We help a lot of clients select, install and adopt a Business Rules Management Systems (BRMS). These clients are looking to get automate decision-making with transparency, deliver business control of their critical decision-making logic and establish an ability to drive continuous improvement through simulation and impact analysis. Adopted correctly, these benefits ensure that a BRMS delivers a great ROI.
To maximize this ROI our clients are looking to get the benefits of their BRMS faster, spend less on implementing their BRMS and increase the size of the benefit they see. Here are some tips based on our experience:
Faster
The best way to get benefit from a BRMS faster is to get to a working decision service faster. More than anything, our experience shows this means capturing the requirements for that service faster.
For this we use decision modeling and the Decision Model and Notation (DMN) standard as well as our decision modeling software, DecisionsFirst™ Modeler. Experience is that this can reduce the time to get your decision requirements and business rules right by 5-10x, getting you to an ROI months earlier than traditional rules-first analysis approaches.
Cheaper
Decision modeling also dramatically reduces the amount of re-work by getting the requirements right the first time. This lowers cost too. More importantly, it creates the kinds of rules that business users can maintain themselves, reducing IT costs by eliminating the need for projects to make rule changes. It let’s you take more advantage of simulation tools in your BRMS, reducing the need for and cost of testing.
Small, regular changes also cost less than waiting until there are enough changes to justify a project. And these updates are themselves much cheaper because a decision model makes it easier to tell what change is needed.
Bigger
Bigger ROI comes from using the BRMS on a larger scope, something that getting faster, cheaper projects will help ensure. But it also comes from creating an environment in which the business can truly take advantage of rapid business rules updates – something a BRMS is really good at but that goes unused all too often. The role of a decision model in creating an environment where this kind of rapid iteration is the norm really can’t be overstated – it’s the key.
So, if you want a bigger, faster, cheaper ROI from your BRMS, don’t forget to add decision modeling. Check out our recently updated white paper Maximizing the Value of Business Rules for more. If you’d like our help with selecting or adopting a BRMS, drop us a line.
Decision Management Solutions is growing and looking for a Delivery Manager for its projects.
The Delivery Manager will be an experienced hybrid agile project manager and will be responsible for managing several concurrent, discipline based, high visibility projects using agile, and fixed milestone methods in a fast-paced environment that may cross multiple internal business divisions and services engagements.
Goals
Deliver agile projects that provide exceptional business value to users
Achieve a high level of performance and quality, and
Further the delivery of discipline and supporting methodologies
The team at Machine Learning Week/Predictive Analytics World has announced the schedule for 2021 (virtual conference, May 24-28, 2021) and issued their call for speakers. This is a great conference and will be a great opportunity to present. As always those with case studies and real experience will be particularly welcome!
I will once again be chairing a business-oriented track focused on operationalization of models, business management of machine learning and best practices for extracting real business value from machine learning, AI and predictive analytics. So if you’d like to talk about THOSE issues, I’d really like you to apply! Feel free to reach out to me directly with questions but I encourage you to apply.
Topics you might think about presenting on:
Success stories on how you build analytic models that added real business value
Horror stories on how to build models that don’t add value
Project management approaches to engage the business and IT in analytic projects
What other technology you use besides your favorite ML/analytic workbench and why it helps you get to production
What you’ve learned about hiring, developing and training analytic talent
How your company learns and improves when it comes to machine learning and analytics – communities, wikis etc.
Rollout best (and worst) practices
Experience with ML Ops and other operationalization steps
Plus of course anything around best practices and experience actually building the models is always welcome!
Eric Siegel and I had a great discussion about doing Machine Learning BACKWARDS recently – you can watch the recording below or on our YouTube Channel. Eric, if you don’t know, is the founder of Predictive Analytics World, a leading consultant, and author of “Predictive Analytics“. You can also check out Eric’s new Coursera class.
This discussion was prompted by Eric and I talking about the rate of failure in Machine Learning projects. For instance, one survey said that 85% or more of machine learning projects fail to add business value and that number has gone up, not down, in recent years. Our premise is that the best way to avoid these failures is to do machine learning backwards – to begin with the outcome you want, an improved decision, and work back to the models you need and the data that will let you build them.
