Skills frameworks without the theatre: mapping critical thinking, AI fluency, and leadership for midmarket orgs

The framework arrives as a poster. Forty competencies, four levels, a colour for each pillar and a request that every role be mapped by the end of the month. The working group is proud of the structure. Six weeks later the spreadsheet is half filled, the managers have guessed the levels, and nobody has changed what they practise.

A skills framework is useful when it helps a manager decide what to coach and helps a person see the next rep. It is theatre when it exists to be complete. Mid-market organisations do not have a team of organisational development specialists to maintain a cathedral. They need a map small enough to survive a busy quarter.

Three skills, not forty

If you are rebuilding capability for an AI heavy workplace, most of the urgent gap sits in three places. Critical thinking: can the person tell a useful answer from a fluent one, and can they frame the problem before they ask for options. AI fluency: can they use the approved tools on their actual work, inside the data rules, and can they tell when not to. Leadership: can a manager coach the first two, set a standard and make it safe to show a rough draft.

Other skills still matter, including domain craft, writing, customer conversations and commercial sense. A three part map does not replace a profession. Do refuse to bury these three inside a generic digital pillar where a manager cannot coach them. Name them, define them in verbs, and leave the rest of the catalogue alone until these are actually in use.

Levels a manager can see

Four levels is plenty, and only if each level is a behaviour you could observe this month. Vague labels like emerging and transformational invite people to tick the flattering box.

1. Needs a guide: they can follow a worked example with someone sitting next to them, and they should not send the output alone.
2. Can run the rep: they do the skill on their own work, and they know the check they still owe a colleague or a source.
3. Can teach the rep: they can sit with someone else, spot the miss and raise the standard without taking the work back.
4. Can set the system: they decide where the skill is required, what good looks like and what the team will stop doing.

Most individual contributors need to reach the second level on critical thinking and AI fluency. Most managers need the third. A few people, often in L&D, risk, product or operations, need the fourth. If your framework says everyone must be visionary, it is a poster.

Map roles to work, not to job titles

Take the roles you are actually training this year. For each one, write the work sample that would prove the level. A critical thinking sample is a marked draft: what they would not send, and why. An AI fluency sample is a workflow they ran for two weeks inside the rules. A leadership sample is a coaching note, or a decision to stop a use that looked busy and was not safe.

Do not ask people to self rate on a five point scale and call it a skills inventory. Self ratings drift toward the middle, or toward whatever the promotion guide rewards. A work sample is slower and it is honest. You do not need one for every employee in month one. You need one for the cohorts you are about to spend money on, so you know whether the training moved the sample.

The critical thinking workshop is a rep for the first skill, not a substitute for the map. People should leave with a marked piece of their own work, which is already a better record than a level selected from a dropdown.

Keep it out of the theatre

Theatre has a look. A launch event. A new logo for the framework. A mandatory course that explains the framework instead of practising the skills. A dashboard of completion that is reported to the executive team in place of the work samples. If you recognise the list, cut the launch and keep the samples.

1. Publish the three skills and the four levels on two pages, with one example per level, written in house language.
2. Attach a work sample to each cohort you train, before and after, and file it where the manager can see it.
3. Refuse to add a fourth skill until a manager has used the first three in a development conversation.
4. Review the map twice a year, with the people who coach, and delete any line nobody has used.

A framework that gets shorter is healthier than a framework that gets a subcommittee. Mid-market teams do not fail at skills because they lack a taxonomy. They fail because the taxonomy is too large to use on a Thursday afternoon. Critical thinking, AI fluency and leadership will cover the argument you are actually in. Map those, practise those, and leave the poster until a manager has used the short version in a real development conversation.