Got a quick update today from the new co-CEOs of SAP – Bill McDermott and Jim Hagemann Snabe.

Jim focused on their innovation strategy – making significant steps into on-demand business applications, aiming to support a hybrid approach allowing customers to mix on-demand and on-premise software. In addition they aim to increase support for running the applications on new mobile devices – this, of course, requires a separation of decision-making business logic from front-end logic. Hopefully this will see SAP investing more in its business rules capabilities (described under the SAP tag on the blog). All of this requires that processes and MDM can be orchestrated across this increasingly complex environment, even when non-SAP application components are involved. They are also rolling out more agile development methodologies (like those being described in the new SAP BPM book on which I am working with various other SAP folks).

Lots of interesting questions got asked and here are some of the responses that seem most interesting from a decisioning perspective:

• In memory analytics will change the way high end analytics are deployed. Focused on a variety of partners to bring new approaches, new techniques into high-end analytic space. Still expect to work with SPSS in this regard but also looking for new technologies that take advantage from the ground up of in-memory analytics.
• Business by Design, SAP’s easy to configure on-demand (SaaS) offering, is coming out this summer. Will be interesting to see the extent to which business rules are used to make it configurable.
• Interesting challenge for a company like SAP is that different product lines, different deployment options have a different cycle. On premise software, for instance, cannot be updated too often as customers don’t want to constantly re-install. On-demand software, however, gets updated more often and on-device software is driven by a very dynamic consumer technology market. This is a large scale change, ensuring that different parts of the company can operate on the right timescale while remaining part of the same company. Personally I think that rules-based decision evolution is a key element of this and I hope to see some sign that SAP thinks this way too.
• Asked about mergers and acquisitions – the point was made that Oracle has been much more aggressive – Bill and Jim acknowledged that they are going to be more aggressive going forward while remaining focused on innovation and an integrated, coherent business application suite rather than generating growth through acquisitions. As more and more established customers have been acquired (up to the acquisition of Chordiant by Pegasystems today) this is an interesting topic – increasingly SAP will have no option to grow through acquisitions but this may suit their corporate culture better anyway.
• Asked about the trend (Oracle, IBM) to mix hardware and software they replied that they see a heterogeneous world that is in constant flux – customers never own one vendor’s complete set – so being good at working in this environment is key. Eliminating layers using hardware is good but they see working with multiple partners not owning their own. This requires collaboration with a mix of hardware partners rather than acquiring and integrating their own hardware. Customers don’t want vendor lock-in, they buy a business outcome not a “stack”.
• SAP is not worried about the ownership of Java by Oracle – they see a vibrant, open, multi-company ecosystem around Java and don’t expect Oracle’s ownership to impact this. Interestingly they made the point that programming languages come and go and that Java is not therefore the be-all and end-all.

Interesting conversation, nice degree of openness and responsiveness – much improved over SAP TechEd where avoiding questions was the order of the day.

The news today is that Pegasystems (rules-based business process management) is acquiring Chordiant (decision-centric CRM). This is interesting news as it merges a company (Chordiant) with a very decision-centric/decision services separate from process mindset with one (Pega) that has mixed rules and process together much more.

Chordiant have been one of my companies to watch for a while, with a great decisioning platform. Their clear separation of decisioning, their support for rules and analytics in combination, their strong adaptive analytics engine for self-learning models, their recent integration of real-time conversations with decision management and their powerful business simulation tool (Visual Business Director, see below) are enough to put them at or very near the top of the decisioning vendors.

Pega, of course, have been best known for their business process management focus. They have always addressed this from a rules-centric perspective and we have had some active disagreements about the role of decision services and the value of a clear separate of processes and decisions (see this post and comment thread, for instance). Nevertheless we agree on the power of business rules to drive more agile and smarter systems and Pega has been one of the rules vendors active in supporting PMML (Predictive Model Markup Language) to allow the integration of business analytics with business rules.

