How to Use ARC for Analysis: A Series

Decision-makers in governments, corporations, and think tanks all face high-stakes questions where a poor decision can have immense consequences. Of course while cliche at this point, the world is indeed getting more complex, and with that complexity comes increased uncertainty. To try and reduce uncertainty and present to decision-makers a realistic view of what may happen in the world, intelligence agencies for decades have relied on disciplined, structured analytic techniques to bring clarity, rigor, and foresight to their work.

ARC was built to bring a select type of these same intelligence methods, refined and proven in the most demanding environments, to anyone needing to try and reduce their uncertainty about the future world for which they are trying to compete and be successful in. Enabled by the unprecedented investment in AI and Large Language Models, we have built an entirely new class of analytic application that can perform these types of techniques in seconds while still facilitating a collaborative environment for humans to iterate and improve upon what we've asked AI to create. Unlike trying to use chat systems like ChatGPT or Perplexity, ARC monitors and updates current sources so you can understand how events are impacting the future in the context of your research question and also assesses the likelihood of future scenarios and indicators via AI and human crowd judgments.  

The foundations of structured analytic techniques

At its core, intelligence analysis is about structured thinking–breaking down problems, challenging assumptions, and systematically reducing bias. Over time the field has developed a comprehensive suite of methods called “Structured Analytic Techniques”(SATs) designed to help analysts move beyond gut instinct and unstructured brainstorming.

SATs were codified decades ago in works by Richards Heuer and Randolph Pherson from the CIA, and are now recognized as essential for any organization facing ambiguous, challenging problems.

These are some of the foundational techniques supported by ARC:

• Decomposition: Break down a complex issue or question into major drivers and observable signals that can inform your analysis on an ongoing basis. This makes thorny issues manageable, and even forecastable.
  
• Scenario Building: Generate a range of possible future scenarios, and assign probabilities to them to inform your analysis. Iterate as you add new sources and introduce your own expertise.
  
• Forecasting: Research shows that crowd-sourcing forecasts on a specific topic can consistently generate more accurate predictions than any one individual. Use ARC to formulate forecast questions that monitor key signals, and invite your colleagues to submit probabilities, generating unique and diverse insights.
  
• Red Teaming: Testing ideas adversarially, poking holes, and pulling them apart to find their weaknesses can create a much more rigorous analysis. Use ARC to seek out contrarian perspectives and identify cognitive biases in your work so that you can adjust, improve, and be confident in your product.
  

What makes intelligence-grade analysis robust is not just the use of the techniques, but the discipline with which SATs are applied. For example, these techniques are designed to be repeatable so that they can produce consistent results. They add transparency to analysis by enforcing documentation so you can easily spot errors or biases. Collaboration is supported to question assumptions and strengthen a final product. SATs are adaptable to new information, making it easier to update assessments and forecasts. ARC makes it easy to maintain the discipline for practicing these techniques as a routine part of your analysis.

It starts with the right question

No matter how sophisticated your tools or techniques, the quality of your analysis depends on the quality of your starting point: the research question. Your analysis in ARC will begin with a question - so, what guidelines should you consider?

Here are some tips for a strong analytic question that is clear, focused, and conducive for guiding your collection of information.

Balance depth and breadth to get useful signals without missing out on important context. ARC is designed to handle big, strategic questions, through decomposition, but it’s still important to be specific to get the information you need. Broadening and narrowing the scope, changing the focus, and rephrasing can help you define your question more suitably.

Tune your question to get information specific to your case	Instead of asking “How will the China-US relationship develop?”, try narrowing your focus: “How will China-US trade policies change over the next 4 years?”
Avoid going too narrow - ARC is built for the big questions	Replace a very narrow question like "Will the US have the fastest supercomputer?" for a bigger-picture question like "What country is best positioned to lead the world in supercomputing?" The latter would broaden the ARC analysis to help you understand the global competitive landscape, not just the U.S.

Align with stakeholder needs. Take the time to identify who will be using the insights you create, what they’re most concerned with, and the decisions and actions they need to take.

Identify your target audience	Even in an hierarchical organization, you may be writing for multiple audiences. Who will be the direct recipients of your work? Will they be using it to advise a third party - or to make an important decision?
Consider the stakeholders’ needs and provide analysis that enables them to take concrete steps	Asking “How will the ongoing trade war develop?” may be interesting, but if your stakeholder is a private investment firm, it could be more relevant to ask “Which countries are most likely to open their markets to US companies as part of the ongoing tariff war?”
Consider how consequential your analysis will be	Don’t shy away from unlikely possibilities if they’re likely to have an outsized impact. Asking about a potential Chinese military strike on Taiwan could provide useful analysis, even if the event itself is unlikely.

Keep your analysis forward-looking. While it’s important to look back to understand the past and even the present situation, backward-looking research is done for the purpose of creating future-focused insights.

Exclusively historical analysis can limit stakeholders’ ability to consider the full range of future possibilities.

Ask questions that make use of historical information while still looking forward. A question like “How will interest rates change over the next year?” will take into account past rates changes while still exploring how different possible economic futures could affect them.

Example research questions

ARC suggests research questions right when you sign in, but here are a few ideas across different industries:

• Geopolitics:
  
• • Which maritime "choke point" is at greatest risk of having its operations disrupted?
  • How will the outcome of Canada’s election affect Canada-US trade relations?
• Finance & Markets:
  
• • How will capital markets and dealmaking evolve over the next 3 years?
  • How will the Trump administration’s tariff policy affect the position of the US dollar as the global reserve currency?
• Science & Technology:
  
• • Which quantum computing architectures or platforms are most likely to become commercially viable?
  • Which new battery technologies will be most economically viable in the next 5 years?
• Business:
  
• • Which lines of business at a large insurance company are likely to see the greatest shifts in risk exposure due to climate change?
  • Which countries are most likely to open their markets to US companies as part of the ongoing tariff war?

________

This is an introductory post in our guide on how to use ARC for analysis. Next in our series:

• Decomposition: Breaking down the big questions
  
• Building actionable scenarios (coming soon)
  
• Integrating forecasting with decomposition and scenarios (coming soon)
• Combatting bias with red-teaming (coming soon)