At the end we took some questions and one of the questions we got was:
How do you recommend getting senior executives engaged?
First, we said, you need to focus the discussion on the value from deploying a solution not the core technology. This means you might want to avoid using the words “model” or “predictive model” or “machine learning model”. Instead, focus on is exactly which decisions within which large scale operations are going to be improved and to what degree they could potentially be improved. Then you can start to talk about probabilities such as that these people are much more likely to cancel and how these probabilities are going to help make decisions more profitably. After all, you can place customers into at least two very different groups based on those probabilities and treat them accordingly generating differentiation.
I discussed one useful exercise we have done with executives. We start by asking them how they are measured – how they measure their own personal success, which metrics they care about because those metrics drive their bonus. Then we’ll ask them to identify the decisions that get made in the organization that have an impact on those metrics. The first few are always big strategic decisions that the executive team make.
If you keep pushing on it, though, gradually they realize that there are decisions made by all sorts of people in the organization and indeed by bits of software infrastructure that matter to the metric also. And while they trust their own judgment – they don’t need analytics – they are much less sure about the judgment further down the organization or in the IT department. Once they realize that machine learning is not about improving their personal decisions but about improving the quality of decision making at the operational frontline they get much more excited.
Machine learning teams often feel pressure to make a strategic difference to the company. They mistakenly assume that the way to do this is to have machine learning influence the company’s executives and executive-level decisions. This is a mistake. Better, instead, to work with executives to find the high volume, repeatable decisions that make a difference and use machine learning to improve them. Because these decisions are made so often, even small improvements multiply to give you a strategic impact.
Lots more good tips in the video. If you are interested in how we approach this why not read our white paper on Framing Analytic Requirements.
A few months back, Scott Adams posted a great Dilbert that I have been meaning to write about for a while (click on the image to see the original).
In the strip, Dilbert says “You don’t go to war with the data you need. You go to war with the data you have.”
Now Scott Adams was being funny but in fact there is a kernel of truth here. We come across many companies that are failing to apply data to their decision-making, delaying building predictive analytic models or postponing their adoption of machine learning because they don’t have the data they “need”. It’s not integrated enough, clean enough, precise enough or just not as good as it “will be soon”. This is a mistake. You should do as Dilbert advises, and “go to war with the data you have“.
The trick is to start with the decision you want to improve, rather than with the data. Understand the decision, model how you think you make that decision today, work with those who make the decision every day to capture your current approach. This decision making is possible with the data you have – it must be, as this is how you decide right now.
Now you can ask some interesting questions like:
What would help you make this decision more accurately?
Which pieces of the decision give you the most trouble?
Where do you spend your time in this decision?
Is the data you need to make this decision presented the way you use it in this decision?
Which pieces of this decision are data analysis – places where you decide something about the data so you can base some other decision on that analysis?
Sometimes the answer to these questions will lead you to new data or identify that your data needs to be improved. If it does, at least you can show exactly WHY you need that new data and so calculate an ROI. But often it reveals that you need to use the data you have in different ways.
The biggest benefit comes from identifying possible predictive models. Because you know how the decision is made, you will be able to see how accurate a predictive model must be to be useful. Often this is a lot lower than you think. We have had clients realize they only needed a model that as a little better than a coin flip and others who only needed 70-80% accuracy. You might need 99.99% but you probably don’t.
Until you know, you can’t answer the question if your data is good enough or not. Without a business-driven target for accuracy, your data team will assume something must be really accurate to be useful and they could easily overshoot. Plus many predictive models cope with missing and bad data quite well or can at least degrade gracefully when the data is of poor quality, allowing reasonable predictions even when data is less good.
So, don’t wait for the data you think you need, start improving decisions with the data you have. It’s noble, its heroic and it works.
