The IBP paradigm has hardly changed for 30 years, the decision needle has not moved enough.

Intelligent IBP can change this.

In this blog I will introduce 27 Intelligent IBP principles I’ve highlighted the last four years in blogs and articles. They will likely change and there will be many more to come, but this is where it starts.

A bit of history

This goes back to early 2000’s implementing APS for many years across many countries in my early consulting days. After that being supply chain and IBP manager for many years and implementing many IBP processes as consultant. I’ve experienced the strength and shortcomings of both APS and IBP over 20 years. I’ve seen the pain of planners spending 50% of their time on tinkering with data and working overtime to get the deck out for an IBP meeting.

I’ve been writing about IBP in my blog since 2009 and have shared many ideas to challenge the status quo or improve IBP across company culture, process, and behaviours. I’ve conducted seven S&OP pulse checks over the years to get field insights from practitioners. As mental toughness coach, I developed a questionnaire to measure S&OP mindset and created unique insights that an effective mindset has correlation with a self-proclaimed effective S&OP.

My first published article was in 2012 for JBF and I started to write for Foresight in 2016. I have published many articles for the advance of IBP knowledge. I became Foresight IBP editor in 2020. In short, I’ve done this S&OP/IBP stuff for a while from many different angles.

Intelligent IBP in the making

As early as 2016, I wrote about autonomous IBP, comparing IBP to an autonomous car. Then, as a response to the ‘touchless planning’ hype, I published an article in 2019 about what is really required for ‘autonomous planning’. When my colleague Hein Regeer and I in 2021 then envisioned a new type of IBP, where decisions are segmented between humans and machines, with increased levels of planning and decision automation and augmentation, I was pleased with the feedback from supply chain professor Nada Sanders:

 ‘You offer an excellent path toward reinvention of traditional IBP that companies should follow.”

Nada is co-author of the HUMACHINE besides plenty of other books. Alain Perrot, a 30-year IBP veteran with over 200 implementations on all continents in all industries, congratulated me with changing the paradigm, but also challenged me to describe a full planning symbiosis between human and machine. A similar idea that Nada describes in her book HUMACHINE. A 10-year Oliver Wight veteran shared that he agreed for 95% with my ideas. Their feedback gave me some reassurance that I was not talking nonsense!

When I thought about what I could call this evolution of IBP, I simply landed on Intelligent IBP, as I believe both machine and human intelligence have to be exploited more. In February 2022, I wrote my first blog using the name Intelligent IBP and launched the term and hashtag #IntelligentIBP on LinkedIn.

Enter Decision Intelligence

Working for Aera Technology, applying intelligent automation to decision making, helped shape these early ideas further since I joined the company end 2019.  The launch of a Decision Intelligence (DI) definition by Gartner later in 2022 further cemented my believe that this new discipline will become mainstream and can be applied to every business decision.

Aera defines decision intelligence as “the digitization, augmentation, and automation of decisions.” In that order, because we first have to be able to orchestrate and digitize the decision process, including all data, plans, parameters, and decision policies that feed into it before we can augment or automate.

Google’s Cazzy Korzakov already had her DI definition in this blog from 2019, and Lorien Pratt published the book Link about decision intelligence in that same year. These two ladies are real Decision Intelligence trailblazers, both have roots in AI/ML. A lot of good decision practices go back to the pioneering paper; Decision Analysis: Applied Decision Theory from Ron Howard in 1966.

Cazzy defines Decision Intelligence as “the discipline of turning information into better actions at any scale”, highlighting that the days of information, insights and BI are over, or at least not enough anymore. We need decisions and actions.

We all stand on the shoulders of giants. What I learned over the last years about decision intelligence, is that you can look at it from many angles as there is such a wide range of decision types and circumstances out there.  And there a plenty of different sciences and management practices to support or improve it. So, there are many possible approaches.

I’m just the guy that tries to apply it to the range of decisions covered by IBP using a synthesis of the HUMACHINE, decision intelligence, good decision practices and my own IBP experience.

Principles of Intelligent IBP

Let’s start with a definition.

Intelligent IBP effectively combines human and machine strengths with good decision practices to make timely, high-quality business decisions.

This definition should become clear once you go through the principles. These principles are a first guide around the WHAT not necessarily the HOW of Intelligent IBP.