The potential of this merger is real. Clearly the merged company will be larger, important as the big players (IBM, SAP especially) get more serious and rules and decisioning. Chordiant’s decision management and simulation components are, in my estimation, better than Pega’s for specific purposes but not as general purpose. An intelligent combination of the two is therefore potentially very powerful. In particular, bringing Chordiant’s adaptive analytics and simualtion capabilities to the broader rules-based platform that Pega offers could be great. In addition both are very focused on CRM or at least on customer treatment decisioning, and this should help give the merged company a clear focus.

The risk, of course, is that the fairly serious difference of perspective between decision-centric / decision management on one hand and rules-driven BPM on the other will derail the technical integration or cause the merged company to merge its operations without merging its products. Either will ensure that the talented people behind the products will not stay and that would be a pity. The merged company must figure this out and make some clear statements on product direction and positioning in this respect – though I appreciate that this can’t be done right now, as it must wait for regulatory clearance etc. There is a lot of overlap in technology. This could be good – giving the merged company enough of a common vocabulary to build a powerful solution – or bad, resulting in lots of infighting about which version to keep.

Check out these posts on Chordiant for more details. The folks at Pega have never seemed to want me to blog about them so I don’t have anything about them on the blog. Hopefully this will change…

• First Look – Chordiant Decision Management
• First Look – Chordiant Recommendation Advisor
• First Look – Chordiant’s Visual Business Director
• Chordiant Decision Management Update

Disclosure: Chordiant was a customer of mine in 2008/2009

I just heard from a colleague that you can check out Oracle’s Data Mining tools on the amazon.com compute cloud.  The Oracle Data Mining development team has set up an instance for prospective customers who want to try the in-database data mining algorithms via SQL/Java APIs or the Oracle Data Miner user interface. You can launch an Oracle Data Mining Amazon Machine Image (AMI) directly through Amazon Web Services (AWS) and your only cost is the standard Amazon EC2 charges.

To get started go to the Amazon Cloud with Oracle Data Mining or click here for a step-by-step visual guide. There’s more on the Oracle Data Mining blog – http://blogs.oracle.com/datamining/

eBureau is a predictive scoring and information service provider founded in 2004, focused on technology for very rapid model development and deployment. Using their own purpose-built modeling software, a small group of modelers developed 900 predictive models in 2009 alone. The company has been applying this capability for real-time and interactive marketing like contact centers, consumer lead generation, risk and fraud management, and display ad targeting. Not only can the models be built fast, they can be deployed in the cloud quickly for real-time scoring applications. Typical transactions take less than a second round trip. They have found that this approach works in a variety of industries like education, financial services, automotive, telecom etc. For instance, online universities are using the eBureau solution to predict which consumer leads will apply, enroll, and stay enrolled.

At its core, eBureau is focused on new customer acquisition whether helping clients understand payment risk or propensity to respond to an offer. They take historical performance data (leads, who converted, how valuable they were) and data from 50 other sources before running their predictive modeling technology. eBureau develops a predictive score (for fraud, probability to convert, payment risk, etc.) that can then be used to:

• Improve the online marketer’s cost-per-lead advertising decisions e.g. buy or no-buy decision on leads or right-pricing based on the score
• Improve contact center conversion efforts e.g. offer path management or how to route calls most effectively based on a consumer’s profile
• Improve display ad decisions e.g. find an audience that looks like your best customers and target the right creative at the right time using predictive models

eBureau has some 50 databases enabling them to cross-reference data like addresses and phones, customer purchase data like aggregated catalog sales data, demographic data, aggregated financial data like Zip+4 household wealth, and interactive data like social graphs and e-mail addresses. These 50 databases add up to a combined 300Bn records and 200TB of data covering 99% of US adult consumers. All of this is available for every modeling project – some 50,000 attributes applied to every problem. Obviously this has to be integrated in terms of identity matching and in terms of managing data granularity to ensure summary data can be used as well as individual data. This is all done in-house in a highly secure data center in St. Cloud, Minnesota just north of Minneapolis.

