AI for mid-market companies: why one-size-fits-all diagnostics fail

You’ve probably seen this type of report before: twenty pages, a four-quadrant framework, sector-wide automation percentages, and a general recommendation to “upskill your teams on AI tools.” It likely cost a lot, it was presented with confidence, and it tells you nothing at all about your organization. This isn’t a problem of consultant quality — it’s a problem of method. A generic framework cannot, by construction, reflect the reality of any single mid-market company. Here’s why this format fails exactly where mid-market companies need it most.

The same report for everyone, tailored to no one

A generic diagnostic starts from a fixed architecture — four quadrants, five job families, a risk/impact matrix — and forces your organization into it. The result looks rigorous: boxes, colors, scores. But the architecture was designed before it ever encountered your company, not built from it. It would produce nearly the same output for your main competitor, even though your org charts, your tooling and your volumes have nothing in common.

That’s the implicit — and untenable — promise of the generic framework: selling you precision in a form that, by definition, cannot be precise about a specific case.

Why generic sells well to mid-market companies first

It’s no accident this format sells particularly well to mid-sized organizations. A generic framework is cheaper to produce, faster to deliver, and it looks just as serious as a bespoke study when presented to an executive committee. For a mid-market company with no in-house data scientist and no unlimited consulting budget, the pitch is appealing: a ready-made report, presentable in a single meeting, that creates the feeling of having addressed the subject.

The problem shows up six months later, when you actually need to decide where to allocate the training budget, which role to redesign first, which division to prioritize. The generic framework was never built to answer those questions — it was built to reassure a committee, not to steer a transformation.

What a generic framework cannot see

• Your actual digitalization level — two companies in the same sector, with the same job title on the org chart, can have completely different levels of tooling. A sector-wide framework averages the two and gets both wrong.
• The exact composition of your roles — a “payroll administrator” in your organization might spend 60% of their time on already largely automatable tasks, or 15% — it depends entirely on your current payroll system, not on the job title.
• Your own volumes and cost structure — the freed-up time and its value in dollars or euros depend on your real volumes, not a market average.
• The automation you’ve already put in place — a company that has already digitized part of its processes doesn’t have the same remaining potential as one starting from zero, even at identical headcount and sector.

A generic framework averages these four dimensions across an entire sector. The result can be statistically true and individually wrong for you — in either direction.

A diagnostic that would hand your main competitor the exact same result hasn’t measured anything specific to your organization.

Mid-market size cuts both ways

Large groups can afford to fund an in-house data team to build their own AI-exposure model, role by role, division by division. Very small companies, on the other hand, have few enough roles that a qualitative review led directly by leadership is often sufficient. Mid-market companies are caught in between: organizationally complex enough that an informal, qualitative approach no longer holds up — several hundred roles, multiple divisions, different collective agreements across sites — but not large enough to justify a dedicated internal data team.

This is exactly the segment the generic framework rushes into, for lack of a visible alternative. And it’s exactly the segment where the gap between the report and reality costs the most, because the resources to correct a bad prioritization call are more limited than at a large group.

A tailored diagnostic without pulling in your IT department

The right answer isn’t to give up on methodological rigor for lack of internal resources — it’s to build the diagnostic from your real data, without requiring the technical integration a mid-market company can’t afford. In push mode, the company sends an Excel export of its roles, headcount and volumes; no access to internal systems is needed, and no HRIS project has to be launched beforehand.

From that export, each role is broken down into elementary tasks and checked against a reference library of roles already pre-scored — producing an automation score anchored in your reality, not a sector average. A first-level diagnostic (N1), run across the whole organization in about three weeks, gives the executive committee an actionable overview without mobilizing a dedicated technical team. A deeper, division-by-division diagnostic (N2) then refines the prioritization wherever the stakes justify it.

The real cost of a bad diagnostic

A generic framework that gets it wrong doesn’t just cost the price of the engagement. It costs a training budget misallocated toward skills that will be obsolete within a year. It costs roles redesigned in the wrong order, starting with the ones that looked like priorities on a sector chart rather than the ones actually under strain in your company. It costs, in labor negotiations, the credibility of a leadership team presenting numbers that any sharp employee representative can immediately challenge with: “sure, but what about here?”

Methodological rigor isn’t a luxury reserved for large groups. It’s precisely what a mid-market company needs to make decisions it doesn’t have the resources to make twice.