IBM Research Scenario Planning Advisor (SPA)     


 Timothy R Dinger photo photoOktie Hassanzadeh photo photoSHIRIN SOHRABI photo photo

IBM Research Scenario Planning Advisor (SPA) - overview

The Scenario Planning Advisor (SPA) is a technology created by IBM Research that automatically projects many plausible high-impact future scenarios, to provide insights for strategic decision making. SPA allows domain experts to generate diverse alternate scenarios of the future, enhancing their ability to imagine various possible outcomes, including unlikely but potentially impactful futures.

Short video                Longer version video                The SPA service, deployed under a non-commercial use license


Unlike most forecasting approaches, SPA does not rely on traditional statistical forecasting techniques. This is especially important where a key event or impact has never happened (or happens in a new way) and where little or no data exists from which to forecast its occurrence. Instead, SPA uses Artificial Intelligence Planning to generate scenarios, and uses Natural Language Understanding to extract knowledge from documents to craft the planning models needed to generate the scenarios. SPA provides the ability to engage in what-if analyses – whereby users identify risks to be explored – and SPA generates, prioritizes, prunes and presents scenarios that can be instrumental in strategic decision-making. SPA also employs current state awareness, to provide hints about where to focus what-if scenario analyses.

Potential users of SPA engage in Scenario Planning, to help them envision and plan for important future risks and opportunities. They look ahead to future events that can have positive or negative impacts on the business, and look for ways to:

  1. encourage/promote/enable the positive impact
  2. prevent the negative impact
  3. remediate, where prevention isn't possible or doesn't work

The users currently do this manually/cerebrally/socially and, since the process is fatiguing, they tend not to create many scenarios.

SPA helps, by automatically generating many more scenarios, scenarios that explore unanticipated outcomes and/or unanticipated ways to getting to those outcomes. SPA helps to expand decision-makers' access to a diversity of future business impacts, and to identify paths from now to those impacts and to get a sense of what to look for along the way. And when the underlying models used to generate these scenarios are assembled from a diversity of industry/sector knowledge, those scenarios can also be generated in a much less biased way. By comparing a variety of “what if” scenarios, a user can better understand and plan for the impacts of possible policy decisions and possible future events.


Here are a few of SPA's key capabilities:

  1. Many more scenarios:  exposes users of SPA (e.g., decision makers) to more scenarios than they have the stamina to generate manually
  2. Modernizes Strategic Planning to be more responsive to a rapidly evolving world:  Because SPA accelerates and expands the scope of scenario generation, SPA can completely change the strategic planning cycle, from yearly to, e.g., quarterly.  Or it can be applied on demand, as new or unforeseen world events arise.
  3. Ability to focus on the new/unexpected: There's a dangerous comfort zone in describing familiar scenarios. The prospect of seeing a greater variety, and seeing unexpected futures helps decision-makers go beyond their comfort zone and encourages them to consider new/different outcomes, and/or new paths to outcomes
  4. Reduces/minimizes bias: A small number of scenarios, generated by a small number of individuals, tends to carry the bias of these people. Expanding the scope by bringing in a diversity of causal perspectives and mechanically generating scenarios tends to identify unbiased results (5 people can say "that will never happen" - more people, with a diversity of experience can say "that's not at all likely, but what if it does happen?")
  5. Transparency: SPA's causal models are accessible to anyone who cares to examine them, and easy enough for most people understand, and even to contribute to. This allows decision makers to see and influence the underlying model (not a black box) and express their own emphases (specifying Likelihood, Impact, and Duration of causation)


Example: Chief Risk Officers (CROs) are concerned with future risks to their companies – both negative (significant revenue losses) and positive (discovery of viral new products) - and how those risks will impact their business.  A CRO anticipates the need to project future events that might be subject to, e.g., supply shocks and/or demand shocks, disruption of supply chains, etc. The CRO builds a model capturing these elements and capturing the causal links between these risks (High unemployment rate could lead to Social Unrest). In addition, the CRO identifies potential triggers for such risks, such as Natural disasters, Risks to the workforce (including diseases and pandemics), etc.

With those various model elements built out, the CRO and/or strategic decisionmakers could then engage in what-if analyses, selecting one or more risks to explore (Pandemic, Increasing unemployment), requesting that SPA generate multiple scenarios to identify paths to the potential business impacts of the designated risks (Low morale due to reductions in force, Revenue loss in the services sector, ...). SPA automatically generates these scenarios. While some are normal and anticipated (and some not relevant), some scenarios present high-impact business implications, including some surprises. The CRO can then further refine these scenarios by generating related scenarios with additional hypotheses about various other risk factors.

When the CRO is satisfied with the relevance and importance of a set of scenarios, these scenarios are then brought forward to a planning group for strategic decision making and, possibly, for tracking and action (prevention and/or remediation).


Selected publications (see the publications tab for the full list):

IBM Scenario Planning Advisor: A Neuro-Symbolic ERM Solution
Mark Feblowitz, Oktie Hassanzadeh, Michael Katz, Shirin Sohrabi, Kavitha Srinivas, Octavian Udrea
Proceedings of the Demonstration Track at the 35th Conference on Artificial Intelligence (AAAI-21), 2021

IBM Scenario Planning Advisor: A Neuro-Symbolic ERM solution
Mark Feblowitz, Oktie Hassanzadeh, Michael Katz, Shirin Sohrabi, Kavitha Srinivas, Octavian Udrea
Proceedings of Application Showcase Program at the 30th International Conference on Automated Planning and Scheduling (ICAPS-2020)
Won first place

An AI Planning Solution to Scenario Generation for Enterprise Risk Management
S. Sohrabi, A. Riabov, M. Katz and O. Udrea
Proceedings of the 32nd Conference on Artificial Intelligence (AAAI-18), 2018

IBM Scenario Planning Advisor: Plan Recognition as AI Planning in Practice
S. Sohrabi, M. Katz, O. Hassanzadeh, O. Udrea and M. D. Feblowitz
Proceedings of Application Showcase Program at the 25th International Joint Conference on Artificial Intelligence (IJCAI-18), 2018

Scenario Planning for Enterprise Risk Management
S. Sohrabi, O. Udrea and A. V. Riabov
Proceedings of Application Showcase Program at the 27th International Conference on Automated Planning and Scheduling (ICAPS-17), 2017



Contact: Shirin Sohrabi (