Why the First Steps of the Analytical Process Matter More Than We Think
In my previous article about advancing analytical methods, I outlined the core steps of the analytical process. Today, I want to dive deeper into something I’ve been observing closely: the friction points in the early stages that can make or break strategic analysis before it even begins.
When analysts support strategy development, they’re not just doing research—they’re building the foundational infrastructure that will determine whether decision-makers get insights they can trust and act upon.
The Complexity Hidden in “Planning”
When we look at the Planning & Research Design phase, it appears straightforward on paper. But in reality, analysts face a fascinating challenge: they must create multiple, sometimes orthogonal structures.
“Analysts need to create one structure for strategy guidance and another for developing supporting facts and references—these can be completely different from each other.”
Here’s what actually happens: Analysts spend significant effort creating guidance for strategy development, including analytical models and work directives. They need input from decision-makers about goals and objectives. Should these focus only on their unit’s activities, or include broader social and environmental goals?
This requires important dialogues where the bigger business context directs how they set up their analytical work. They often create synopsis formats—one outlining strategy guidance, another focusing on developing supporting facts and references.
The main challenge? Capturing this complex structure digitally in a way that can actually be leveraged throughout the analysis work.
The Spreadsheet Phenomenon
What I find fascinating is how often Excel spreadsheets emerge in analytical workflows—not for calculations, but as metadata management tools.
“Spreadsheets provide an easy way to start working, but with multiple people creating multiple ad hoc structures, they don’t scale well.”
Analysts use spreadsheets to gather information, label content according to frameworks like PESTLE, and write comments. They copy document names or web links into cells. There’s something about the visual overview of having data grouped in rows and columns that they seem to need.
But here’s the problem: while spreadsheets offer an accessible starting point, they don’t scale when multiple people create ad hoc structures. And many organizations still rely on file shares with limited collaboration support.
The Data Capture Friction
Moving to source management and data collection, I’ve identified several specific friction points that analysts consistently encounter:
The Download Dilemma: When analysts find a PDF report on a webpage, they face immediate friction. They need to download it somewhere—usually the downloads folder—because they want to ensure it’s not lost and because specialized analysis tools often require file uploads.
The Selection Hesitation: There’s a fascinating psychological element here. Analysts hesitate to import documents when they’re not yet ready to commit that it’s relevant for analysis. This hesitation is both good and bad.
“This hesitation creates a curated digital library, which enhances quality. But it also means we might miss important diverse content that could reduce bias.”
The Duplicate Fear: Analysts worry about importing the same document twice. But “duplicates” are complex—they can involve file names, binary signatures, or content similarity across different file formats.
The AI Era Changes Everything
With AI and digital technology, we can actually handle large amounts of content while providing sophisticated tagging and labeling for different quality requirements. Moving from simple question formulation to structured hypothesis generation broadens collection scope by exploring different hypotheses.
“We need to remove friction in capturing different types of data and eliminate unnecessary hesitation in committing this data to storage.”
But there’s a crucial distinction to make about AI’s role. Research agents are now included in most public chatbots, which can be perceived as an easy way to automate everything. However, when it actually matters, we need to ensure all facts are correct and conclusions are sound.
This requires both data behind firewalls/paywalls and much richer context of the actual problem at hand—something only humans can provide through human-in-the-loop verification.
Supporting the Messy Reality
The analytical process is often depicted as a straight line (or in the intelligence cycle, a circular process) where each step happens sequentially. The reality is messier and more complicated, with small iterations and interactions between all steps, feedback loops, and the need to rephrase questions and key concepts.
“We need to support the actual nature of analysis work and remove unnecessary friction while helping analysts in these process steps as much as possible.”
This messy reality can actually provide a new level of transparency based on “logs” of what actually happened in the process. As Andrej Karpathy has noted, we can’t make these processes fully autonomous—human-in-the-loop verification is key.
We need to think much more about how to assist humans in doing these verifications effectively.
The Trust-Building Imperative
I believe it’s crucial to get the basic steps right by making them more efficient, rather than trying to automate everything and potentially failing—thus losing trust.
“In a world where facts are more scarce, they become more important. The road towards more facts runs through empowered analysts who get the tools they deserve.”
People today struggle with inadequate tools and want to work more professionally, but that can be really hard when only basic software is provided. This makes working with carefully selected methods more difficult.
AI is not only the problem but can also be the solution—helping analysts work very deliberately while still meeting tight deadlines.
Looking Forward
Supporting the early stages of the analytical process is one step toward empowering professional analysts. There are clear frictions that can be removed, and we’re trying to do our part in solving this important problem.
Because ultimately, the quality of strategic decisions depends on the quality of the analysis that supports them. And that quality is determined not just in the final synthesis, but in these crucial early steps where the foundation is laid.
The invisible profession of analysis deserves tools that match the importance of their work. The methodology matters, the tools matter, and getting the early steps right matters most of all.
What friction points do you encounter in the early stages of your analytical work? How are you balancing efficiency with rigor in your research processes?
Alexandra Kafka Larsson is the CEO and Co-founder of Parsd, a digital research platform that helps analysts create trusted insights. She has over 30 years of experience developing intelligence systems and methods across military, academic, and civilian domains.
