a Data Management pro and you're likely to hear: "Data was invented by the Pharaohs - you ever hear of a little thing called 'numbers'?"  

In upcoming research we tackle this vexing question and conclude that process and data are of course inseparable (shocking we know).  Unfortunately, most Business Process pros, Data Management pros, and vendors we interviewed couldn't see the forest for the trees.  In fact, in our recent MDM Survey, only 11% of respondents said that their MDM and BPM initiatives share the same cost center and team members that work together on a daily basis to develop solutions for the business.

In many ways, master data management (MDM) and business process management (BPM) represent two different sides of the same coin when it comes to business optimization and transformation.  So, why is there so little collaboration and interaction between Business Process and Data Management professionals?  There's plenty of blame to go around:

• Business Process pros don't find data sexy. While most process improvement initiatives pay lip service to data quality, only a small handful take data seriously and incorporate data modeling and data mapping into upfront discovery efforts. BPM teams typically focus on process first, user experience second, and data is an afterthought - no one ever asks if the data is clean until it’s too late!
  
• Data Management pros value clean data over process context.  "If we create clean data, then they will come" is often the mantra of Data Management pros. Unfortunately, the person coming is the CFO - asking "why are we spending so much money on something when we can’t articulate its business impact?"  In other words, Data Management pros often fail to provide cross-enterprise context for MDM.
  
• BPM Suite vendors conveniently MIA on the issue.  Some argue that SOA is the silver bullet to bridge the gap between MDM and BPM initiatives.  While SOA can definitely address reusability and architecture issues, many BPM teams come to realize the importance of master data too late into their initiatives.  While BPM suite vendors provide basic tools to define process variables, only a few (Lombardi/Siperian, Appian, and Polymita) accept some responsibility for keeping process data in sync with master data sources.
  
• MDM vendors focused on data governance and stewardship only.  The good news is these MDM vendors do in fact recognize that business process plays a major role in their success. The bad news is they are only actively addressing one side of the problem: the business processes and stewardship roles that govern the administrative workflows within the MDM tools themselves. The MDM vendors are not actively influencing the business processes that consume and depend upon the master data these vendors provide.

Although process and data professionals may not see it, like your brain and your heart, BPM and MDM are interconnected and one cannot survive for long without the other.  Our upcoming research, "Warning: Don't Assume Your Business Processes Use Master Data," highlights the fact that process improvement initiatives face a vicious cycle of deterioration and decline if master data issues are not addressed from the outset. And MDM initiatives face an uphill battle and certain extinction if they're not connected to cross-cutting business processes that feed and consume master data from different upstream and downstream activities.

Earlier this year, Forrester highlighted the emergence of "process data management" in our BPM Tech Radar report.  This new category in the business process landscape organically merges BPM and MDM disciplines and capabilities to provide the enterprise with one version of "process and data truth".  

Over the next several months, we will publish a body of research that scopes the process data management challenge, provides best practice for combining process improvement and data management initiatives, and presents case studies highlighting teams that have successfully integrated these two critical capabilities.

Why It Matters?

Call it what you like, the consolidation of process improvement and data improvement is inevitable.  Organizations looking to minimize operational risk and boost adoption on BPM initiatives must incorporate data modeling and data management activities for cross-functional process improvement projects.  Additionally, organizations looking to increase the visibility and importance of master data must identify and harness the most critical business processes that generate and consume mater data.  These are just starting points for bringing process and data closer together.  Most importantly, senior management must foster collaboration and provide cross-training between these two siloed disciplines to begin this long overdue paradigm shift...

What’s Your Take?

We want to hear from you.  Let us know whether you think there is a real need to synchronize MDM and BPM activities? Do you think BPM vendors should accept some responsibility for keeping process data synchronized with master data sources? Alternatively, do you think MDM vendors should be doing more to educate the marketplace? Also, let us know if you have a process data management success story, or lesson learned that you want to share.  Post your thoughts in the comment section or feel free to shoot us a quick e-mail at crichardson@forrester.com or rkarel@forrester.com.

HP and Informatica’s Expanded Relationship: Portent of Bigger Deals to Come?

By Rob Karel and James Kobielus

Strategic partnerships are a dime a dozen in the data management industry. Vendors announce partnership deals so often that it’s a challenge to distinguish the mundane from the noteworthy. Case in point: Earlier this week, HP and Informatica entered into a partnership that is not entirely new—after all, their relationship long predates this announcement—but could be the start of  a relationship of considerable strategic importance to both partners.

