How to Build Expertise: From Experience to Deliberate Improvement
How to build expertise is often answered with a number: practise for years, accumulate thousands of hours and eventually become an expert. Time unquestionably matters because difficult fields require repeated exposure, knowledge and practice. But elapsed time is not the mechanism that turns experience into expertise.
A person can perform essentially the same job for ten years while repeating methods learned during the first year. Another person can progress much faster because difficult assignments are followed by feedback, mistakes are analysed rather than forgotten and each cycle of work changes how the next problem is approached.
Expertise is better understood as reliably superior performance within a particular domain. Experts do not merely possess more facts. They notice meaningful patterns that novices overlook, organise knowledge around deeper principles, recognise which information matters, retrieve relevant knowledge efficiently and choose actions with greater precision.
The National Academies' synthesis of expertise research makes this distinction particularly clear. Experts acquire extensive domain knowledge, but what matters is how that knowledge is organised. They recognise meaningful patterns, connect individual facts to underlying concepts and understand the circumstances in which particular knowledge or procedures apply. (nationalacademies.org)
Becoming an expert is therefore not simply a process of accumulating information or repeating a task.
It is a process of changing what you notice, how you represent problems and how effectively you learn from the consequences of your decisions.
Expertise Begins With a Defined Domain and an Organised Knowledge Base
“Become an expert in business” is not a useful learning objective.
Neither is “master technology”, “become good at finance” or “learn marketing”. These areas contain too many different problems for performance to be measured coherently.
Expertise develops within more specific domains: infrastructure-project financial modelling, B2B exhibition sales, employment-law research, paediatric emergency nursing, industrial product design, cybersecurity incident response or paid-search advertising.
The narrower definition creates a learning advantage because recurring decisions become visible. You can identify which problems repeatedly appear, which mistakes matter, which knowledge experienced practitioners rely on and what distinguishes good performance from merely acceptable work.
That does not mean expertise must remain permanently narrow. A highly skilled professional often expands into adjacent areas over time. But depth usually develops around a defined class of problems before it becomes useful elsewhere.
This is consistent with decades of expert–novice research. The National Academies notes that experts are not simply general-purpose thinkers who possess superior reasoning strategies applicable everywhere. Their advantages arise largely from detailed and organised knowledge within a particular domain. (nationalacademies.org)
A strong learning programme should therefore begin by creating a map of the field.
That map can include the field's core concepts, recurring tasks, important tools, accepted standards, common failure modes, important edge cases and the relationships among them. Someone learning recruitment, for example, might map sourcing, screening, interview design, assessment validity, compensation, candidate experience, employment law and workforce planning. Someone developing expertise in sales might map prospecting, qualification, discovery, value creation, objection handling, negotiation, account economics, closing and retention.
The point is not to memorise a giant taxonomy before doing any work.
It is to prevent learning from becoming a random collection of tips.
Experts tend to organise knowledge around underlying concepts or “big ideas”, while novices are more likely to focus on surface features. In classic studies reviewed by the National Academies, physics experts categorised problems according to underlying physical principles, whereas novices were more likely to classify them according to visible features of the problem. (nationalacademies.org)
The same distinction appears in professional work.
A novice salesperson may think, “This customer asked for a discount.”
An experienced salesperson may instead recognise, “This is a perceived-risk problem disguised as a price objection.”
A novice programmer may see an error in one function.
An expert may recognise a poorly defined system boundary that will generate similar errors elsewhere.
The expert has not merely learned more answers.
They have learned how to represent the problem differently.
Practice Matters, but “Just Put In the Hours” Is Too Simple
The research associated with psychologist K. Anders Ericsson made deliberate practice one of the most influential ideas in discussions of expertise.
The concept is frequently simplified into the claim that enough hours of practice eventually create exceptional performance. That misses important parts of the original theory.
Ericsson's stricter definition involved training activities specifically designed to improve particular aspects of performance, typically guided by a qualified teacher or coach who could identify errors, assign appropriate tasks and provide informative feedback. The activities were demanding, focused and adjusted as the learner improved. (frontiersin.org)
That is very different from simply doing the activity repeatedly.
A salesperson making the same pitch to fifty customers may accumulate experience. A salesperson who reviews unsuccessful calls, identifies where discovery failed, develops a revised questioning sequence, tests it and has an experienced manager critique the result is practising in a much more improvement-oriented way.
A lawyer completing familiar filings is working. A lawyer taking a weak argument, comparing it with stronger examples, identifying the specific reasoning failure and rewriting it under expert review is deliberately trying to improve a component of performance.
A programmer writing routine code is gaining exposure. A programmer studying recurring architectural mistakes and repeatedly designing systems that force practice of that weakness is doing something closer to purposeful improvement.
The distinction matters because work and practice optimise for different things.