One of eBureau’s education clients wanted to predict which leads would result in enrolled students. Over a period of 6 months, this university purchased 537,000 consumer leads which ultimately resulted in 6,000 enrolled students, representing a 1% conversion rate. eBureau found some 120 attributes that were predictive across demographic, property, purchase history, etc.  The average cost-per-enrollment was $3,100 across the whole portfolio but by focusing on the highest scoring segments they were able to reduce this to $2,300 saving them tens of millions of dollars; Classic predictive analytic segmentation.

Depending on sales cycles, it can take several months to know if a lead converts or not, but once the model is built eBureau clients get immediate feedback on the quality of a given lead. This allows eBureau clients to rapidly assess a new lead source, understand quality across a portfolio of lead sources, or simply optimize internal campaigns and creative.

Another example is a company using it to segment inbound leads from ads bought on the spot market (where they don’t control the time to run the ads). Using only the prospect’s phone number, this client used a simple green/yellow/red score to prioritize the incoming calls. When volumes are low, the red (low conversion) ones get handled but when things get busy (because lots of ads are running) only the green and yellow get handled and the green’s (top few segments) get prioritized and routed to the appropriate sales people.

Direct marketers have been doing this for years with direct mail lists. For example, credit card marketers don’t send letters to everyone, they pre-screen to find the best potential audience for the offer. Online display ads can be managed similarly, but using predictive models in real-time to understand who it is worth showing a display ad to. eBureau uses their data to build “look-alike” models for a company’s best customers – online consumers who look (statistically) like your best customers. eBureau protects privacy by placing anonymous cookies allowing advertisers to identify a high-propensity “look-alike” prospect and serve exactly the right display ad in real-time without the advertiser ever knowing who the target is or anything about them. This is a nice example of a predictive model keeping private data private while letting people use that data effectively.

eBureau has also invested in web-based reporting and analysis tools to help their clients understand the impact of models such as score trends by source over time. As with many predictive model and scoring solutions, the education of marketplace and helping new clients with organizational change implications are critical to success.

Personally I think this kind of hosted decisioning/analytics is a great way for many companies to get started with analytics, with using external data sources to enrich their own and to apply analytics in their operational systems.

Optimization is a mathematical process for finding the best decision for a given business problem – usually highest profit, lowest cost given a set of constraints. Involve applying an algorithm to data, decision variables, constraints and an objective function. In financial services and insurance optimization is still fairly new (unlike, say, supply chain) but the complex regulatory environment and tradeoffs between risk and reward are ideally suited to it. It also helps with champion/challenger and experimental design. Combining optimization with business rules, and with predictive analytics, is a growth area and I got an update from FICO on how they see these technologies working together.

Business rules and optimization allow you to use rules for flexibility and agility and optimization to get to “best” faster. FICO provides  FICO Decision Optimizer as a packaged solution that solves specific banking optimization problems, and FICO Xpress-Mosel as an optimization modeling tool for solving a wide range of industry problems.  Both of these products can be combined with  FICO Blaze Advisor to leverage business rules management.   With these tools, FICO can offer a number of ways to integrate rules and optimization:

• Use Blaze Advisor to deploy an optimized strategy tree created using Decision Optimizer or Xpress optimization results
• Invoke a configured Xpress-Mosel optimization model from a Blaze Advisor decision service (Mosel is the Xpress modeling and programming language)
• Use rules to configure the parameters of the Xpress-Mosel model and then execute it
• Blaze Advisor could provide the input data to Xpress-Mosel Models
• A core model in Xpress-Mosel with Blaze Advisor providing pieces using rules execution t assemble the pieces
• A skeleton model in Xpress-Mosel with Blaze Advisor providing the Xpress-Mosel Code and data using rules to assemble it

For example in debt consolidation use inputs and preferences to find the best payoff loan. The customer has several exiting debts, at different rates of interest, and wants to optimize for payment or pay off period etc. The optimization model does the tradeoff while the rules manage the eligibility of the customer for specific products that might be available for the pay off choice. The optimization engine gets to pick only from the eligible products (the rules for this are already being used elsewhere in most companies so this allows the rules to be reused not repeated for the optimization problem). This can also use predictive models e.g. to predictive price sensitivity (the model is used to calculate a value that is input to the model).

