THE IDEA
When expertise is the answer
Not every situation requires experimentation. Some problems, however difficult, have analysable causes and knowable solutions. A bridge that’s failing has a structural problem that a structural engineer can diagnose. A financial irregularity has a cause that an auditor can find. A supply chain disruption has a bottleneck that a logistics expert can identify. These are complicated problems, and the right approach is to sense (gather information), analyse (apply expertise), and respond (implement the solution the analysis reveals).
This is the domain of good practice - not best practice (which belongs to simple, clear situations) but good practice: multiple right answers that require expertise to find. The doctor who runs tests and diagnoses. The mechanic who listens to the engine and identifies the fault. The architect who assesses the site and designs the building. Expert analysis isn’t guessing. It’s applying deep knowledge to a knowable problem.
The approach works when cause and effect are discoverable - when there’s a relationship between the parts that an expert can trace, even if it’s not obvious to a non-expert. The key is “discoverable.” In complicated domains, the answer exists and can be found with enough expertise. In complex domains, the answer doesn’t exist in advance - it emerges from interaction. Using the wrong approach for the wrong domain is where the expensive mistakes happen.
IN PRACTICE
When analysis gets you the answer
A manufacturing line is producing defective parts. The quality team gathers data: which parts, which machines, which shifts, which materials. They analyse the patterns: the defects cluster on one machine, during night shifts, using a particular batch of raw material. They respond: recalibrate the machine, adjust the night shift procedure, reject the material batch. The problem is solved. Expert analysis was the right approach because the system was complicated (many variables, but discoverable relationships).
A company’s website is loading slowly. The engineering team senses: they gather performance data, run diagnostics, examine server logs. They analyse: the bottleneck is a database query that scales poorly with traffic. They respond: optimise the query, add caching, adjust the architecture. The problem was complicated but knowable. The right expertise, applied to the right data, produced the right answer.
A patient presents with a cluster of symptoms. The doctor senses: medical history, physical examination, test results. Analyses: cross-references symptoms with possible diagnoses, runs differential diagnosis. Responds: prescribes treatment for the identified condition. If the condition is within the doctor’s expertise and the relationship between symptoms and disease is known, this approach works reliably. The expertise makes the complicated diagnosable.
WORKING WITH THIS
Knowing when this is your tool
Sense-analyse-respond is the right approach when the problem is complicated but not complex. The test: can an expert, given enough information, determine the answer? If yes, this is the right domain. Bring in the expertise, give them the data, and trust the analysis.
The critical skill is knowing when you’ve left the complicated domain and entered the complex one. The signals: experts disagree fundamentally (not about the details, but about the nature of the problem). The system responds to your intervention in unexpected ways. The analysis produces a confident recommendation that doesn’t work in practice. These are signs that the problem has complexity that analysis alone can’t handle.
The mistake to avoid: applying sense-analyse-respond to everything because it feels rigorous and professional. In complex domains, more analysis doesn’t produce better answers - it produces more confident wrong answers. The discipline is matching the approach to the domain: analysis for the complicated, experimentation for the complex.
THE INSIGHT
The line to remember
Expert analysis is powerful when the answer exists and can be found. It’s dangerous when it’s applied to problems where the answer can only emerge from interaction with the system itself.
RECOGNITION
When this is in play
You’re in sense-analyse-respond territory when the problem is difficult but knowable. When experts exist who have solved similar problems before. When gathering more information genuinely helps. When the system is complicated enough to need expertise but stable enough that the analysis will still be valid when you act on it. You’ve left this territory when the experts keep being surprised, when the analysis doesn’t survive contact with reality, and when more data produces less clarity rather than more.