Hiring for Potential: The AI Promise Almost Nobody Keeps

Detecting potential is not talent management

In recent years, the number of AI tools promising to detect a person's potential, whether in a company or in a club's youth academy, before that potential fully shows itself, has grown fast. The promise is appealing. The reality, almost always, falls short. Here is what we have observed:

• Detecting potential is only the first step, and many organizations stop right there.
• Potential that is not paired with a real development plan gets wasted just the same, with or without AI.
• The more precision a tool promises, the more responsibility it demands from whoever receives its results.

The report arrives, the follow-through does not

We have seen companies invest in AI platforms that score employee potential with sophisticated indicators, and clubs invest in systems that flag, at a young age, which players show the most promise. In both cases, the technology investment is usually well handled. What is almost never handled is what happens after that score comes in.

We come across high potential reports that get filed away, promising young players identified early who then receive the same generic treatment as everyone else in the academy, and employees flagged as key talent who never get a differentiated development conversation. The tool kept its promise to detect, the organization did not keep its promise to follow through.

• High potential reports that never translate into a concrete development plan.
• Promising young talent spotted early and then treated exactly like everyone else.
• Team leads unaware of who has been flagged as key talent on their own team.

A ranking, on its own, develops nobody

Our read is that detecting potential without a development system behind it is not talent management, it is just a ranking. And a ranking, on its own, develops nobody.

That is why, when we work with organizations and clubs on potential detection, we insist on designing the support system that will receive that information first: who has the development conversation, how often, and what real resources go to that person. Without that system, any AI tool, however sophisticated, becomes just one more report.

Three steps to turn detection into development

Turning potential detection into real development requires designing the whole process, not just the technological part.

1. Define the support plan before implementing the tool
   • Decide what will happen at each level of detected potential before the first report arrives.
   • Without that prior plan, the information arrives with nowhere to land.
2. Assign a specific development conversation, not a generic one
   • Each person identified with high potential needs a different plan than everyone else, not a standard chat.
   • That conversation should repeat on a defined cadence, not remain a one-time event.
3. Measure whether the detected potential turned into real progress
   • Check, over time, whether the flagged people actually grew more than the rest.
   • If they did not, the problem is not in the detection, it is in the follow through.

Before investing in another potential tool

Before investing in another potential detection tool, it is worth asking:

• What actually happens today to people identified as high potential, beyond the initial report?
• Do those people have a development plan different from the rest of the staff or academy?
• Do direct managers know who has been flagged as key talent on their own team?
• Have you measured whether the detected potential has translated into real progress over time?
• Are you ready to follow through on what your next AI tool is going to detect?

Follow through better on what you already detected

You do not need the best potential detection tool on the market to start developing your people better. What you need, above all, is a system that knows what to do with what you already know.

If you have information about your talent's potential that has not yet turned into real development, let's talk. Sometimes the challenge is not detecting better, but following through better on what you have already detected.