Work usually optimises for producing an acceptable result efficiently.
Improvement practice temporarily sacrifices efficiency in order to change capability.
This is one reason experience alone is unreliable as a measure of expertise. Once people become competent enough to perform routine tasks successfully, ordinary work can stop forcing adaptation. Procedures become automatic, familiar solutions are reused and mistakes that do not produce immediate consequences may persist unnoticed.
Deliberate improvement deliberately interrupts that comfort.
It asks: Which part of my performance is still weak enough to deserve focused attention?
Deliberate Practice Is Important, but It Does Not Explain Everything
Popular accounts sometimes present deliberate practice as though it explains nearly all differences between ordinary and elite performers.
Research is more complicated.
A widely cited 2014 meta-analysis by Brooke Macnamara, David Hambrick and Frederick Oswald found that accumulated practice explained different proportions of performance differences across domains: about 26% in games, 21% in music, 18% in sports, 4% in education and less than 1% in the professional studies included in their analysis. The authors concluded that practice was important but could not by itself explain most differences in attained performance. (sagepub.com)
Ericsson later disputed how broadly that research had defined deliberate practice. He argued that many included activities did not meet his stricter criteria involving individualised diagnosis, training specifically designed to improve a weakness and expert guidance. Reanalysing the evidence under narrower criteria produced substantially stronger associations between qualifying practice and performance. (frontiersin.org)
The disagreement is useful because it prevents two opposite errors.
The first is believing that talent makes structured improvement almost irrelevant.
The second is believing that practice hours alone determine eventual performance.
Fields differ. People differ. Opportunities differ. Access to coaching, training environments, prior knowledge, physical characteristics, motivation and the quality of practice all matter. Some activities also provide much clearer feedback than others.
A chess player can often determine relatively quickly whether a decision produced an inferior position. A violinist can repeatedly practise one passage and hear specific technical errors. A surgeon, manager or policy analyst deals with messier situations in which outcomes may appear months later and are influenced by many people and variables.
That makes expertise in professions particularly difficult to develop.
The lesson is not that deliberate improvement does not apply to professional work.
It is that professionals often have to build the feedback environment themselves.
The Fastest Improvement Usually Comes From Finding the Bottleneck
Trying to “improve everything” rarely produces focused learning.
Performance is usually constrained disproportionately by one or two weaknesses at a particular stage of development.
A salesperson may know the product extremely well but fail to diagnose what a customer actually values. A writer may research deeply but structure arguments poorly. A manager may understand operations while avoiding difficult conversations. A programmer may write clean code while designing fragile system architecture.
The useful question is therefore not simply:
What should I study next?
It is:
What weakness currently prevents my performance from improving?
Several diagnostic questions can help:
What error keeps appearing in my work? Which tasks require disproportionate help or revision? What can stronger performers do reliably that I cannot? Which decision am I still making largely through intuition or guesswork? Where do outcomes repeatedly disappoint me despite significant effort?
Once identified, the bottleneck can become a practice target.
Suppose a B2B salesperson discovers that prospects respond positively to presentations but rarely advance afterward. The temptation is to learn better presentation techniques. A review of conversations may instead reveal that the salesperson is presenting solutions before establishing urgency, authority or business impact.
The real bottleneck is discovery.
Practice should therefore shift toward diagnosing customer needs, quantifying problems and asking better follow-up questions—not toward polishing slides.
This is what makes feedback valuable.
It converts poor outcomes from disappointment into information about which capability needs work.
Feedback Must Be Specific Enough to Change the Next Attempt
“Excellent presentation.”
“Needs improvement.”
“You need to be more confident.”
These comments may be encouraging or discouraging, but they are poor learning tools because they do not specify what behaviour should change.
Diagnostic feedback answers questions such as:
What specifically worked? What specifically reduced performance? What information was missed? At what point did the decision go wrong? What should be done differently on the next attempt?
The best source of feedback also depends on the problem.
A customer is authoritative about whether a proposal convinced them, but may not know why its structure failed. An experienced sales manager may recognise that the proposal was introduced before the customer's needs were sufficiently understood.
A reader can report that an article became confusing halfway through. A strong editor can diagnose that two distinct arguments were mixed inside the same section.
Outcome data, customer reactions, expert review, peer comparison and self-review can therefore provide different information.
None is perfect in isolation.
Experts themselves also require feedback because experience can create confidence faster than accuracy. A professional who repeatedly makes decisions without reliable information about the outcome can become highly experienced in repeating an error.
One practical method is to conduct short after-action reviews:
What did I expect to happen? What actually happened? Where was my reasoning accurate? Where was it wrong? Was the error caused by missing information, weak execution or a poor mental model? What will I change next time?
This turns completed work into training material.
Without such review, the same case disappears into memory as soon as the next task begins.
