Why Banks Keep Missing Top Talent — and What High-Performing Firms Do Differently

Banks are paying more per head than at any point since 2021. Johnson Associates projects advisory and equities bonuses up 10–20%+ for 2026, on top of last year's increases, and is explicit about the mechanism: lower overall headcount means fewer people sharing a larger revenue pool (Johnson Associates, May 2026). The same firm expects industry headcount to fall 10–20% over five years (Johnson Associates, November 2025).

Fewer seats, more money attached to each one. That arithmetic should concentrate minds, because it means the cost of losing a hire has gone up while the number of chances to get it right have decreased.

It has not concentrated minds much. When a bank loses a candidate it wanted, the loss is almost always booked as competition — a rival paid more, or moved faster, or had the better brand. In the searches we run across London and Paris, that is usually not what happened. The candidate was lost internally, at a specific and identifiable point, and the firm never found out where.

1. Slow Decision-Making Is an Ownership Problem, Not a Calendar Problem

Processes are getting longer, not shorter. Banks across Europe are pushing modelling tests and case-study assessments earlier into their 2026 recruitment cycles (Selby Jennings, The Future of Financial Services in Europe 2026), adding stages at precisely the moment speed matters most. And the silence that fills the gaps is now the norm rather than the exception: in Greenhouse's May 2026 survey of 2,950 candidates, 51% of those who completed an interview received no response at all — not a rejection, nothing (Greenhouse, May 2026).

Every firm knows it is slow. Very few can say why, because the honest answer is rarely "we needed more information." It is that no single person could say yes without asking someone else, and each person in that chain was rational to defer. Speed is not a scheduling discipline; it is a governance one. Layers of beaurocracy kills momentum and desire.

Solving this is relatively simple - before a search opens, banks need to know who can approve this hire without a further conversation.

2. A Vague Proposition Loses to a Specific One, Not to a Bigger One

Randstad's Workmonitor 2026, covering 27,062 workers across 35 markets, contains the clearest statement of the problem we have seen. 81% cite pay as the leading factor in joining an employer — but only 23% cite it as a reason for staying, against 46% who name work-life balance (Randstad, Workmonitor 2026, January 2026).

Pay opens the conversation. It does not close it, and it does not hold. What holds is a candidate's belief that they understand what they are joining: the mandate, the reporting line, where the revenue comes from, what the next three years look like. There is a further mismatch to navigate — 72% of employers now describe the linear career path as outdated, while 41% of workers still want one.

Croft & Co has argued elsewhere that ambiguity at offer stage is one of the most common reasons senior candidates withdraw. It remains true, and it is entirely self-inflicted. A firm that cannot describe the seat in specific terms is asking the candidate to price the uncertainty, and they will price it at more than the firm wants to pay.

3. The Interview Process Is the Only Sample of the Culture a Candidate Gets

Firms describe their culture in interviews. Candidates ignore the description and read the process instead — how prepared each interviewer is, whether the panel tells a consistent story, how long the gaps are, whether anyone senior bothers to show up.

The data on where this is heading should worry anyone automating the top of their funnel. Greenhouse's May 2026 survey found 63% of candidates have now been interviewed by an AI, up 13 percentage points in six months — and 38% have walked away from a process because of it. Seventy per cent were never told in advance that they would be assessed by a machine; 21% found out when the interview began. Only 19% want less AI in hiring overall, but 46% want the option to ask for a human instead (Greenhouse, May 2026).

Read together, those numbers say something precise. Candidates are not objecting to automation. They are objecting to being processed without being told — which they correctly read as a statement about how the firm will treat them once they are inside it.

At VP level and above this compounds, because candidates are joining a person rather than an institution. Randstad found trust in senior leadership has fallen to 72% from 77%, while 72% report a strong relationship with their direct manager — up from 64% in 2024. The manager is the relationship that carries weight. Firms that keep the hiring manager at arm's length until a final round are withholding the only asset that reliably converts.

4. The "Perfect Fit" Brief Needs Elasticity

Some briefs are narrow because the work genuinely demands it. Many are narrow as a direct reaction to the abundance of AI-applications and because no one has pushed back on a brief that is too specific from the outset. A specification assembled from every attribute of the last incumbent, plus every capability the desk currently lacks, will describe a person who does not exist — and the search will spend four months proving it.

The workforce is moving against this. Randstad found 38% of workers now expect to hold different types of role across a career, and 72% of employers themselves say the linear path is finished — yet job specifications are still written as though careers run in straight lines (Randstad, Workmonitor 2026). The profile that has done exactly this job, at exactly this kind of firm, for exactly this long is a shrinking population. The profile that could do it well is not.

This is where a search should earn its fee before a single candidate is presented. Mapping the relevant desks tells you within a fortnight whether the brief describes five people, fifty, or nobody. Discovering that in week two is a strategy conversation. Discovering it in month four is a failed search, and the firm will conclude — wrongly — that the market was thin.

5. Compensation Is the Floor of the Argument, Not the Argument

The bonus data cuts both ways. Advisory and equities up 10–20%+ means the bank across the street has the same headroom. When the whole market moves pay together, pay stops differentiating and simply becomes the price of being taken seriously.

What differentiates is everything a candidate cannot verify from a compensation sheet: the quality and consistency of deal flow, who they report to and whether that person is staying, how revenue is credited, and — increasingly — whether the platform intends to remain in their business at all. The headcount forecast that is inflating this year's bonuses is the same forecast candidates are reading as a risk. A projected 10–20% fall in industry headcount is a pay rise if you survive it and a career interruption if you do not, and senior candidates now price that explicitly. A bank that cannot say what its franchise will look like in five years is competing on money alone against firms that can.

A Note of Balance

Not every loss is self-inflicted. Counter-offers have become more aggressive as pay pools have grown, some candidates were never really movable, and occasionally a competitor simply had the better seat. Speed, too, can be overdone: a bank that compresses its process by removing judgement will hire faster and worse, and front-office mis-hires are expensive in ways that do not show up for eighteen months. The argument here is not for haste. It is for knowing, in each case, which of these actually applied — most banks do not, because nobody asks the candidate afterwards.

Conclusion

Reach is not the constraint. Almost every bank in London and Paris can get a message to almost any banker they want. Conversion is the constraint, and conversion is decided by things entirely within the firm's control: who owns the decision, how precisely the seat is described, whether the process reflects the culture being sold, whether the brief is buildable, and whether the argument extends past money.

There is a sharper version of this for anyone hiring through search. A candidate who applied to you arrived with motivation to spare and will forgive a mediocre process. A candidate we went and found — someone who was not looking, is well paid, and has nothing to gain from the inconvenience — will not. That is the standard worth designing to. It is also, not coincidentally, the standard that wins the hires that were never going to apply.

‍

Sources: Johnson Associates, financial services compensation projections, via eFinancialCareers (November 2025 and May 2026); Randstad, Workmonitor 2026, survey of 27,062 workers and 1,225 employers across 35 markets (January 2026); Selby Jennings, The Future of Financial Services in Europe 2026: Investment Banking (2026); Greenhouse, survey of 2,950 candidates on AI in hiring (May 2026).

‍