Business rules, optimization and predictive analytics are also being used in FICO Retail Action Manager to optimize marketing spend. Uses the business rules management system to ensure consistent and targeted messages across channels, predictive models to predict who will buy what and an optimization model to pick the optimal offer and channel given the constraints you have.

Retail space planning was another solution that included rules and optimization. Retailer was trying to maximize profitability in the “planograms” or shelf layouts that were being developed. This used predictive analytics to predict how likely customers might be to pick more expensive products if they are positioned correctly, rules for defining the constraints like competing products or store layout consistency as well as contractual requirements from suppliers. Optimization handled the tradeoffs.

I see more and more use cases for business rules, optimization and predictive analytics in combination. The move to considering these complementary technologies as a platform for decisioning is welcome.

TRIAD 8.5 has just been released and is the latest version of FICO’s combined account manager and customer manager platform for financial services companies (its focus is on accounts/customers where credit risk is a critical issue). I got an update recently focused on Decision Graph. Decision Graph is  one of the new capabilities. Decision Graph  is a new strategy management and visualization tool – replacing or enhancing a decision-tree centric approach.

Because strategies in Decision Graph can be represented as graphs, they allow the reuse of elements – sub trees – preventing the growth of strategies into large, unwieldy decision tree structures. The technology has 7 patents pending and was described by one of the inventors of CART (Professor Jerome H. Friedman) as the best strategy visualization tool he has ever seen. I blogged about the technology while it was an R&D capability and it has now been released commercially. The new capability has a number of key features:

• Users can visualize a strategy as a Directed Acyclical Graph (de-duplicates sub-trees that are exact) to simplify the strategy and more clearly see exactly how many distinct kinds of outcomes there are.
• Users can visualize it as an Exception-Directed Acycical Graph (EDAG) that takes the most complex logic and makes that the exception path – eliminating complexity by emphasizing the simpler paths.
• A strategy can be viewed through its component action graphs that allow you to focus on the logic that lead to a particular action – each one only focuses on the logic for a specific action. When a typical decision tree for credit management might have 1000s of nodes and actions being able to see the sequence of decision points building up to a specific action makes it much easier to manage the complexity.
• Strategy slicing (specifying that they want to see only the logic for a particular segment/set of conditions) is key to explicability and compliance. This allows a user to focus, for instance, on how sub-prime customers with a particular range of credit scores are being treated.
• Analysis of the outcome distribution for a node is also interesting – users can click on a node and see how the population who reach that node will break down across the various actions. Users can see how data “flows” through the strategy.
• One of the best features is the ability to compare two strategies based on the actual logic not the structure. Comparisons show what is the same and what’s different between two strategies (see figure). Unlike decision tree comparisons, the way the strategy has been documented does not affect the comparison. This allows Champion/Challenger comparisons, different divisions or channels’ approach to customer treatment and much more to be compared. A key use case for this is in mergers and acquisitions where knowing the differences between customer treatment between the two companies is critical. Users can also use this for comparing treatments across product types, or for showing regulators exactly how a given strategy meets compliance requirements.

The reduction in complexity is the most critical outcome of this collection of features. They improve the ability of the software to spot errors automatically and any of the visualizations can be used to build and edit a strategy (all share an underlying model). Levels can also be reordered automatically to simplify the strategy (improving execution and understandability), without changing the underlying logic of the strategy.

Many financial services companies have found that their strategies have become so unwieldy that editing them is too hard. Using this new feature to reorder and revisualize strategies can result in dramatic reductions in complexity of up to 80% (though 25% is more typical – still very useful). The key advantages for customers are that it allows for more rapid change (improving agility), increased likelihood of finding improvements, lower errors due to reduced complexity and easier explanations/execution. FICO estimates this could be worth $1M per year on a 2M account portfolio.