Working with companies that are investing in becoming analytic enterprises, we have determined that there are three critical success factors. Whether you are focused on business analytics, data mining, predictive analytics, machine learning, artificial intelligence, or all of the above, these factors will be critical. Check out these videos that talk about them:
Analytic Enterprises Put Business Decisions First The first critical success factor for analytic enterprises is keeping the focus on business results by beginning (and ending) with business decisions, not analytic technology. https://youtu.be/DQ9GHSOxd9s
Analytic Enterprises Predict, Prescribe, and Decide The second critical success factor for becoming an analytic enterprise is moving beyond reporting and analysis of the past to prediction and action by using more advanced analytics to predict, prescribe, and decide. https://youtu.be/2128C4p8wVM
Analytic Enterprises Learn, Adapt and Improve The third critical success factor for becoming an analytic enterprise is recognizing that applying analytics is not a one-time exercise, and focusing on how to use analytics to learn, adapt, and continuously improve. https://youtu.be/rlnNtk9bSyc
And if you enjoy the videos, check out our white paper on building an Analytic Enterprise.
Gartner recently published a piece “Top 10 Trends in Data and Analytics, 2020” that you can currently get from our friends at ThoughtSpot (registration required). It’s an interesting report you should definitely check out.
My favorite section was the one on Decision Intelligence, within which they include the kind of digital decisioning or decision management I’ve been doing for the last couple of decades (and in which the firm I founded, Decision Management Solutions, specializes).
In this section they correctly point out that, while automating decisions is a critical component of digital decisioning, it’s not necessary to automate 100% of the decision 100% of the time. Often we find that sometimes an automated decision requires some human inputs or that only a certain percentage of transactions can realistically be handled by a decision engine. We build decision models to understand the problem well enough to make these calls – to decide on the automation boundary – and it was great to see the shout out for decision modeling (and the Decision Model and Notation standard) in the report. The team at Gartner linked decision modeling to improved agility (faster changes), transparency and business user enablement – all key benefits we see in client after client.Personally I always get the biggest satisfaction from seeing how digital decisioning enables continuous improvement, generating the data you need to review, improve, simulate and compare decision-making approaches. As the report says, the key is to pass actionable insights directly to decision engines to act and then enable humans to review the effectiveness of this and close the loop. Putting business owners in the driver seat for improving their own automated decisioning systems is a powerful tool that generates a huge ROI.
There were also some good pieces of advice on how to scale your Machine Learning (ML) and Artificial Intelligence (AI) efforts. I would add to their advice that scaling ML/AI in a fast changing world requires more than just adopting the right ML/AI techniques. It needs the active engagement of business domain knowledge through decision modeling and business rules too. No matter what you do to improve your AI/ML, there’s no substitute for combining it with in-house business knowledge. I also appreciated their comment that the approaches that got you to an AI pilot won’t get you to production – you need an approach like the one Cassie Kozyrkov discussed and I comment on in Some great advice on Machine Learning from Google (and me) or the ideas in this post on Most companies are not succeeding with advanced analytics. But you can.
Anyway, it’s well worth registering for and downloading. If you want to learn more about decision modeling, check out our great white paper on Decision Modeling with DMN and if you want an overview of our approach to machine learning, check out this paper on Enabling the Predictive Enterprise.
The work from home movement, voluntary or mandatory quarantining, retail store closures, and limits on public gatherings all serve to significantly increase our dependence on digital capabilities.
He goes on to talk specifically about insurance
Digital interaction capabilities: Self-service portals for agents and policyholders, websites that are easy to navigate and built using responsive design approaches, mobile apps for policy service and claims, and world-class call center technologies will become more critical than ever. Volumes are likely to increase as fewer face-to-face interactions occur by necessity.
We do a lot of work with insurers and they have historically been challenged when it comes to digitization:
Self-service portals only allow customers to do limited things – generally simple data updates or document review.
Agent portals are likewise often very passive, presenting data to agents but not really helping them manage their business.
Mobile apps focus on reporting the status of a policy service update or a claim when the customer just wants the update made or the claim paid.
Call centers increasingly have access to all your data but must constantly refer you to others for approvals.
The problem is that insurers have added digital channels, lightly digitized their data (scanned documents) and automated (digitized) processes without reviewing how they make decisions. They have paved the cowpath. And their use of Robotic Process Automation tools increases the odds that they will continue to do so. This is going to have to change. They are going to have to digitize decisions.