Study How Experts Think, Not Merely What Their Finished Work Looks Like
Expert output hides the process that produced it.
A polished financial model does not show which assumptions the analyst challenged first. A successful sales conversation does not reveal why the salesperson ignored one customer comment but pursued another. A good diagnostic decision does not show which possibilities the clinician considered and rejected.
Simply copying final output can therefore teach style without teaching judgment.
The more valuable questions are:
What did the expert notice first?
Which information seemed important and which was ignored?
How was the problem categorised?
Which risks were considered?
What alternatives were rejected?
What previous cases made this situation look familiar?
What new evidence would have caused a different decision?
These questions reveal mental models.
Research on expertise repeatedly finds that experts recognise meaningful relationships that novices do not. Their knowledge is organised in ways that allow them to identify deeper structure instead of reacting primarily to visible surface features. (nationalacademies.org)
This is why observing an expert can be far more useful when the expert thinks aloud.
Instead of seeing only the decision, the learner sees the sequence:
I noticed this signal. That made me consider these possibilities. I ruled this one out because of this evidence. This case resembles another pattern, but this feature is different, so I would not use the normal solution.
That reasoning is where much professional expertise lives.
Build a Case Library So Experience Becomes Searchable Knowledge
Professionals often encounter valuable cases and then lose most of the learning they contain.
A difficult negotiation ends. A campaign performs unexpectedly. A hiring decision succeeds or fails. A software deployment breaks. A client objection appears for the first time.
The event is remembered vaguely, but the exact reasoning is not documented.
A case library turns experience into reusable evidence.
For significant cases, record a short structure:
Context: What situation were you dealing with?
Signal: What information seemed most important?
Decision: What did you decide and why?
Outcome: What happened?
Error or surprise: What did you misunderstand?
Lesson: What should you notice sooner next time?
Pattern: Which future situations might this case apply to?
Over time, this becomes far more valuable than a collection of generic notes.
Experts rely heavily on pattern recognition, but useful pattern recognition requires enough well-encoded examples. The National Academies notes that experts organise experience into meaningful structures that help them recognise which information is relevant and which principles apply. (nationalacademies.org)
A case library helps build that structure deliberately.
It also reduces survivorship bias in memory.
People naturally remember dramatic wins and painful failures. Quietly mediocre decisions disappear. If cases are documented systematically, you can discover patterns that intuition alone may miss.
Perhaps your largest clients did not come from your best presentations but from unusually strong discovery conversations.
Perhaps projects you believed were delayed by technical complexity were actually delayed because requirements were unclear at the beginning.
Perhaps forecasts fail repeatedly in the same type of market condition.
Written cases make those patterns visible.
Increase Difficulty Without Turning Challenge Into Chaos
Once routine work becomes easy, continued repetition produces diminishing learning.
The solution is not to jump immediately into problems so difficult that failure becomes uninterpretable.
Improvement happens most effectively when difficulty increases enough to expose the next weakness while still allowing the learner to understand what happened.
A salesperson comfortable managing ₹5 lakh accounts might begin handling more complex organisations with multiple decision-makers. An analyst who can build standard models might add uncertain scenarios or unfamiliar financing structures. A writer comfortable with explanatory articles might take on subjects requiring contested evidence and source reconciliation.
The new task should create productive difficulty.
If everything succeeds automatically, there is little reason for the current mental model to change.
If everything fails simultaneously, it becomes difficult to know which skill caused the failure.
Mentors and good managers are especially useful here because they can calibrate challenge. They can recognise when a task is slightly beyond current capability rather than hopelessly beyond it.
This is one reason access to strong coaching can accelerate development disproportionately.
The coach does not merely provide answers.
A good coach helps select the next problem worth struggling with.
Teaching Exposes the Difference Between Recognition and Understanding
Reading an explanation can create a strong feeling of familiarity.
Explaining the same concept without looking at the source reveals whether that familiarity has become understanding.
Teaching is useful because it forces knowledge into a coherent structure. You have to define terms, connect causes and effects, anticipate confusion and respond to questions that may expose assumptions you had not noticed.
If the explanation repeatedly collapses into jargon, vague analogies or “it just works that way”, the underlying model is probably incomplete.
This does not mean people must become teachers to become experts.
In fact, expertise does not automatically make someone a good teacher. The National Academies specifically notes that knowing a field deeply does not guarantee the ability to communicate that knowledge effectively to novices. (nationalacademies.org)
But attempting to explain is still a powerful diagnostic tool.
A useful exercise is to take one important concept from your field and explain it at three levels:
to a beginner;
to a competent practitioner;
to another expert.
The beginner explanation tests whether you understand fundamentals.
The practitioner explanation tests whether you can connect principles to decisions.
The expert explanation exposes precision, exceptions and unresolved questions.
Different levels reveal different weaknesses.