Decision centric approach

Although IBP is an executive decision-making forum, historically the focus has been on information, process & planning. Not so much on decision making, decision quality and good decision practices. Unfortunately, only 2% of managers apply best practice decision practices. Additional benefits of good decision practices are reducing (non automated) decision time with 75%, reducing wasted analytics with 60% and improving employee engagement with 50%. Intelligent IBP will focus on the decision first.

Principle 1: Intelligent IBP takes a decision centric approach and has a thorough understanding of decisions to be made, decision quality and good decision practices.

Decision Cycles & Decision Types

In a 2020 Foresight article, I highlight the different horizons in common business planning cycles and how they are related. Based on six planning & decision drivers (frequency, data generation, granularity, impact, system complexity, human centricity), I then extract the relative efficacy of automation (in both process and decisions) vs. augmentation across the business planning horizons. I conclude that the longer the planning horizon, the more human centric decisions will be.

Principle 2: Intelligent IBP can be applied to every business planning & decision cycle.

Figure 1: Business planning cycles

Figure 2: Six automation and augmentation drivers and their relative impact by planning horizon

Figure 3: Relative preference for automation & augmentation by business planning horizon

In a 2021 commentary on an article about the HUMACHINE, I segmented decisions in operational, planning, strategic and cultural decisions. I would now split planning into basic planning decisions (change a price to close a projected value gap) and complex planning decisions (align internal functions, supplier, and customer around a new product introduction, relationship complexity that a machine can’t solve). In my presentation at the 2021 International Symposium on Forecasting (ISF), I then combined decision types with augmentation and automation efficacy as per figure 4 (which I would also slightly change now: I would add augmented analytics, which is more applicable for strategic decisions than decision augmentation).

Figure 4: Slide from the ISF 2021

Now there are other ways to segment decision types, you can check McKinsey decision types here. The point is that to describe a full symbiosis between human-machine across horizons and decision types, all decision types are part of the intelligent IBP decision scope. Even when there is no planning involved for a decisions, other types of augmentation (machine support) are still possible.

Principle 3: Intelligent IBP includes every decision type in a business planning cycle.

There are different ways to segment decisions, however the key is to segment IBP decisions in machine centric IBP decisions: guided by the human and Human centric IBP decisions: augmented by the machine. This is what my colleague Hein Regeer and I proposed in our 2021 Foresight article. See figure 5.

Principle 4: Intelligent IBP segments between machine-based decisions, and human-based decisions.

Figure 5: IBP decision type segmentation by human and machine

Decision automation versus augmentation

What decisions can/should be automated?

Not everything can or should be automated. But where we can, we have to consider it. Data gathering & cleansing, descriptive analytics, dashboards, reporting, operational and basic planning decisions, can be automated where possible (mostly in the S&OE horizon, but also beyond). Master data maintenance can be highly automated. Planning parameters (like a lead time, safety stock, etc) can be maintained continuous and become probabilistic in nature.  Short and mid term demand & supply balancing decisions, can be automated, including volume and value gap closure. All between cross functionally agreed boundaries. The human still decides how the machine operates these decisions and has to maintain machine boundaries (and negotiates these across functions).

Principle 5: Intelligent IBP automates repetitive planning tasks, basic planning problems and decisions.

Principle 6: The human provides the boundaries on where and how the machine should automate.

What decisions should be augmented?

Where automation is not possible or the preferred option (too high value or impact, contextual or relationship complexity), the human leads and the machine helps the human by automatically providing multiple options to choose from. I see two steps in augmentation:

1. Augmented analytics runs (automated) advanced analytics to provide unique insights or different possibilities to a planner/user. For example a monthly triggered risk analysis (set of simulations) from your e2e supply chain. No suggestion or decision will be made, just insights.

Human centricity remains paramount around cultural decisions & visioning. It is the only basis for defining purpose, values, and behaviors, but also sustainability, corporate responsibility, diversity, and inclusion, elements which are beyond the reach of the machine. Augmented analytics here is possible, augmented decision making not.

I see more strategic decisions (close a plant or DC, enter a new product category, change in strategic sources) more in this category as well (which is different than what I show in figure 4).