If policy update approvals are automated, customers can use the portal to do things not just request them.
If agency management decisions are automate, the agent portal can suggest how to grow the business and improve customer service not just passively support an agent who might not know what to do next.
If claims handling decisions are automated, mobile apps can support submission of claims and then respond immediately with “this claim is approved and will be paid in XX amount”.
If underwriting decisions are automated, websites and mobile apps can issue binding quotes and kick off the onboarding process 24×7.
If call approvals are thought of as the decisions they are not processes then call center reps can immediately assist customers, not just promise to talk to their supervisor.
COVID-19 is driving digital transformation. Insurance is a decision-centric industry and only if decisions are also automated can it be transformed.
A survey in CIO magazine on IT leaders’ thinking in the current crisis revealed that a plurality (37%) chose digital transformation as their first priority to help the business persevere through the current disruption. Moreover, a full 61% of respondents agreed with the statement that the effects of the pandemic are actually accelerating digital transformation efforts.
We certainly see this in some of our customers. The movement of staff to working from home, the need for clients to interact remotely, restrictions on travel and meetings – all these are increasing the value of digital channels while also providing concrete motivation to get over hurdles previously seen as insurmountable.
What’s interesting, though, is the extent to which digital transformation driven by COVID-19 is decision-centric – how much it relies on digital decisions not just digital data, digital channels and digital processes.
COVID-19 means digitizing decisions about customer interactions so you can build and sustain profitable interactions with your customers. Decisions about how to personalize and target the content you display, the emails you send and the offers you make are essential. Companies are finding that they have neglected these digital decisions, investing in digital channels only to deliver cookie-cutter digital content when they could be engaging customers directly and precisely. Digital “micro” decisions need to be made for each customer, each time.
COVID-19 means that manual approvals, manual discount calculations, manual eligibility checks and manual pricing are all problematic. Many companies have found that their highly automated digital processes just route work to a person for critical decisions. With more remote workers, more workers having to flex their schedules to cope with home schooling and customers increasingly doing likewise, trying to coordinate human decision-makers for these transactional decisions is not getting it done. These decisions need to be digitized.
Finally COVID-19 means that the way people and resources are assigned needs to change. When everyone was in the office together, informal ways to assign work and manage resources worked OK. Now they’re a recipe for delay and confusion. Decisions about assignments and allocation need to move beyond first-in,first-out queues and informal group discussions to precise, data-driven and digitized decisions.
Digital transformation is happening faster thanks to COVID-19 but the real opportunity is for companies to digitize their customer and transaction decisions.
With everything else going on at the moment, you may be finding it difficult to think about improving your decision-making. Yet COVID-19 has added a new set of constraints on your decision making and changed you objectives. This makes the management, and optimization, of decisions even more valuable.
A few weeks back we published a new white paper – The Customer Journey to Decision Optimization. Sponsored by FICO, this is available on their website. This paper lays out steps to adopt decision optimization, walking through how to codify your current practice, systematically improve your decision strategy and apply mathematical optimization.
We’ve published a couple of follow-up blog posts over on the company blog too:
We blogged previously about the need to react quickly and accurately to change and the importance of building and sustaining customer relationships remotely. One final area of focus is the need to be more flexible and dynamic in how resources are assigned and managed.
Companies are going to have more staff working remotely. Remote workers can’t share work as readily with their co-workers, or bounce assignments to others by simply standing up and waving across the room. A group of collocated workers could be assigned work pretty mindlessly and left to sort it out for themselves – simply queuing up the work for the group worked OK. This approach had problems before but now these problems become too serious to ignore. Organizations need to rethink how work is assigned and move away from first-in, first-out queues to more sophisticated allocation and assignment.
For instance, one call center we know supports remote workers and multiple locations by investing in its call routing approach. Instead of just dumping calls into a queue, the stated or predicted topic of the call as well as customer details and value are used to route it appropriately. To maximize the engagement and value of these calls, the organization also determines the best possible upsell or cross-sell to be made at the end of the call (including the possibility than not making one was the best option). Where possible, calls are routed not just to someone who can solve the problem the customer has but also to someone who’s good at selling the identified upsell. This focused assignment is highly dynamic and can be changed whenever new products or campaigns are launched or when the kinds of calls being received is impacted by outside events. Staff are given work they are good at, customers get agents who can help and who are enthusiastic about the products being up-sold, and results improve.
