Talent is where conviction is heading. The most reliable forecast in any economy is not a figure on a balance sheet. It is a resume, filed eighteen months before the press release. Learn to read who is quietly changing jobs, what they are studying, and which job titles did not exist three years ago, and you can see the next growth sector before the money admits it is there.
Money is a lagging indicator of conviction. Talent is a leading one. The interval between the moment skilled people commit to a problem and the moment capital agrees with them is a forecast that costs almost nothing to read, which may be why almost nobody reads it. What follows is how to read it, why South Africa turns out to be the most revealing place on earth to test it, and the four ways it will mislead you if you trust it without thinking.
Sit in enough South African boardrooms and a pattern starts to repeat itself. The investment case arrives about a year after the best people did. By the time a sector is obvious enough to fund, its pioneers have been working in it for eighteen months, the recruiters have been circling for twelve, and the analysts are writing it up as though they found it. We keep calling capital allocation the smart money. Most of the time it is the rear view mirror with a research budget.
There is a mechanical reason for the delay, and it has nothing to do with intelligence. A fund deploys against a thesis written years before the money actually moves, runs a diligence process that rewards whatever fits in a spreadsheet, and answers to limited partners who would rather be late and right than early and exposed. Capital cannot commit until a committee agrees, and committees are built to agree slowly. A person carries none of that ballast. A good engineer does not wait for a sector to be de-risked before she joins it; she joins because she has sat in the meetings and read the room, and the room is full of problems she finds more interesting than the ones already on her desk. One resignation is noise. Forty thousand of them pointed the same way are a forecast that no analyst commissioned and no investment committee ever signed.
The academic record is unusually blunt about this. Tracking America's most admired employers across a quarter of a century, the financial economist Alex Edmans found that a portfolio of the best companies to work for beat the market by roughly three and a half percent a year. The detail that should nag at you is not the size of the number. It is the visibility. That list was never proprietary; Fortune printed it every January, on a public newsstand, and the market still took four to five years to price what it implied. You can argue about the mechanism. Perhaps the best workplaces simply belong to better companies and the contented workforce is along for the ride, and Edmans spent much of the paper closing that door. Take even the modest reading and it unsettles: when a public, named, annually updated signal about talent inside one company takes years to be valued, a quiet signal about talent moving across a whole sector is, for any practical purpose, a secret.
Push the question back far enough and it stops being about share prices at all. In 1991, Kevin Murphy, Andrei Shleifer and Robert Vishny asked a stranger version of it: where does a society send its ablest people, and does the choice feed back into how fast it grows. Their reading of the data said yes. Economies that steered talent toward engineering grew faster; those that funnelled it into rent seeking professions, the business of dividing wealth rather than enlarging it, grew slower. The fair objection is that this is one regression across a cross section of countries, and wealthy countries can simply afford more engineers, so the direction of cause is genuinely arguable. Yet the pattern has aged unnervingly well, and it leans the uncomfortable way. Where a country points its talent is not a souvenir of growth already banked. It behaves like a cause of the growth still to come.
So the money lags. The sharper question is why it lags so predictably, and the answer is that an economy does not switch on. It accretes in layers, and the layers surface in a fixed order. A field needs a vocabulary before it needs a balance sheet. You could not have put a prompt engineer on an org chart in 2019, because the words for the work did not exist yet; by the time the title was common enough to recruit against, the first movers were three jobs deep into the thing. Words, then the skills, then the jobs themselves, then the funding, then at last the revenue everyone can finally see and cite. The ladder below stylises that sequence. Read the month figures as a rough heuristic rather than a measurement, because the true lead time swings wildly from one sector to the next. What holds is the shape. Read from the top and you are early; read from the bottom, where the dashboards are clean and the figures audit to the decimal, and you are studying the past in high resolution.
Tap any rung. The higher you read, the more warning you get, and the messier the data looks. The numbers get cleaner as the opportunity gets deader.
How to read this: each rung is a stage an economy passes through, ordered top to bottom from earliest to latest. The right-hand figure is the typical lead time that stage still gives you before it shows up as revenue. Reading from the top means acting on vocabulary and skills while the signal is faint; reading from the bottom means waiting for capital and output figures, which are clean and precise but arrive last.
If the ladder tells you where to look, the gap tells you how much time the looking buys. Pick a sector busy remaking the world and watch the people arrive before the money believes them. Drag the line to read the distance between the two curves at any point in the adoption. Where the gold runs ahead of the blue, talent has committed and capital has not, and that shaded wedge is the only quantity on the chart you can still act on. A caveat the chart will not show you: the gap is an opportunity, not a promise. Talent pours into plenty of things that never pay off, and a chart cannot tell conviction apart from a stampede. Reading it well is mostly the work of the section after this one, learning when the current is lying to you.
