Learning Velocity: Why Some People Double Their Skills Every Two Years

Between 2018 and 2023 I watched two colleagues — similar age, similar starting point, both analysts at the same consulting firm — take measurably different career trajectories. One is now a director at a top-tier fund, running a portfolio worth nine figures. The other is still an analyst, at a different but comparable firm, earning roughly 1.4x what she earned in 2018. Their IQ scores, if anyone were to measure them, would almost certainly be close. Neither is working noticeably harder than the other. The difference is learning velocity — the rate at which they've been converting experience into compounding professional capability.

Learning velocity is an underrated variable in how careers play out. Most people assume that the rate at which they learn and grow is a fixed personal characteristic, determined by native intelligence and motivation. The research on expert development — most of it coming out of Anders Ericsson's lab at Florida State and K. Anders Ericsson's follow-on work summarised in Peak (2016) — suggests the opposite. Learning velocity is highly teachable, and the specific habits that produce it are well-documented. They're just rarely practised, because they're uncomfortable, and because most professional environments reward competence rather than growth.

What Learning Velocity Actually Measures

A working definition: learning velocity is the rate at which you're expanding the range of professional problems you can handle competently, relative to the time you've been in the field. High learning velocity means your third year feels qualitatively different from your first year — you can handle classes of problem, or levels of responsibility, that the you-of-twelve-months-ago could not. Low learning velocity means year three feels like an extended version of year one — you're more efficient, you know more people, but the actual scope of what you can do hasn't meaningfully shifted.

The key word is qualitatively. Quantitative growth is cheap. Handling more of the same work you handled last year, faster, is not learning velocity — it's efficiency gain. Real learning velocity means taking on work that would have been beyond you two years ago, and doing it well. You can only do that if the underlying capability has shifted.

Most people confuse efficiency gain with learning velocity, and then wonder why their careers plateau in year seven or eight. Efficiency is linear. Learning velocity compounds.

The Four Variables That Predict It

Ericsson's research, and related work by Robert Bjork at UCLA on "desirable difficulties," converge on four specific habits that separate rapid learners from slow ones. None of them are about raw ability.

1. Deliberate practice on edge-of-ability tasks

The single biggest variable. Most professional work, after the first year or two, is done at the comfortable middle of your ability. You're doing what you already know how to do. You're good at it. It doesn't stretch you. This is pleasant and it produces flat learning curves.

Rapid learners deliberately seek out work at the edge of their current capability — work where failure is genuinely possible, where they have to stretch to produce a competent result. In Ericsson's framing, the task must be at a level where success requires focused effort, not one where success is guaranteed by routine.

Practically: the analyst who volunteers to build the first draft of the client memo, despite never having written one before, is putting herself on the learning curve. The analyst who only does work he's already mastered is not. Over five years, the gap between these two compounds into a large capability difference.

2. Rapid feedback loops

Learning velocity is proportional to the speed and specificity of the feedback you're getting. In a slow-feedback environment — long projects, infrequent reviews, ambiguous success criteria — you can spend years getting comfortable with habits that are subtly wrong. In a fast-feedback environment — daily work with explicit review, clear criteria for success — you converge on better practice quickly.

The move for the ambitious professional: deliberately find or construct fast-feedback loops. Write a short essay. Post it publicly. Get feedback in days, not months. Pitch an idea to a senior colleague. Get a reaction inside a week. Take on work where the result is measurable in a quarter rather than a year. Feedback that arrives later than six months is almost useless for learning — by the time you get it, you've moved on and can't meaningfully connect it to the decisions you made.

3. Explicit reflection on mistakes

The default response to failure is emotional processing — feeling bad about it, then moving on. The rapid-learner response is structured diagnosis. What specifically went wrong? What specifically did I miss? What would I do differently if I faced the same situation tomorrow?

This is the decision-journal habit, from the Kahneman / Tetlock tradition. It sounds obvious and almost nobody does it. Because doing it well requires a tolerance for sitting with your mistakes in detail, which is uncomfortable. The discomfort is precisely what produces the learning — by forcing you to stay with the specifics until you've extracted the lesson, rather than letting the emotional protection system paper over them.

Research on expert forecasters (Tetlock's Good Judgement Project) shows that explicit post-mortem on predictions is one of the strongest predictors of forecaster improvement. The same principle applies to any professional domain where outcomes are observable. If you're not doing written post-mortems on your meaningful decisions, you're not learning from them in the way you could be.

4. Exposure to people meaningfully better than you

Your learning rate is heavily constrained by the calibre of people you spend professional time with. If everyone around you is at your level or below, your sense of what "good" looks like is capped by them. You'll ceiling out at their ceiling.

The single highest-leverage career move in your first ten years, as described in different language by Peter Thiel, Marc Andreessen, and many others, is arranging to work alongside people who are measurably better than you at the thing you're trying to learn. You'll accelerate simply by being able to see what better looks like, and by absorbing — often unconsciously — the specific habits and judgements that produce it.

This is often a job-selection decision more than a within-job decision. The firm, team, or mentor you're working under matters more than your role title or your pay bracket. Early-career, optimise for learning adjacency to the best people you can plausibly work near. Compensation catches up later.

The Habits That Kill Learning Velocity

Equally important is what stops you learning fast. Common failure modes:

• Staying in a role where you're the smartest person in the room. Flattering, comfortable, and terrible for your growth. You stop having anyone to learn from. Leave.
• Avoiding feedback because it's uncomfortable. The senior person who told you your draft was weak is the most valuable person in your professional life. Most people avoid them. Rapid learners seek them out.
• Optimising for visible output at the expense of hard learning. There's always a way to look busy without growing. Writing the fifth variation of a document you already know how to write produces visible output and no learning. The visible work can be the enemy of the important work.
• Protecting your ego by not trying things you might fail at. Stretch assignments, public presentations, writing in domains you're new to — all involve potential failure. The people who avoid all of these never make the mistakes that produce the biggest lessons.
• Never slowing down to think about what you're learning. Learning requires processing. If you're perpetually at 100% utilisation on delivery, there's no cognitive space for the patterns to consolidate. The weekly reflection, the quarterly review, the annual retreat — these aren't luxuries. They're the infrastructure that turns raw experience into actual capability.

The Second-Derivative Test

A simple honest test: look at the work you're doing this quarter. Look at the work you were doing 24 months ago. Are they qualitatively different, or just a faster version of the same thing? If the latter, your learning velocity is low and you need to do something about it.

The "something" is usually one of three moves: take on a stretch assignment in your current role that would be clearly beyond you; find a mentor or peer meaningfully better than you and arrange for regular substantive contact; or, in more extreme cases, change roles. The first is cheapest. The last is most effective. In between is the professional-development middle ground most people occupy, which works but compounds more slowly than they realise.

The Long-Run Compounding

Learning velocity compounds non-linearly. Consider two professionals. One is learning at 10% a year — modest growth, roughly keeping pace with a corporate environment. The other is learning at 25% a year — deliberately practising, seeking feedback, exposed to better peers. After five years, the first has grown their capability by about 61%. The second has grown theirs by about 205%. The gap gets bigger year over year.

These numbers are illustrative, not measured. The actual compounding effect is hard to quantify because capability doesn't have a clean metric. But the qualitative observation is visible to anyone who's watched a cohort of peers develop over a decade. Some of them, at year ten, are doing work that's qualitatively beyond what anyone in the cohort could have done at year three. Others are doing a more polished version of what they were doing at year three. The difference wasn't talent. The difference was the specific habits, compounded quietly, over several thousand working days. It was learning velocity.