Even when organizations have built dynamic allocation and assignment approaches, there are often overrides. When call volumes get high, when the quarter-end is looming or when marketing campaigns are in flight, all the sophistication is thrown away as the override kicks in. This is never great – overriding things like this inflicts long term damage on data-driven improvement for instance – but it becomes very high risk when circumstances may often result in new overrides. Like now, when the constant changes to COVID-19 rules and regulations are causing companies to keep overriding their systems and process to cope.
The new environment is going to require much more managed overrides and more rapid change. One of our clients, for instance, has a routing algorithm that considers a wide range of factors before deciding how to handle a particular product return. Various situations exist that “override” the default algorithm but these are built into it. Business users can simply identify that a particular circumstance has occurred and the built-in changes to the algorithm are triggered. The business gets the appropriate response, users still rely on the system in exactly the same way and all the usual data gets collected. The “override” gets 100% compliance from users because its built in and the cost and impacts of the override are transparent because the system still runs and still creates the data needed for analysis.
All of this brings up one last point. Precise resource assignment increasingly involves algorithms – machine learning and predictive analytics. Which customers are churn risks, which products will be appealing, who’s likely to be able to handle this problem – all of these are candidates for advanced analytic models and machine learning algorithms. Focusing on the business problem – how to do the assignment or allocation – frames your need for these analytics so you can not just build the algorithm, you can also get it across the “last mile” and into production.
These systems are going to require new decisions to be made – dynamic, managed, automated assignments and allocations will replace first in, first out queues and simple distribution. Processes will change to add these decisions. Analytic insight will matter in these decisions, as will expertise and experience. Batch analysis will be replaced with real-time scoring and simple logic with managed business rules. Decision models will help you integrate these elements to deliver the real-time decisions you need.
Dynamic, managed assignment is something you can build into your systems now. We’ll be posting regularly on ideas and approaches and producing some great content. To stay up on it, why not sign up for our newsletter.
We blogged about the need to react quickly and accurately to change already as this is one of the key ways you can adapt your operations. A second is to ensure that you can build and sustain your customer relationships even as your customers are more remotely and have fewer in-person interactions with you.
The new era is going to require much more automation to support remote interactions. Customers are going to be more remote. They’re going to be under more stress too, so quicker responses will be more valued than ever. And the staff you would need to support a manual decision may be remote themselves, making manual responses problematic. Automated customer interactions can’t be cookie-cutter, though. Customers still want to feel that you know them and value them. You can’t pick between automated and personalized when your customers want both. And you still need the flexibility we talked about earlier, because all of this is a moving target.
An organization we work with recently began investing in building and sustaining customer relationships online, having for years relied on in-person interactions. When prospective customers start looking online at information about products, this organization makes sure it presents the right first product. Using everything it knew about the prospect, it presented targeted (and compliant) “first best offers” to these clients. Engaged customers could buy the product online but it was a relatively complex product to buy so many would choose, in the end, to talk to someone. So now the lead, the initial offer and the partially completed application are all routed to the right point of contact. Rapid feedback on customer behavior and flexible automation let them keep the initial conversation feeling personalized and relevant even as products, marketing campaigns and the day-to-day business context changes. The automation meant that customers could choose to no people when automation was what they wanted and could be routed to the right person when it wasn’t.
This organization applied this same mindset to helping its agents in this new, online engagement approach. Agents used to in-person meetings are threatened by digital channels, and by fully online direct-to-consumer competitors. Helping agents manage their customer portfolio, identifying opportunities to engage with existing customers around special events or campaigns, and helping the agents target customers and prospects precisely all help agents add value to the customer relationship, even as it becomes an increasingly electronic one.