You can see a FICO Tech Talk on TRIAD 8.5 and Decision Graph on YouTube.

I got a briefing from Quantivo recently. This is a company focused on behavioral analytics – uncovering patterns within the mountains of customer data that companies have – web analytics and point of sale data for instance. They help companies find these patterns, find the insight that they are not seeing with their current tools, and then help them make better decisions. Quantivo started with Retail (market basket analysis and promotion analysis) then expanded beyond that, including web analytics, and acquired new customers throughout 2009. The focus now is on their latest product release, Quantivo 4, and a strategic partnership with Webtrends (who now resell Quantivo). Quantivo has signed some good retail, B2B, marketing and insurance customers, including OSH and Cisco WebEx. Quantivo describe themselves as offering advanced analytics at scale in the cloud.

Quantivo sees companies trying to find who did what, when and why so they can target customers and promote more effectively. Companies want access to their data and effective answers without having to go through the IT department. The need is to democratize access to data and the answers hidden in data. People have a thirst for answers that is not being met by the BI infrastructure IT departments have implemented. In particular there is a gap between analytics and action – data is too far from decision makers who anyway can’t use the analytic tools that are available. Quantivo has tried to re-think the current data/ETL/Data Warehouse/BI/Data mining tool stack and do this re-thinking in the cloud to take advantage of the flexibility and elastic computing power available that way.

Their target user is a business analyst who wants to know things like who purchased movies and games together or what coupon users bought the 10 days following their use of the coupon, which campaign drove high-value repeat customers etc. A marketing analyst, for instance, trying to figure out what works and what does not. These are “advanced” analytics not because the questions are conceptually difficult to ask or because the representation of the answer is complex but because they are hard to answer using classic OLAP/reporting tools.

Quantivo 4 has focused in a few key areas:

• Dynamic Behavioral Segmentation
  Context filtering and context-specific queries (over a web session, lifetime of a customer, product range etc), multi-attribute segmentation and segmentation comparison
• Drag and drop web UI to make the solution accessible to business analysts
• Instant export so can load into some tool to take action using downstream applications

The web environment allows business analysts to create and manage worksheets (which can be shared between users). These worksheets can be built using drag and drop feature from lists of dimensions and measures in an OLAP-like way. Performance is good, with large numbers of records being processed quickly and filters can be easily added to restrict the data and see results. Within the results users can start to select elements (one department, say) and make them a comparison target. This allows them to see, for instance, what else people who bought from a specific department purchased at the same time. Or what people bought in the week following a purchase from that department.

Users can drill down, navigate around etc in an easy to use and pretty responsive interface. This is the kind of analysis most people would do in data mining or high-end analytic tools but made available in a very easy to use end-user analyst interface. These worksheets are live and updated when new data is uploaded and they can be shared across users. Customers’ data is uploaded to Quantivo, which is hosted on Amazon EC2.

At any point the user can take the population (of people who might be a good target for instance for an offer) and export to a marketing application etc. Quantivo makes it easy to access the result of a worksheet programmatically and they are working on more advanced APIs also.

Wolf Frameworks is a USA/India PaaS company started in 2006 as a pure play cloud computing platform. They have a front end (AJAX) using XML to communicate to a .NET backend on C#. They have about 3,000 plus people designing software using the platform and have about 13 plus solution providers covering 7 countries. They have customers across a wide range of industries with lots of smaller companies and ISVs. They are primarily working with partners who develop and support software and services. As with most PaaS vendors, they offer a no capex, minimum op-ex approach.

They position their product as a rapid application development platform – code free with no scripting language. They offer a nice hybrid model in that cloud applications can be developed and then moved to a private cloud. They offer their standard On Demand deployment using iweb (http://iweb.com/) and offer premium customers their choice of deployment with Rackspace, Amazon.com, etc.

