Expert ecosystems: the new way to run a transformation
- 1A transformation now requires narrow, varied expertise in short bursts, which neither a single large provider nor an internal team alone covers well.
- 2An effective ecosystem rests on three pillars: an orchestrator accountable for the result, experts engaged on demand, and internal relays with genuinely protected time.
- 3The real investment is not the external budget but internal days: in a typical example they are of the same order as the days billed.
- 4The main risks are diluted accountability and dependence on individuals; a clear mandate, required documentation and planned skills transfer address them.
Most transformations we see in companies of 100 to 1,000 people do not stall because the idea was weak. They stall because of the shape of the team asked to deliver it. Two shapes dominate. The first hands everything to one large provider, with its pyramid, its methods and its day rates. The second hands everything to an internal team that is already saturated by day-to-day operations. Neither fits the problem as it now presents itself.
Our view is clear. The most reliable way to run a transformation in an SME or mid-cap is to assemble a small ecosystem of experts, placed under a single orchestrator, and connected to internal relays who keep ownership of the business.
Why the two classic models are running out of road
A current transformation draws on narrow and varied skills: process mapping, data quality, technical integration, law and data protection, training, change management. No mid-sized organization holds all of them permanently, and no large firm supplies all of them at top level when the teams are staffed with juniors in proportion.
The single large provider brings structure, but the executive often pays for coverage rather than expertise. The purely internal team brings knowledge of the ground, but it learns on the job in areas where mistakes are expensive. The regulatory context sharpens this. In Switzerland, the revised Federal Act on Data Protection (FADP) has been in force since 1 September 2023. In the European Union, the AI Act entered into force on 1 August 2024, with phased application and most obligations arriving in August 2026. A Swiss company that sells or operates in Europe now has to settle precise legal questions, for instance how a given use case is classified. Those questions call for a few days of a real specialist, not a quarter of a generalist.
What we mean by an expert ecosystem
This is not a stack of vendors. A functioning ecosystem has three components.
An orchestrator accountable for the result
One person, internal or external, answers for the objective: timing, budget, adoption. That person arbitrates, sequences interventions and rejects deliverables that do not serve the result. Without this role, the ecosystem becomes a row of suppliers who are each right within their own scope.
Experts engaged on demand
Each expert comes in for a short sequence with a defined output: a process diagnostic, an integration design, a legal analysis, a training scheme. They leave behind usable documentation. This resembles what is called the fractional model in the US, and it suits the Swiss and European landscape well, which is rich in specialist boutiques and senior independents.
Internal relays with real time
This is the component executives most often underestimate. Relays are people from the business who validate what is designed and then carry it to their peers. Without protected time, they are only names on an organization chart. We covered this in our thinking on designing the organization around AI and how roles change.
A worked example
Take a Swiss industrial distributor of 350 people that wants to make order entry more reliable and relieve its customer service, using an AI-based assistant where it adds value and plain organizational changes elsewhere. The assumptions below are illustrative.
- Duration: five months.
- Orchestrator: 35 days, of which roughly 15% goes to coordination.
- Process expert: 15 days.
- Data and integration engineer: 25 days.
- Lawyer or data protection officer: 5 days.
- Training and adoption lead: 12 days.
That makes 92 external days. At an assumed average rate of CHF 1,600 per day, the order of magnitude is CHF 150,000. On the internal side, eight relays each give 10% of their time over five months, about 10 days each, so 80 days in total.
The point that matters is that internal days are of the same order as external days. The visible budget is the provider's, but the decisive investment is the time of your own people. A company that has not planned those 80 days will get a fine deliverable and little usage. The case we described on getting an AI assistant adopted in customer service illustrates the logic: the technology weighs less than the ownership of those who use it.
The real risks, and how to handle them
An ecosystem is not free. We see three risks.
Diluted accountability. When ten parties contribute, each can believe the failure sits elsewhere. The remedy is the orchestrator, with a written mandate and a shared outcome indicator, not just deliverables.
Dependence on individuals. An independent expert who leaves takes knowledge along. Require documentation, reviews with the relays and a skills transfer planned from the start.
Coordination cost. It is real, in the order of 10 to 20% of external days in a setup of this size. It is justified as long as the number of contributors stays limited. Beyond six or seven, we recommend consolidating.
What we recommend to executives
Name the orchestrator before choosing the experts. The first hire of any transformation is the person who answers for it.
Plan internal days as a budget line. If the relays' time is not freed and visible, it does not exist.
Buy deliverables and handovers, not presence. Every external intervention should end with a document, a decision or a skill acquired.
Bring in legal early, and briefly. A few days of a specialist in the FADP or the European regulation upfront avoid costly rework, especially in regulated trades, as we showed for healthcare and medtech.
Size training to what teams can absorb. An ecosystem that trains too much, too fast, produces fatigue rather than use. Our article on training teams in AI without drowning them develops this point.
A question of governance before a question of tools
The expert ecosystem is not an organizational fashion. It is a pragmatic answer to a simple fact: the skills required are many, brief and specialized, while accountability for the result must stay single and close to the ground. This is the core of our approach to organization and adoption: start from strategy, deploy with the right people, then steer the results.
If you are unsure what shape your next transformation should take, who orchestrates, which experts, which relays, our diagnostic is a simple starting point to discuss it on the basis of your actual situation.
A personalised first reading in three minutes.