No Algorithm Signs the Player

The model will not walk into the locker room for you

Automation is here to stay in talent selection: automatic CV screening, interview scheduling, online testing, first-round filters run by artificial intelligence. The real question is no longer whether to automate, but which part of the process can safely be handed to a machine and which part still demands trained human judgement.

• Automating repetitive tasks frees up time, but it doesn't free anyone from the responsibility of deciding well.
• The more a selection process is automated, the easier it becomes to lose sight of the judgement that used to guide it.
• You sign talent, you don't process it, confusing the two costs organisations dearly further down the line.

When the score replaces judgment

We keep seeing organisations and clubs that have adopted automated selection tools without first deciding what criteria they actually wanted to protect. The result is a faster process, yes, but also one that is blinder to anything that doesn't fit a predefined box.

Unconventional profiles, the ones that don't follow the usual pattern but often turn out to add the most value, get filtered out before anyone with real judgement ever sees them. And when the moment to decide finally arrives, many hiring managers lean on a compatibility score generated by an algorithm they can't even explain.

• Processes that reject unconventional candidates because they don't match the programmed keywords.
• Interviewers who reach the final conversation having read nothing beyond an automated summary.
• Hiring decisions justified by a compatibility percentage and little else.

Useful data, non-negotiable human decision

Elite sport understands this well. A scout can use data and footage to narrow hundreds of candidates down to a shortlist, but the decision to sign a specific player still runs through people who know the context, the dressing room and the project. Business should work the same way.

This isn't about distrusting technology, but about using it where it genuinely helps: cutting through noise, speeding up filters, spotting patterns across large volumes. The final call, whether that person fits this specific team at this specific moment, still belongs to the people who will actually work alongside them.

How to use the algorithm without giving up the wheel

This isn't about choosing between automating everything or automating nothing. It's about deciding, with method, exactly where the line sits.

1. Define what the machine decides and what the person decides
   • Put in writing which stages an algorithm can resolve and which always require a human conversation.
   • Revisit that line periodically, since technology moves faster than processes do.
2. Audit the algorithm before trusting it
   • Understand which variables it uses to score a candidate and which biases it might be carrying.
   • Ask whoever built it to explain its decisions in plain language, not just in metrics.
3. Always keep a human conversation before deciding
   • No compatibility score should replace an interview with the person who will lead this hire.
   • That final conversation is also where you catch what no test measures well.
4. Measure the quality of the hire, not just the speed of the process
   • A faster process that increases turnover isn't an improvement, it's a problem disguised as efficiency.
   • Track each new hire's real performance through the first year, not just how quickly the role was filled.

Questions before automating a signing

Before adding another automated selection tool, it's worth asking:

• Can you explain, in plain terms, how the algorithm you use today scores candidates?
• How many unconventional profiles have you filtered out this year before a person ever reviewed them?
• Who makes the final hiring call: a person or a compatibility percentage?
• Have you measured whether a faster process is also driving higher early turnover?
• Does your technology free up time to think better, or just to decide faster?

Let's review your selection criteria together

We don't think automating well and deciding well are opposites. They only become opposites when automation happens before anyone has defined what criteria deserves protecting.

If you'd like to look at where that line sits in your own selection process, let's talk. You don't need the whole system figured out to start seeing it differently.