When the wedge is at its widest, the people have arrived and the money has not. The gap is the alpha. When it shuts, the money has caught up and the edge is gone, priced into every term sheet in the room.
How to read this: the gold line tracks talent flowing into the sector; the blue line tracks capital following it. Both are read as a share of eventual inflow, left to right across the adoption curve. The shaded wedge between them is the foresight gap: where gold sits above blue, talent has already committed and capital has not yet caught up. A wider wedge means more warning still available; a wedge that has closed means the advantage is gone.
None of this needs proprietary data or an expensive terminal. The public record, the World Economic Forum's survey of employers and LinkedIn's Economic Graph, is already loud enough to act on; the redrawing of work to 2030 sits in profiles and job titles years before it surfaces in the output figures that make the news. One honest wrinkle before the numbers seduce you. The roles growing fastest by raw headcount are not the ones on the magazine covers; the WEF's own tables put farmhands, delivery drivers and care workers at the top of the absolute pile. Growth in numbers and growth in opportunity are two readings of the same table, and mistaking one for the other is the most reliable way this whole discipline goes wrong.
Direction of movement per WEF Future of Jobs 2025. Bar lengths show relative ranking within each column, not absolute headcount. The tell is in the titles: in 2025 the two fastest rising roles on LinkedIn in the United States were both built entirely around artificial intelligence, names that barely existed a decade ago.
South Africa is where the comfortable version of this story breaks, and where the useful one begins. It is also, for an analyst, the cleanest natural experiment on the planet. Hold the two headline numbers in one hand at the same time. They look like a contradiction. They are actually a proof.
Hold a labour surplus and a skills shortage in the same hand and the contradiction dissolves into something more useful. If both are true at the same instant, the shortage cannot be about the number of bodies; it is about the gap between what those bodies can do and what the economy will pay to find. Read coldly, the 8.1 million people out of work are the bill for forecasts the country declined to make a decade ago, the compounded cost of training pipelines nobody built. The shortage is the next forecast, arriving free of charge, in the one language a treasury cannot help but hear: a line of employers holding money they are unable to spend. It is the rare prediction that turns up with a price tag already attached.
The scarce skills register looks like a list of human resources complaints. It is closer to an industrial policy printed in the negative. Engineers, data scientists, electrical and software talent, artisans, renewable energy specialists: the jobs a country cannot fill this year sketch, with some precision, the sectors that will define it in five. Finance and construction were among the handful of sectors still adding formal jobs into late 2025; the energy transition is already bidding for skills the local universities have barely started to produce. The talent is voting, with its feet and its course enrolments. Whether anyone tallies that vote before the result becomes obvious enough to be worthless is, on the evidence so far, an open question.
There is a human texture the dashboards miss. The graduate who emigrates is not only a line lost on a spreadsheet. The destination she chooses is information. It tells you which skills the global market is bidding hardest for, and therefore which capabilities South African firms are about to fight to keep. The matriculant who walks away from a generic commerce degree toward a data qualification is making a private forecast about where the work will be. Aggregate ten thousand of those private forecasts and you have something no capital expenditure model will hand you for another two years.
Widen the lens past one country and the signal sharpens instead of blurring. Africa is now home to roughly 716,000 professional software developers, more than the state of California, with South Africa holding the largest single share. The figure that ought to reorganise a policymaker's morning is the next one: something close to two in five of those developers already draw a salary from a company headquartered off the continent, and most have never left home to do it.
Look hard at that 38 percent, because it quietly breaks the instrument we use to measure this kind of loss. Brain drain, as the phrase was coined, required a departure: a flight booked, a border crossed, a name struck from a tax roll. This version requires none of it. A developer in Nairobi or Cape Town keeps her flat, her citizenship and her Friday routine, and ships the full economic value of her work to an employer five time zones away. Her country keeps the citizen and loses the output. Remote work was sold, not least to African governments, as the thing that would finally let talent stay. It let the talent stay and exported the contribution, and because nobody passes through an airport, every statistic built to count brain drain now reads close to zero while the drain itself speeds up. The gauge went blind at the precise moment the thing it measures got worse.
