Why Training Is Not Enough: the Five Conditions for Autonomy

The scenario every organisation knows#

A problem appears: incomplete files, customers sent to the wrong place, disputed sales, forgotten controls. The quickest response is training. Participants enjoy it, the end-of-session evaluation is good, and three months later the problem is back.

It was not the training that failed. It was the diagnosis. The problem lay elsewhere: in a rule that prevented the adviser from deciding, a tool that did not show the relevant information, an individual target that rewarded speed rather than quality, or the absence of any regular meeting to keep the new practice alive.

The five conditions#

These five conditions spell out what the Empower step of the KAPER method means: passing on a capability is not just passing on knowledge.

What research and data say#

Incentives are the most frequently neglected condition. As early as 1975, Steven Kerr (Academy of Management Journal) described organisations that reward one behaviour while hoping for the opposite. A sales team paid on volume and asked to sell responsibly is a textbook example.

The link between teams and customers is real but conditional. The review by Hogreve, Iseke and Derfuss (Journal of Service Research, 2022) of the chain linking employees' situation to customer satisfaction concludes that the link is neither automatic nor linear: it depends on the context, the processes and the tools provided.

In Morocco, the World Bank's 2023 Enterprise Survey puts the management practices index at 32 out of 100, against 46 for the Middle East and North Africa region. The index covers concrete practices, such as tracking indicators, setting targets and incentives. In other words, part of the gap lies in management systems, not just in individual skills.

The one-hour diagnosis#

Before deciding on training, bring the manager and two or three members of the team together around a real case, and ask five questions in order:

1. Would they know how to do it correctly if their life depended on it? If not, knowledge is missing.
2. Are they allowed to take the necessary decision? If not, authority is missing.
3. Do they have the information when they need it? If not, information is missing.
4. Is the desired behaviour the one that is measured and rewarded? If not, incentives need reviewing.
5. Is there a regular moment when you check that the practice is applied? If not, routine is missing.

It is common to find two or three conditions missing at once. The response then combines several actions, and training is only one of them.

When training is essential#

Some situations do require training, and sometimes the law demands it. Article 4 of the AI Act, applicable since 2 February 2025 and amended in 2026, requires providers and deployers of AI systems to take measures to support the AI literacy of their staff; a Moroccan company is concerned when the outputs of its systems are used in the European Union. Our guide to training teams in AI explains what it involves.

Even then, training has more effect when it comes with the other conditions: a designated owner for each AI system, clear information on permitted uses and a regular review of incidents. The same logic applies to compliance culture in general.

Measure the effect, not satisfaction#

The usual training evaluation measures what participants thought of it. It says nothing about what has changed on the ground. To know whether an initiative has worked, you need to:

• choose, before starting, an operational indicator linked to the problem (repeat contacts, complete files, cancelled sales, controls on time);
• record its baseline value;
• measure it again at an agreed date, usually two to three months later;
• examine the gaps in the team's routine.

This is the logic of the Reflect step: an initiative that has not changed the indicator is adjusted or stopped, however satisfied participants were.

Your checklist#

• The problem is described from real cases, not from an impression.
• The five diagnostic questions were asked before deciding on training.
• Each missing condition has an identified response and an owner.
• Targets and variable pay have been checked so that they do not contradict the desired behaviour.
• An operational indicator has been chosen, with its baseline value.
• A regular routine is planned to keep the new practice alive.
• A measurement date has been set.