In these cases, the machine can still augment the human with analytics, providing options, what if’s, long term trend analysis, scenario’s, simulations, war games to choose from. It is less likely to give ‘advice’. Still, providing the different options to choose from before deciding, improves decision quality.

2. Augmented decision making. This is where the machine gives an ‘opinion’, advice or recommendation to the human about the best choice to make between different options but leaves it up to the human to agree or not.

These will be operational and basic planning decisions that otherwise could be automated, but go beyond a certain threshold. Hence the human needs to be in the loop to make the final decisions, but the machine will give the ‘advice’ around what’s best in the circumstances (which it has learned over time).

Add to this more complex planning decisions (For example with relationship complexity, NPD or price decisions with suppliers and customers). The machine can automate a price change for an online retailer easily. However, for a manufacturer a price or promotional change might first need to be aligned with the customer. Hence the machine can give this advice, but the human need to stay in the loop for the final decision.

All augmentation itself can be automated through a time, variance, probability trigger, or if not, by an ad-hoc trigger by the human. These are rough guidelines. In the end, every business has to take a decision centric approach and decide what can be automated and what should be augmented.

Principle 7: Intelligent IBP augments the human by providing multiple options for decisions that are not automated.

Principle 8: The human provides the boundaries on where, when, and how the machine should augment.

Human & machine strengths

Moravec’s paradox describes that human’s and machine have complementary strengths. Intelligent IBP will need to exploit both strength in human & machine segmented decisions.

Human centric decisions can include human qualities like creativity, empathy, innovation, contextual awareness, relationship management, meaning and purpose. Creativity (brainstorming) and innovation are important to seek multiple, or new options/angles to solve issues. Having multiple options is an important element of decision quality. Empathy and relationship context are important for human change and implementation/execution aspects of decisions.

Machine planning & decisions will focus on machine strengths like computing power, frequency, speed, scalability, consistency, and endurance.

Principle 9: Intelligent IBP exploits machine strengths in automation and augmentation

Principle 10: Intelligent IBP seeks to exploit human strength in human centric decision making.

Human-Machine collaboration & trust

Kasparov’s law dictates that a good process between and average human and machine is superior to great machines and great humans with an average process between them. Hence, human-machine collaboration becomes critical.

It should be easy for the human to provide goals and automation and decision policies to the machine. It should be easy for the human to receive and understand advice from the machine in order to make and execute a decision.

Principle 11: Intelligent IBP requires an effective, near frictionless human-machine interface.

A good human-machine process builds trust and enhances collaboration, but to maintain that trust, the human also needs to understand to some extend what the machine does (although, how many supply chain prefoessionals can really explain how exponential smoothing works, or Kings formula for safety stock?).

Principle 12: Intelligent IBP provides explainability for automated and augmented decisions.

Good Decision Practices

For human centric and augmented decisions, emphasis will be put in negating the over 200 human biases in decision making. Good decision practices can minimize human biases and should be used to get the most out of human strengths. Althought speed of high impact decisions (which should be the only decisions in an exec IBP meeting) is still relevant, the reduction/prevention of bias for high impact decisions is critical. Bias reducing decision practices are a major gap in traditional IBP.

Principle 13: Intelligent IBP uses good decision practices to minimize human bias and speed up decision making.

Decision Quality

Although IBP is an executive decision-making forum, there has been limited focus on decision quality. Decision quality will be included in the design for automated decisions making. There need to be controls for machine ethics and bias in automation and augmentation. Although machine bias has a trade-off against variance, so that’s more optimizing both, rather than only minimizing bias. For human centric decisions, it will be part of the orchestration process for high impact decisions. Decision quality will become a performance metric in IBP.

Principle 14: Intelligent IBP seeks to integrate decision quality for every decision type.

Principle 15: Intelligent IBP integrates decision quality in the design of automated decisions.

Principle 16: Intelligent IBP seeks to optimize machine bias in decision making.

Principle 17: Intelligent IBP measures decision quality and decision impact for every decision type

IBP decisions and its impacts will need to be digitally recorded, be transparent, accessible for al IBP stakeholders, and be able to learn from decisions over time (by applying machine learning)

Principle 18: Intelligent IBP digitizes every decision type.

Principle 19: Intelligent IBP learns from decisions made.