Turn the same fact over and it becomes a gift. The destinations of all that exported work form a live, free, constantly refreshing map of which skills the planet is bidding hardest for. You no longer wait for an emigration figure to confirm that the world wants your data engineers; you watch them being hired in real time, from Berlin and Dubai and Toronto, while they sit in Johannesburg. None of this is hypothetical. When Stripe paid a reported two hundred million dollars for the Lagos company Paystack in 2020, it was not discovering African fintech. The engineering had been compounding in Lagos for five years; the cheque was capital turning up, characteristically late, to a market that talent had already built. For a founder, that map is a recruiting list. For a policymaker, it is the demand side of an industrial policy, delivered without commissioning a single survey.
If reading talent flows were just a matter of counting movement, every recruiter would be a prophet and this brief would be a spreadsheet. The signal is real, but it is noisy, and the people who make money from it are the ones who know exactly where it lies to them. Four assumptions worth retiring before you trust the current with a budget.
The founders move first. Capital chases the team that has already assembled. Treat talent as downstream of investment and you get the sequence exactly backwards in the sectors that matter most.
So decide: when you scout a sector, weigh where senior people are quietly landing above where the funding was last announced.
Delivery driving is among the fastest growing roles on earth by sheer headcount, but it is a story of necessity, not of a frontier. In a country with 60 percent youth unemployment, most movement is push, not pull.
So decide: separate the inflow of aspiration from the inflow of desperation before you call anything a trend.
The 2021 crypto rush, and the metaverse hiring spree that chased it, pulled in extraordinary talent that had largely melted away inside two years. A spike is not a signal; it is a crowd. What you are hunting for is the inflow that survives the first disappointment, the down round and the layoffs, and keeps compounding regardless.
So decide: wait for the second wave. Back the movement that returns after the hype has left the building.
What is true for San Francisco is often false for Johannesburg. A signal that ignores emigration corridors, the informal economy and a youth unemployment rate this size is not foresight. It is imported noise wearing a confident chart.
So decide: re-weight every global signal for your own geography before you act on it. Never trade someone else's map.
None of this requires a data science team or a subscription to an expensive feed. It requires the older and rarer discipline of reading the cheap, early layers while everyone around you stares at the expensive, late ones. What follows is five habits, and beneath them, what each one changes depending on the chair you happen to occupy.
Watch for job titles that did not exist three years ago. New language is the first visible edge of a new economy. By the time the title is common, the lead is already spent.
People acquire the capability before they change the job. A 140 percent rise in skills added to profiles is a leading indicator hiding inside a lagging one.
Distinguish the inflow of aspiration from the inflow of need. An engineer joining renewables and a graduate driving deliveries are both movement. Only one is a frontier.
Re-weight every global signal for your own market: emigration corridors, the scarce skills list, the training pipeline that does or does not exist. The current bends around local geography.
Your whole advantage lives in the gap before consensus. Act while the gold line still runs ahead of the blue. Once the money agrees with you, the edge is already priced in.
Which capabilities rivals are quietly hiring for
Pay benchmarks and headcount
Stop using pay benchmarks and headcount, both lagging, as your read on the workforce. Track which capabilities your rivals are quietly hiring for. That list is your eighteen month warning, and it is cheaper than the forecast you would otherwise commission.
How to read this: select a chair along the top to see the same foresight gap from that seat. The timeline marks when talent moves against when capital prices it in; the two boxes below show what to watch, the leading signal, against what to stop relying on, the lagging one, for that role.
Every figure that will eventually move a market, a funding round, a quarterly print, a GDP revision, is downstream of a decision some person made first: to learn a thing, to join a team, to leave for a problem that paid less but mattered more. By the time that decision reaches a balance sheet it has been laundered into a number, stripped of the eighteen months of warning it once carried. The number is true. It is just late.
The discipline that falls out of all this is almost embarrassingly plain, which is probably why so few people hold to it. When the room reaches for the cleanest number, reach one rung higher, for the messier and earlier thing the committee has not priced. There is a trap worth naming on the way out the door. We tend to trust a number in proportion to its precision, and precision is exactly what the late indicators have in abundance. GDP is quoted to a decimal and describes a quarter that closed months ago. The cleaner and more auditable a figure looks, the more of its life is already behind it. So watch the vocabulary, the skills, the moves. Read a shortage as a forecast and an emigration as a bid. The map made of people updates before the territory it describes, and nowhere on earth is that map being redrawn faster than here, on this continent, right now.
If your read on the future still begins with a balance sheet, it begins one rung too low. That is the conversation Diagelo exists to have, and we would rather have it with you before the money does.
Skilled people do not spread evenly. They pool, and the pools deepen. Four forces feed the same loop, which is why a lead, once established, compounds.
This is why talent policy that only trains people, and never builds the place that keeps them, funds the next city's rise.