They offer Integration, Billing, Presentation, Application Development and Database layers – all code free. They aim to allow software development to be 70% faster and 50% cheaper. In their approach, business analysis and application design are the critical pieces – coding is not required. They argue that this changes the equation from traditional development – 25% design, 50% develop and code, 15%test and 10% deploy – to 25% design, no development/code, 7% (functional) testing and no deploy. As a result they get to be 70% faster. They also focus on the role of a domain expert/business analysts. Finally their XML/web-services architecture also makes it easy to develop components as part of mashups and to link with other SaaS providers or any web portal as embeddable functional widgets. Some of their solutions today connect with SAP, Quickbooks, Microsoft suite, etc

Unique value propositions:

• Business rules as a service
  More on this later
• Customizable UI
  Data and UI rendering separate so can be customized
• Minimize lock-in
  Data is not locked, you can click to extract and use data, design is stored in Wolf but design extraction is a supported process that produces XML, hosting is open as can move to your own environment
• Standard software
  AJAX, XML, IIS, .NET etc.

They also offer support for multiple devices and reporting. They talk in terms of having data at the center of the solution with a wrapper of business rules around that and definitions for input/output around that (mobile devices, web etc).

From a design perspective they offer what seems like a nice UI with widgets and graphs, tabs, basic forms, printing, access to Excel, CSV, RSS Feeds, etc all built in. The Entity Designer lets you define the information you are managing in the application. Entities can be grouped for management and each has relations and an edit screen. Designing the screen, as usual in PaaS, defines the entity. Fields can be grouped into sub-tabs and repeating groups of fields are supported. You can define multiple screens & tabs for an entity. A Navigation Designer allows you to define the structure of the user navigation in an application and the Reporting Designer does exactly what you would expect.

From the business rules perspective they have made some good progress. All pieces of business logic in the system are identified as business rules. Each is associated with an entity defined in the system for management. Rules can also be linked to events (such as saves or updates to the entity) or to explicit service calls. Each rule can be as simple as a single If.. Then rule or a long procedural script. The logic is easy to specify with a nice point and click interface, easy access to the data involved etc. Rules can invoke other rules within looping constructs or in a simple sequence and the invoked rules have direct access to the data available within the calling rule.

While Wolf does not offer a “true” business rules management environment, what they have actually let’s you get pretty close:

• You can write lots of simple rules
• You can group and manage these rules using the entities (all claims rules, for instance, could be associated with the claims entity)
• You could write a rule for each ruleset or group of rules that invoked the rules in that ruleset in order
• You could write a rule for each decision that simply invoked all the rulesets involved in that decision
• Rules and rule sets can be reused across multiple decisions in this way

For instance, one could write a set of business rules to validate a claim. You could then write an additional rule representing the ruleset “Validate Claim” and another for the decision “Approve Claim”. The Approve Claim rule can be triggered from a user action (so that the user could decide when to call it in the UI) and can then call Validate Claim before executing the rules / rule sets for approving the claim. Because rules can be exposed as services, the Validate Claim and Approve Claim rules could be exposed as Decision Services for use elsewhere in your system.

Obviously this only supports sequential execution and there is no support for decision tables or trees or other higher level rule representations, so it’s not perfect. But, for a PaaS solution, this is a pretty good start. You COULD just use it to write code like any other PaaS but you could also use it with a decision management mindset, which is cool.

Analytics in Wolf is focused on reporting and dashboards rather than decisioning so there is no support today for data mining or predictive analytics. You would need to take the data out and into a cloud-based data mining solution like Data Applied or Clario

It will be interesting to see where Wolf take their rules capability in the coming months.

SAS Real-Time Decision Manager (RDM) is designed for inbound communications, complementing outbound communication solutions. It aims at real-time delivery of decisions and recommendations during a customer interaction to optimize that interaction to improve revenue, growth and retention. For example, in retail banking, a customer might come in with a new job with very different income and needs financial guidance – this must be real-time as the bank did not have this information earlier.