The IBP operation model

Many planning & IBP decisions are actually repetitive. Maybe not frequent, but they have been made before. If it is not in your business or industry, it will be in another industry or business. Most business decisions are not as unique as we would like to think. This means we can digitize and orchestrate many of them and work our way back what is needed to make this decision in the best way possible, including decision quality & bias reduction. If we identify new and unique high impact decisions, a Casual Decision Diagram (CDD) can be used, as defined by Lorien Pratt. Once a team works through the CDD, the decision is defined and can be digitised in a library.

Principle 20: Intelligent IBP maintains a digitized library of decisions, how they are made, what the common options are, what inputs are required and who decides.

Below picture shows an overview of demand & supply balancing options in a global beverage business. Many decisions will be familiar, even across industries. The options to choose from will largely be known, so are the data, plans & insights required. Intelligent IBP will have these documented and highlighted where there is automation and augmentation potential.

Organizational design (OD): there will certainly be changes in the planning and IBP OD. For IBP, I foresee the creation of a decision squad to tackle urgent, high impact decision, that otherwise simply can’t be solved timely and with enough quality. There will be less planners, working with the machine covering more planning scope, supported by analyst and data scientists to improve automated and augmented planning.

Think about this anecdotal example of a very large CPG with a portfolio of 25,000 SKUs with 800 planners for a certain geography, versus an mostly digitized online retailer with 5 million SKUs, 15 planners and 150 analysts.  Sure, they have different supply chains to manage and to plan, but you can guess what the long-term trajectory will be for the planning structure in the CPG.

Roles & responsibilities: the role of the planner will change as I describe in this blog. We will see the birth of the decision role and the automation role, and we see how the planning & information role will become less transactional and more strategic.

Principle 21: Intelligent IBP minimizes dull repetitive planning tasks and makes planning roles more strategic and more value adding.

Principle 22: Intelligent IBP requires new and changing roles for a planner.

Planning incentives: IBP incentives will shift from policing process compliance like in traditional IBP, to provide strategic business scenarios, find incremental levels of planning & decisions automation and augmentation. Measure and improve decision quality.

Principle 23: Intelligent IBP requires new incentives for planners around decision automation, augmentation, speed, quality and learning capability.

The IBP cycle: The frequency and duration of IBP meetings will be more driven by decision requirements, rather than fixed process requirements. Ad-hoc, high impact decisions can be facilitated outside the IBP cycle but will stay aligned. I described this in my 2023 Foresight article using below diagram.

The traditional planning cycle (PR, DR, SR, IR, MBR) might change, but I’m not sure yet how. I still see value for humans to come together in a monthly cycle and apply human strengths (and minimize bias) to business problems and decisions.

Augmented decisions that can’t find a resolution are centrally stored and visible at any point in time for IBP stakeholders. Plans, gaps & risk are live, transparent at any time and can’t be functionally or organizationally hidden! Cross functional augmented decisions are supported by a smart workflow, where changes in decision policies and decisions are visible and auditable (think blockchain). Anything in the decision library that can’t be solved, is automatically the input for the IBP meeting.

Principle 24: Intelligent IBP makes high impact executive decisions at a higher clock speed than the traditional IBP cycle but stays aligned with it.

Principle 25: Intelligent IBP uses a smart, auditable workflow to sign off augmented decisions.

Principle 26: Intelligent IBP maintains a centralized, transparent library of planning outcomes, gaps and risks for decisions that require executive attention.

And last but not least,

Principle 27: Intelligent IBP seeks a full symbiosis between human and machine to make decisions; however, the human will always be leading.

Even when we 100% automate (not likely) decisions in the shorter horizon, a human still has to maintain the machine, provide direction, polices and goals. AI without knowing what to solve can be rather dumb.

So, even in automation the human is still a leading factor. With augmentation the human still makes the decisions, and with 100% human centricity (decision about values or behaviours for example), the human is absolutely leading, and the machine is pretty much out of the loop of making the decision. At best will provide augmented analytics.

There is still a lot of ground to cover, but I’ve seen enough in both human and machines capabilities to believe Intelligent IBP can become a reality.

Image: Created with Hotpot.ai – “A human and machine making a complex plan and make decisions together”