To deliver systems that engage remotely with customers in this way requires automation of the underlying decision and the targeted application of machine learning. You can’t just look at the data though, you need to combine what the data tells you with what your expertise tells you and filter all this through the current (changing) state. The analytic insight you develop must be in service to the customer treatment decisions you are trying to make. Decision modeling will frame up the analytics you need and, integrated with business rules, deliver targeted customer treatment across all your channels.
We have a proven, established way to build systems that will engage customers remotely and we’ll be posting regularly on ideas and approaches and producing some great content. To stay up on it, sign up for our newsletter.
One of the key ways you can adapt your operations is to develop systems and processes that can react quickly and accurately to changes as they happen. The response of governments at the city, state and federal level is going to evolve as the situation changes. The behavior of customers, partners, suppliers, agents and distributors is going to change too. You need to keep operating through all this change, without creating unnecessary manual clean up, without “hacking” temporary fixes every few weeks and without putting yourself at risk for legal or compliance (or publicity) problems.
One of the persistent myths in systems development is that systems and processes can’t be both flexible and compliant. COVID-19 puts this old saying under real pressure. If you work in a regulated industry (insurance, financial services) or deal with regulated consumer data, your regulators are not suddenly going to give you a pass because there’s a pandemic on. They might be a little more flexible, give you a little more time to report things or adopt new policies, but they’re going to expect you to remain compliant – even as what it means to BE compliant changes all the time. Now you really need to be flexible AND compliant.
One marketing department we worked with has the kind of flexibility you need. They can change the marketing offers they make to new customers whenever they need to. They can try multiple approaches to picking an offer to see which works best in rapidly changing circumstances. They have the transparency and auditability they need to show regulators that they never tried to sell products to people not eligible for them. This kind of system allows you to turn on a dime as states relax and tighten restrictions, guidance changes and new polices and regulations come out.
The key to compliant flexibility lies in business enablement. Like the marketing department in this example, you need business owners to understand and be engaged in managing the system, to be empowered to make the changes they need. They are the ones closest to the regulations, closest to the customer, setting the policies. When they can see how the system is behaving, change the way the process works, and drive the outcomes they need then you’ll get the quick, accurate, compliant response to change that the new era is going to require.
Building these kind of flexible yet compliant systems requires a focus on the decision-making embedded in these systems and on exposing that decision-making so business owners can change it themselves. The decision making is what changes the most often and the business owners are the ones who see those changes first and who understand them best. Decision modeling, business rules management and a focus on continuous improvement all contribute to developing these systems that can react quickly and accurately.
The good news is that there are proven, established ways to build these systems and we’ll be posting regularly on ideas and approaches and producing some great content. To stay up on these ideas, why not sign up for our newsletter.
The COVID-19 Coronavirus has upended the world economy – and your business. While the impacts to date have been dramatic, we have to face the fact that this is the beginning of a new normal – a world in which this virus circulates. This will have short- and medium-term consequences for your business regardless of the industry you operate in.
You’ve probably spent the last few weeks re configuring everything to function effectively as a remote enterprise. You’ve dealt with the immediate impacts like surges in VPN access, new laptop requirements, making processes work when people are working from home. Plus ,you’ve survived the immediate economic hit. In the coming weeks and months, organizations like yours will begin to focus on cost optimization and seeking opportunities in the new environment. Some companies will succeed at this – they will adapt, survive and thrive in this new environment. Others will fail.
Now is the right time to be thinking about how you should adapt operations to cope with the new normal. It’s time to plan because this is not going away. The problems you are facing will change and evolve but there’s a new normal coming and you’ll want to have new systems and processes to cope.
How will you create targeted, engaging and profitable interactions with customers you will never meet? And sustain relationships you started offline now they must move online?
How will you create resilience in your supply chain and other processes, without tying up more capital? How do you ensure you can adapt rapidly to shifting targets, regulations and opportunities?
How will you shift your adapt operations to handle an increasingly volatile and dynamic world? Can you assign and allocate resources based on what’s happening now, or soon, not on how things used to work?
Our focus here at Decision Management Solutions is on digital decisioning – using our DecisionsFirst™ approach to apply technology and deliver automated solutions to decision-making problems. The systems we help our customers build address exactly these issues: they ensue you can personalize and target the digital interactions you have with customers; they deliver resilience, transparency and agility; and they support dynamic assignment and allocation.
