Three talent management principles every leader gets wrong

Executive overview

Most organizations conflate "people" with "talent" — and manage accordingly, with policies that repel the very people they need. Mark Allen, talent management expert and author of Aha! Moments in Talent Management, outlines three principles that expose where standard practice fails.

Outdated HR rules block exceptional hires. Promotion ladders reward past performance instead of future fit. Training is designed around learning rather than business outcomes.

The core insight: talent management done well is a competitive advantage — done poorly, it actively prevents results.

HR policies as talent prevention systems

• Most organizations only recruit when a seat is empty — reactive, not proactive.
• Top-tier candidates are rarely on the market for more than a few weeks; slow processes hand them to competitors.
• A Fortune 500 example: a known, high-value candidate was lost because no open requisition existed — the gap was purely procedural, not financial.
• Many hiring rules originate from the 1980s (e.g., mandatory 10–15 day posting windows designed for newspaper classified ads).
• Fix: ask "why does this rule exist?" — many were reasonable once but now function as barriers.
• Goal: maintain rigorous standards while building flexibility for exceptional, time-sensitive opportunities.

Promoting the best engineer is two mistakes, not one

• The Peter Principle — promoting people until they reach their level of incompetence — was named in the 1960s and is still standard practice.
• Mistake one: creating a manager from someone without management skills or interest.
• Mistake two: removing your best technical contributor from the role where they excel.
• The assumption that the best engineer will become the best manager has no logical basis — yet it drives most promotion decisions.
• No organization would hire an accountant without accounting skills or a nurse without nursing training; management is treated differently without reason.
• Fix: match manager candidates to documented managerial competencies, not to performance in their prior technical role.
• Google's Project Oxygen — an 18-month, data-driven study — produced eight characteristics of effective managers (good communicator, good coach, cares about employees, etc.); none were surprising, but they're rarely used as selection criteria.
• Ask people whether they want to manage others before assuming they do — many technical professionals actively don't.
• Build two tracks: a practitioner path (senior/master engineer, project leadership without people management) and a management path with deliberate development before the seat opens.

Training as investment, not cost

• Kirkpatrick's four levels — satisfaction, learning, behavior change, business results — are widely known; most programs stop at learning.
• Research from Jack Phillips: 60–90% of what people learn in training is never applied on the job.
• A program that's interesting but changes nothing is still a waste of investment.
• The problem is design, not measurement: programs are built to produce learning outcomes, not business outcomes.
• "Intertrainment" — training designed primarily for a positive experience — is common; more time is often spent on the lunch menu than on the intended business result.
• Fix: start with the end in mind (Covey). Define the specific, measurable business outcome before designing the program — increased sales, reduced turnover, improved engagement.
• Leadership development is a frequent offender: programs aim to "develop leadership" without connecting to what good leadership is supposed to produce.
• Leadership is a means to an end, not the end itself — programs must connect to measurable business results.
• Organizations that treat training as a cost can't answer ROI questions; those that treat it as an investment must be able to.
• Two questions to ask before any training initiative: (1) How will this improve a key business result? (2) How will we measure it?
• Sophisticated corporate universities now refuse to run programs that can't specify a quantifiable business outcome in advance.