SAS Real-Time Decision Manager supports in person, call center, website, IVR, ATM/POS, mobile channels and is tightly integrated with SAS Marketing Automation – it shares a UI, data model, contact and response history etc. SAS Real-Time Decision Manager leverages data from customer profiles, historical data, real-time data being captured by the channel, other systems and SAS analytic models. Real-Time Decision Manager is designed to support business users, analytic users and technical administrators.

Real-Time Decision Manager is independent of the channels and plugs into the channel management systems to provide a decision. Real-Time Decision Manager gets web service requests from channel applications and responds by executing a decision flow and returning an answer – it acts, in other words, like a pure decision service. During the decision flow it can reach out to databases, other applications or external web services. It can execute analytic models and conditions and then returns the answer as a web service response.

The decision flow is specified by a business user and has a nice graphical interface common to many SAS applications. The flow can be defined using drag and drop from a palette of components including branches, filters etc. There is a task bar to manage the process and the diagrams can be annotated with notes and images. The flows start with a node for receiving request (the flow is exposed as a web service) and progress through a series of steps. These steps or activities can:

• Use models (developed by SAS Enterprise Miner, for instance) to score the customer – the business user does not see the details of the model or need to drill into the details.
• Execute a custom process (any web service call or SAS scripts) – provides a means to extend the capabilities of the solution with minimal development.  These could be rules-based servicesto handle things like eligibility
• Access databases with SQL queries that are preconfigured so the user does not need to create queries themselves
• Filter by applying conditions – essentially rules that are highly constrained in terms of action (only yes/no in terms of further evaluating the customer to determine a specific communication ) but allow conditions to be written against all the data being managed.
• Branch based on data values
• Assign transactions to a reporting cell for later analysis

The business user can bring up a test interface and specify input parameters to see what result you get. You can save and manage the tests you want to use and re-run them. The nodes executed and any errors are reported as part of this. Impact analysis can be set up using standard SAS BI/reporting capabilities to act on the response data stored by the system (though there is not an out of the box impact analysis report).

Analytic users can develop models using normal SAS tools and then register models with SAS Model Manager. This makes the model available to the business user working on decision processes. The model can be updated and versioned independently of the decision flow, allowing constant updating of the models for improved performance.

Most customers are focused, obviously, on marketing and many use it in conjunction with SAS’ outbound campaign management component, SAS Marketing Automation. Some however use it for customer service and other customer-related decisions. Competitively this goes up against Chordiant Decision Manager, Unica, Oracle RTD, Convergys etc. SAS sees their product as more visual, easier for a business user to use, and clearly it has much tighter integration with SAS’ modeling tools which is a big plus for companies with a large investment there. They also feel their inbound/outbound integration is stronger. SAS believes that models developed based on knowledge of the domain and the data are still more effective than adaptive analytic models, and this is another difference from their competitors many of whom are very focused on these kinds of analytics.  However SAS also knows that expert analysts can be a rare resource in many organizations, so it is developing a new solution called Rapid Predictive Modeling to enable non-modelers to develop expert models rapidly. I am getting a briefing on this soon and will blog about it when I do.

I got a chance to catch up with Aha! recently. Aha! is based in the Denver Tech Center and was founded back in 2006. Aha!’s premise is that it is now possible to build analytics into a platform and to focus on how to operationalize predictive models and deliver analytics within business processes. Initial customers are in healthcare, telecom, travel and transportation. Their aim is to deliver a complete analytics management system. The pain points of traditional BI solutions that they address are: limited use and access (big focus on self-service), long time to value (SaaS platform or rapid start up with minimal IT), rear-view focus (predictive analytics), piles of data (model-based analytics) and Excel/scattered data (single network). They focus on being “dynamic and aligned” and focused on business users.

Their market is the $2-$3Bn “business embedded analytics” that is part of the overall $26Bn global analytics market. In particular, they provide non power analytic users with access to analytics without having to obtain specialized skills. They see themselves helping the vast majority of business users who don’t use analytics today – the people that dominate the operations of a company like marketing managers, sales managers, customer care managers, product managers, marketers, engineers, and operations specialists. Financials matter to these folks but they don’t dominate the way they do with “traditional” financial department analytics users.

