Germany’s twenty largest management consultancies grew revenue 4.1% in 2025, down from 7.5% the year before. Six of those twenty shrank. The best performers grew 26.4%, the worst fell 18.5%. That is a spread of nearly 45 percentage points inside a market that expanded 0.6% overall, and the average is now the least informative number in the whole dataset.
I read the Lünendonk figures the way anyone who sells days into German enterprises reads them: not as an industry story but as a map of which briefs still get signed. The headline says slowdown. The distribution says something more specific and more useful.
The numbers behind the flat line
The Lünendonk-Liste 2026 covers the German management consulting market for financial year 2025. The relevant figures:
| Metric | 2025 | Prior year |
|---|---|---|
| German top-20 revenue growth | 4.1% | 7.5% |
| International top-20 growth (German market) | 3.0% | 7.3% |
| Total market volume | €49.0bn | +0.6% vs 2024 |
| German top-20 combined global revenue | €3.96bn | – |
| German top-20 firms with falling revenue | 6 of 20 | – |
| Best performer (quattron, Horn & Company) | +26.4% | – |
| Worst performer (Ingenics) | −18.5% | – |
| German top-20 headcount | ~17,170 | – |
Roland Berger led on absolute revenue at €1,013.1m (+3.9%), followed by Simon-Kucher at €606.0m (+5.9%) and d-fine at €312.7m (+0.4%). This is the third consecutive year the sector has grown through a recession, which is itself worth noting: the market did not contract. It stopped moving as one thing.
The detail that undercuts the AI sales pitch
The finding I keep returning to is not a growth rate. Lünendonk reports that revenue per head declined across the German top 20 despite AI adoption, on headcount of roughly 17,170.
Consultancies spent 2025 selling AI-driven productivity to their clients. Inside those same firms, output per consultant went down. Some of that is billing-rate pressure rather than productivity, and one year is not a trend. But it is a genuinely awkward data point for an industry whose current growth pitch is that AI makes knowledge work cheaper, and it matches what I see in client environments: the pilot works, the organisation around it does not change, and the efficiency never reaches the P&L. I wrote about the mechanism in why enterprise AI pilots fail on governance rather than technology.
If the firms best positioned to capture AI efficiency, with the capital, the talent and every incentive to prove the case, cannot show it in revenue per consultant after a year, that should temper what any client expects from their own programme in eighteen months.
What is actually being bought
Lünendonk’s Jörg Hossenfelder names two demand drivers: AI consulting and restructuring. That pairing explains the spread better than any macro argument.
Both categories are non-discretionary. A restructuring is triggered by financial reality and the board has no option to defer it. An AI programme is now defended as competitive survival, which in German boardrooms has become the one budget line that still survives a cost review. Everything between those two poles, the capability build, the strategy refresh, the operating-model redesign with a two-year payback, is discretionary, and discretionary work is where the 18.5% decline lives.
The practical translation for anyone selling expertise into this market: a proposal that reads as improvement is competing against deferral. A proposal that reads as obligation is competing against other obligations. That is a much better fight.
Regulation belongs in the obligation column, which is why security and compliance work has held up while adjacent categories have not. The German NIS2 implementation moved a large tranche of work from voluntary to mandatory, and I covered how unevenly the Mittelstand has responded in the NIS2 readiness gap among German SMEs.
The freelancer market shows the same split
The independent market mirrors the consultancy one, and the mirror is clearer because rates are visible. The freelance.de Freelancer-Studie 2026, surveying 3,300 freelancers, found roughly 70% expecting a difficult year for project availability. Average hourly rates barely moved: IT freelancers at €101.98, the cross-discipline average at €101.70.
Demand fell and price did not follow. In a normal downturn those move together, and when they decouple it usually means the composition of demand changed rather than its volume alone. The projects that disappeared were the ones that competed on price. What remains is specialised enough that the client is not shopping on rate, because there are not five interchangeable candidates.
A flat average rate in a market with 70% pessimism is not a sign of health. It is a sign that the bottom of the market left, taking its lower rates with it.
What changed in the briefs
Three shifts show up consistently in what German enterprises are asking for now, and each one follows from the obligation-versus-improvement split.
- Scope arrives pre-narrowed. Clients bring a defined deliverable and a defined end date instead of a discovery phase. The exploratory front half of an engagement is being cut, which suits specialists and hurts generalist capacity.
- Evidence is the deliverable. In regulated work the artefact that matters is the one an assessor will read. That has always been true in compliance programmes; it is now true in AI programmes too, and most organisations discover late that their controls fail quietly rather than visibly.
- Sovereignty entered the requirements. Where data and models run has moved from an architecture preference to a procurement condition, which I set out as a practical exercise in the cloud exit test.
The crack nobody is pricing
One structural finding in the Lünendonk data has nothing to do with 2025 revenue: 58% of the firms named generational succession as their largest obstacle to filling partner positions.
That is a supply problem arriving in a market everyone is describing as a demand problem. Partner-level capacity is what these firms actually sell, and it takes a decade to build. A cohort retiring out of firms that cut junior intake during a three-year recession produces a shortage that does not resolve when the cycle turns. It also explains part of the spread: the firms growing 26% are not simply better at selling, they are the ones who still have someone senior to put in the room.
What I would take from this
Three conclusions I would defend, from the seat of someone who bills days into German enterprises rather than analyses them.
First, treat the market average as noise. A 45-point spread means the sector number describes no actual firm, and any planning built on “the market is up 4%” is planning for a company that does not exist.
Second, the AI-efficiency story is unproven by the people selling it. That is not an argument against AI programmes, and I build them. It is an argument for measuring your own before believing a vendor’s. The consultancies had the best possible conditions and could not show it in revenue per head.
Third, obligation beats improvement in a flat market, and it will keep beating it while budgets stay defensive. Security, compliance and regulatory work sit on the right side of that line, which is a large part of why they have been insulated. The organisations that get burned are the ones treating a mandatory programme as a discretionary one, then finding out on the assessor’s timetable rather than their own. My 90-day Zero Trust plan for the Mittelstand and my lessons from enterprise security and compliance programmes both start from that premise.
The consulting market is not in decline. It has stopped rewarding the middle, and the middle is where most capacity sits.