We’ve learned a lot about building these kinds of systems over the decade or more we’ve been doing this and we’d like to share some tips and help you see how you can do it too. We’ll be posting regularly on ideas and approaches and producing some great content. To stay up on it, sign up for our newsletter.
Some of our old friends at Gartner have just published some great research on Decision Management. Specifically they have extended their work on Decision Management Suites (blogged about here) and focused on How to Choose Your Best-Fit Decision Management Suite Vendor [Gartner subscription or modest fee]. As they say in the intro:
Decision management suites go beyond business rule management systems by providing more features for designing, deploying, maintaining and auditing decision-making software. This report describes the steps data and analytics leaders should take to identify the best vendor for their business needs.
Gartner
Selecting a decision management vendor or business rules management system is something we here at Decision Management Solutions do a lot so I was excited to see what the Gartner team had come up with.
They suggest that you identify potential providers by considering their support for both business rules and analytics. As I have described in my books (most recently Digital Decisioning: Using Decision Management to Deliver Business Impact from AI ), the decision services you build are going to need business rules, are likely to need machine learning or predictive analytics and may need optimization (though much less often). Streaming analytics and event processing are included in this list by the Gartner team but I see this more as a niche market – generally not relevant but occasionally critical.
Building decision services with a mix of these elements requires decision modeling – specifically decision requirements modeling. You can build a decision requirements model that pulls together business rules, machine learning and optimization to give you an effective, graphical blueprint. Don’t leave home without one. We build a lot of decision services and we would never do so without a decision model. You shouldn’t either.
Below is the list of topic areas they considered and they have some great content in the paper. Based on our experience developing decision services and modeling more than 4,000 decisions, I have a few additional comments:
Ease of Authoring Our experience is that this is all about decision requirements models and associated decision tables. People often evaluate other authoring elements but they don’t really matter – business users like decision tables and decision requirements models are essential for getting decision tables right..
Application Solutions and Templates
Operating Environment
Build, Version and Deploy Always consider this as two separate threads – one focused on how non-technical users do versioning and deployment, one on integrating with IT’s processes like CI/CD. Don’t mix them and don’t assume that being good at one makes a platform good at the other. Some platforms are very programmer-friendly but baffle business users.
Scalability and Latency
Logging, Monitoring and Evaluating Remember that the data you capture to track how you made decisions and how that worked out for you will drive continuous improvement. This is much more important than the logging or monitoring of rule execution which is interesting only sometimes.
Process Orchestration/Workflow Meh. Build stateless, side-effect free decision services and leave the workflow somewhere else.
Simulation This is REALLY important. Don’t miss this. And don’t confuse it with testing, which it is not. Testing is to see if something is broken. Simulation shows the impact of a change. Business users make lots of changes that should not result in test failures. Make sure they can simulate the impact of a change before they commit it.
Rule Validation
Microsoft Excel Support Not a fan. Excel is super flexible but that’s not necessarily a good thing. Either commit to a product that does everything inside Excel or one that provides decent editors. Don’t get stuck in the middle.
Rule Harvesting Never harvest rules until you have a decision model in place. Ever. Really, just don’t. Please.
Supporting this paper is a Toolkit: Decision Management Suite Vendor Profiles [Gartner subscription required]. While I don’t think some of these are really Decision Management Suites – a couple are really workflow or streaming engines with some rules support thrown in – its mostly a good list and a reasonable framework to use. Just remember, as the authors point out, to weight the factors that matter to your project.
If you’d like our help selecting a vendor or platform, contact us and we’d be happy to talk you through what we do. If you want more background on decision requirements modeling, check out this paper.
Balancing multiple profit drivers is essential in many high volume, transactional decisions targeted with Decision Management. Being able to execute decisions, monitor their effectiveness, learn what works, and systematically improve your decision-making results in significant business benefit. Mathematical optimization let’s you go even further.
This paper lays out the proven steps to adopting decision optimization, walking through how to codify your current practice, systematically improve your decision strategy and apply mathematical optimization.