Measure Performance, Not Just Learning Activity
Courses, books, certificates and hours studied measure activity.
They do not necessarily measure capability.
A person can complete twelve sales courses without improving conversion. A programmer can earn certifications while still making poor architectural decisions. A writer can read dozens of books while requiring the same amount of editing.
Expertise requires evidence closer to real performance.
Useful indicators depend on the domain: forecast accuracy, error rates, conversion quality, client retention, defect rates, time to diagnosis, percentage of work accepted without major revision, recovery from failures or expert review scores.
Metrics should be chosen carefully because a single measure can distort behaviour.
A salesperson judged only on the number of deals may close poorly qualified customers who later cancel.
A support team measured only on response time may answer quickly but resolve problems badly.
A writer judged only on traffic may produce sensational content that damages trust.
Quantitative evidence should therefore be combined with case quality and expert judgment.
The objective is not to create a perfect numerical score for expertise.
It is to make improvement falsifiable.
If you believe your skill has improved, something about the quality, reliability, speed or complexity of your performance should eventually demonstrate it.
Deep Expertise Needs Adjacent Knowledge Without Becoming Shallow Everywhere
Expertise is domain-specific, but professional problems rarely remain inside neat disciplinary boundaries.
A technically outstanding engineer may still need to understand cost, customer requirements, regulation and organisational incentives. A salesperson selling industrial technology needs enough technical understanding to diagnose genuine customer problems. A doctor needs communication skills in addition to clinical knowledge.
This is why the familiar T-shaped model remains useful.
The vertical part represents deep capability in one field.
The horizontal part represents enough understanding of neighbouring areas to work effectively with other people and recognise when another discipline changes the problem.
Breadth without depth produces superficial generalism.
Depth without breadth can create technically sophisticated solutions that fail in the real system around them.
The balance should not be interpreted as learning everything equally.
Choose a field in which you intend to become unusually strong, then acquire enough adjacent knowledge to understand the context in which that expertise must operate.
A Practical Expertise Cycle Turns Work Into Continuous Training
A useful improvement system does not require waiting years before evaluating progress.
It can operate in short cycles.
Start by choosing one performance bottleneck and collecting baseline examples. If you want to improve sales discovery, review several existing calls and identify how often you uncover business impact, urgency, decision authority and constraints.
Then practise that specific capability deliberately. Study strong examples, create targeted exercises and apply the revised behaviour during real conversations. Seek weekly feedback from someone qualified to diagnose errors.
After several weeks, increase difficulty. Use the skill with larger or more ambiguous prospects. Compare the new conversations with the original baseline.
Finally, review what changed.
Which mistakes disappeared? Which remain? Did the improvement create a new bottleneck? What should the next cycle target?
A twelve-week version might look like this:
Weeks 1–2: diagnose one bottleneck and establish baseline performance.
Weeks 3–8: practise that capability repeatedly with frequent feedback.
Weeks 9–10: apply it in harder or less familiar cases.
Weeks 11–12: review outcomes, update the case library and choose the next constraint.
The precise calendar is not important.
The loop is.
Performance → evidence → diagnosis → focused practice → feedback → harder application → review.
Repeated over years, that loop changes what experience means.
Ten years of experience can represent ten years of adaptation.
Or it can represent the same year repeated ten times.
Expertise Is an Improvement System, Not a Permanent Status
Experts know far more than novices within their domains, but genuine expertise also changes the relationship a person has with uncertainty.
A novice may not know what they do not know.
A competent practitioner begins recognising exceptions.
An expert often becomes highly aware of where rules stop working.
The National Academies describes metacognition—the ability to monitor one's own reasoning, question an initial interpretation and adjust strategy—as an important part of competent and expert performance. (nationalacademies.org)
That matters because fields change.
Technology changes.
Regulation changes.
Evidence changes.
Customers change.
Competitors change.
An expert who treats previous success as permanent proof of superior judgment can become less accurate precisely because experience creates overconfidence.
The durable advantage is therefore not reaching a point where learning ends.
It is building a system that makes continued correction normal.
Define the domain narrowly enough to practise it. Build organised knowledge rather than collecting disconnected tips. Identify the weakness currently limiting performance. Create practice specifically around that weakness. Seek feedback detailed enough to change behaviour. Study how excellent performers make decisions, not merely what their finished work looks like. Record cases. Increase difficulty as competence grows. Explain what you know. Measure real outcomes.
And remain careful with simple formulas.
Practice matters enormously, but hours are not all equal. Experience matters, but experience without feedback can automate mistakes. Knowledge matters, but disconnected facts do not produce expert judgment.
Expertise emerges when repeated experience continually changes the mental models used to interpret the next case.
That is why the most useful question is not:
“How many years have I done this?”
It is:
“What can I now notice, understand and do reliably that I could not do before—and what am I deliberately improving next?”