Aha! sells direct as a SaaS offering (setup fees and subscription), offers model development and data discovery services and licenses through OEM/SI partners. Partners are typically domain experts and vertically focused.

Some example customers include: a telecom company using analytics to handle the ROI of proactively building out a fiber network and to optimize sales and marketing to light up this fiber; a telecom handling customer retention, product segmentation and customer experience satisfaction; a healthcare company working on customer retention and acquisition.

Their offering (Axel) is a SaaS multi-tenant, multi-hosted system. It is designed to bring models into the business process – business process based models – make the analytics actionable and close the loop between analytics to actions. The whole thing is based on KPIs and designed to help companies actually act on their strategy, using a KPI model that runs from head office strategy to the front line. The platform has 5 core elements:

• Language
  The Aha! Expressions analytic model definition language that allows business analysts to build the models
• Dynamic services
  Secure, multi-tenant, forecasting, simulation and optimization
• Visualization
  Self-service, near real-time and model driven
• Data Engine
  Profiler, designer, ETL, Smart Pub/Sub
• Extensions
  Support for third parties to extend and integrate the platform

The basic process looks like this (for a healthcare member retention example):

1. Customer profile, billing, survey and claims data is used to create a model data file
2. Predictive models are developed based on this data
3. Customers are scored using these models
4. Contact and campaign management define available actions based on these scores
5. KPI-based models are developed using the same data
6. Collaborative analytics link all this together to support decision making and drive ROI

The target for this customer was to reduce churn. They were up and running in 60 days, improved retention by 7.5% (v target of 3%), improved new member retention by 9%. NPV of $43M in a single enrollment period and an all-in ROI of 2447%. This was recognized at the World Health Congress as a top example of using predictive analytics to drive member retention and satisfaction. Users ranged from call center operations to VP level executives.

The model data was used to create retention or churn scores for each customer that were loaded into the operational system in batch. These scores can be updated regularly from the model data file and can be calculated live based on intra-day data or, in theory, even during a conversation (using a standard web-services interface). The use of this model is much the same as the use of any other predictive model except that the data is tightly coupled with the KPI hierarchy. Models can be built from and evaluated against the historical data that drives the KPIs, so that users start off with a valid historical base. Axel also provides a stochastic enrichment engine ( Monte Carlo simulation with category selection, probability, and triangular distributions) that supports PMML, allowing models built outside to be imported using PMML. Models can also be generated via an Microsoft Excel Template.

Aha! is driven by a KPI model hierarchy. In the case of this healthcare company it was Retention Campaign (Strategic), then the health plan a member was in (Tactical) then events within a member lifecycle (Operational). This drives how the data is viewed and KPIs – in this case customer retention measures of various kinds – are tracked against this hierarchy. So, for instance, each KPI could be viewed with respect to a specific member lifecycle step, a particular plan or a particular campaign.

Each KPI has a calculation defined for it and are calculated dynamically. In addition to mathematical calculations, the Expressions language also provides addition functionality that supports the calculation of KPIs based on Year to Date, Quarter to Date, Month to Date, Sum of values for a defined period, Average of values for a defined period, etc.

The interface allows different reference periods to be selected and the KPIs to be viewed within that period along with measures like averages, high/low values for the period, goals etc. For instance, this customer saw a lot of new members were signing up but then being lost. The prediction showed that the trend would clearly exceed their target for such losses and allowed them to see the impact on all their KPIs. This provoked a focus on the reasons for this and they found an external verification service that was needlessly disqualifying people. They had no expectation that this would be a problem and the tool allowed them both to spot it and see the impact on their KPIs quickly enough to take action before the open enrollment period was completed and the opportunity to fix it lost.

The most interesting thing about Aha! for me is the tie to a formal model of KPIs that drive from a high level to an operational level. This allows impact analysis and decision making to be clearly linked to the objectives set at different levels.

For more information on Aha!, you can visit their website at www.ahasoftware.com or download their paper on Business Embedded Analytics.