We’ve published a couple of follow-up blog posts over on the company blog too:
Don’t Just Sit There, Experiment! Continuous and systematic improvement in the quality of your decision-making requires an investment in experimentation.
Why not download the paper and start your own “Journey to Decision Optimization”?
If you’re stuck sheltering in place or bored working from home, we have a great opportunity for you. You can learn a vital new skill – Decision Requirements Modeling – from the comfort of your home FOR FREE!
We are going to take some of our popular online training materials and provide a FREE 2-hour introduction to modeling decision requirements. This session will focus on the most valuable feature of decision modeling, its power to accurate capture how business people want a decision to be made.
Decision requirements modeling captures business decision-making using a graphical representation – a decision requirements model using the Decision Model and Notation (DMN) standard. This powerful visual analogy is used place of technical requirements and enables practical, creative design of solutions to your decision-making problems.
The class is 2 hours, it’s FREE and its on April 3rd, 2020 10am-Noon Pacific (1pm-3pm Eastern)
We’ll introduce decision requirements modeling and the DMN notation, teach you how to define decision requirements using the DMN standard, show you how to drive decision-centric design sessions and help you get started using decision requirements modeling on your projects.
It’s 100% online, 100% FREE and interactive – you’ll be building a model in the class. Sign up now!
I recently got together with Irene Lyakovetsky of Saugatuck Worldwide and well known independent analyst Jim Sinur to talk Digital Decisioning. Irene released our talk on her Saugatalk channel as a great series of snippets:
Balancing key profit drivers such as revenue, risk, & costs is central to success in lending and many other high volume, transactional decisions targeted with Decision Management. The decision improvement lifecycle is central to delivering the full value of decision management. Being able to rapidly and effectively execute decisions, monitor their effectiveness, learn what works (and what does not), and iteratively improve your decision-making results in significant business benefit. Add mathematical optimization and you can go even further.
Designed to help you optimize your decisions and aimed at rules and business analysts, this paper lays out the proven steps to adopting decision optimization. It walks through how you can codify your current practice, systematically improve your decision strategy and apply mathematical optimization to make better decisions. It will show you the value of decision optimization and identify the critical success factors.
Decision Optimization can accelerate your success by helping you develop decision-making approaches that better meet your goals and objectives. This paper will show you how.
Over the next few months we are going to publish some supporting blog posts, focusing in on some key aspects from the paper. Next week is the first, discussing the importance of continuous improvement, and the others will follow every other week.
Meanwhile, don’t forget to download the paper and start your own “Journey to Decision Optimization”.
At Decision Management Solutions we do a lot of work with IBM and an old friend of ours – Harley Davis – just published a nice blog post “How do you best operationalize AI?” We’ve had many conversations on this topic with Harley and his team and it’s no surprise we’re completely aligned on this.
Harley outlines three steps:
Reimagine your process Specifically by identifying and modeling your decisions
Assemble a decisions team One that has all three legs – business operations, IT and data science/ML
Map the data Make sure the ML/AI algorithms can use the data available to make the decision where it is in the process
We have a whole approach for these kinds of projects – what we call DecisionsFirst™. We have successfully applied this with multiple clients, helping them operationalize their machine learning (ML) and AI investments and deliver more effective decision-making. We see three critical success factors:
DecisionsFirst Design Thinking Always begin by identifying your decisions, modeling them directly with the business owner and using design thinking approaches to ideate ML and AI opportunities
Mix and Match Technology Use ML and AI technology for sure but mix it with business rules technology (using a Business Rules Management System like IBM’s ODM) and use a decision model to coordinate all the pieces.
Continuous Improvement It’s not about the first version you deploy, it’s all about how regularly and effectively you can improve it. Invest in business enablement and continuous testing, experimentation and improvement.
Historically, organizations have codified their best practices in rules. Now they are hungry to take advantage of the possibilities of AI to leverage the wealth of information in their historical data and take even better decisions. They are struggling to do so. James Taylor makes it clear how to succeed: By combining business rules and machine learning in a single digital decisioning framework. With clear explanations and examples based on years of practice, this book lays out what can be accomplished with digital decisioning platforms – and how to go about it successfully.