What the two vendors announced was a multipronged joint approach to the data integration (DI) and data warehousing (DW) markets. First, they announced that HP “will sell and deliver a set of data integration solutions with embedded Informatica technology,” including the latter’s flagship data integration, data quality (DQ), and identity resolution software solutions. Also, the partners plan to develop solutions that combine Informatica’s software with HP Neoview, a high-end, massively parallel data warehousing (DW) appliance. Furthermore, HP and Informatica will develop horizontal and vertical solution accelerators through industry-specific data models drawn from their respective consulting ecosystems, with principal reliance on HP’s BI consulting organization, the foundation of which came from HP’s 2006 acquisition of Knightsbridge Consulting.

It’s an important deal for Informatica, but not one that will appreciably improve its already strong positioning in the data integration and data quality markets. HP simply represents another major systems integration (SI) partnership for Informatica, which already does considerable business with such SI powerhouses as Accenture and Wipro. Informatica will leverage its expanded HP relationship to open doors into new large, global accounts. Likewise, Informatica will benefit from any future sales of HP’s Neoview that incorporate pre-bundled, pre-optimized data integration software. So, clearly, there’s a lot of upside, but minimal downside for Informatica.

For HP, however, the deal may offer very strong upside potential, but also comes with considerable downside risk.

On the positive, the deal will enable HP to position Neoview—with integrated Informatica software and HP/Informatica-developed industry data models—as a more complete analytic solution platform. In this way, HP will be better able to sell Neoview into an increasingly solution-oriented DW appliance market, against such rivals as the forthcoming IBM Smart Analytics Systems.

These future HP-Informatica solution offerings will add more credibility for HP to address consulting engagements with deeper data integration and data quality complexity--especially in accounts that have already short-listed or procured Informatica. The future Neoview-based joint solutions will also provide a credible data management platform for HP’s global consulting business to build custom DW, MDM, and other analytics-intensive applications for clients.

Furthermore, the HP-Informatica alliance represents a new lease on life for Neoview, which, though it has achieved respectable adoption in the three years since it was released, has not been a rousing success in the DW market. The vendors’ joint development deal holds out the hope that Neoview will evolve into a more feature-complete data management and analytics platform. This is the best hope for Neoview, which, until the Informatica announcement, appeared to be drifting as a solution platform in HP’s increasingly professional-service-centric data management strategy.

But the downside of the Informatica deal for HP is undeniable. For starters, the HP-Informatica alliance is likely to elicit a wall of indifference from prospective SI customers if it can’t effectively differentiate itself from a multitude of other partnerships between DW platform vendors, SI firms, and data integration solution providers. This is a non-exclusive deal, after all, which sows doubt as to whether HP is truly committed to the relationship on a high strategic level. Yes, Neoview has a new lease on life, but that could quickly fade away if the promised solutions—incorporating Informatica software—take forever to come to market.

The deal also highlights the fact that the HP-Informatica offering will likely, barring addition to other partners in the near future, fail to deliver a joint solution with the functional breadth of the forthcoming IBM Smart Analytics System offerings. Obvious components missing from a joint HP-Informatica software portfolio include MDM, BI, business performance solutions (BPS), predictive analytics and data mining (PA/DM), and text analytics.

One could even make a case for the Informatica deal representing HP’s last brave attempt to mount a credible data management software strategy. It highlights the question in many industry observers’ minds:  Does HP want to be in the business of data management software? For years, analysts, soothsayers, and other fun-loving rumor-mongers have questioned if and when HP will evolve its data management strategy from primarily a services-driven model—plus an also-ran of a DW appliance product (Neoview)--to a full-fledged software and services platform vendor that could compete head to head with the likes of IBM, Oracle, SAP, and others. However, the Informatica deal does not take HP to that critical next strategic step.

If HP is truly serious about becoming a full-fledged data management software provider, now is the time to make strategic acquisitions in the DI, DQ, MDM, BI, PA/DM, and other key solution markets. With over $13 billion cash in HP’s war chest as of July 31, 2009 (according to its Q3 2009 financial statement), HP certainly can afford such acquisitions. It could even afford to acquire Informatica, which, however, wouldn’t come cheap.

But what HP cannot afford is to let the window of opportunity slam shut. It cannot afford to make any acquisitions of this magnitude, and then fail to capture and grow a very significant share of the data integration, data quality, data warehousing, and BI software markets. 

So is the partnership with Informatica a “proof of concept” for future acquisition or is it simply HP BIS’s answer: ‘We are a services business and we will leave software to our partners’?    We don’t know if this is the first step of a broader software strategy, or an admission that HP does not see itself as a major data management solutions provider.  If it’s the latter, Forrester would not be surprised to see Neoview either spun off or sold to a full-fledged data management solution provider (perhaps Informatica, though SAP would also be a potential suitor) that needs a scalable, real-time DW appliance platform.

We don’t expect Neoview to die on the vine of HP’s sometimes unclear data management solution focus. But we would hate to see this impressively engineered DW appliance platform dangle in the shifting winds of HP’s strategy for the business intelligence market.