The unit of a multi-market discovery programme is the country visit, and the argument about it is nearly always about length. Three days or five, one workshop or two, whether the local team can spare a full day. The more useful question is which parts of the work have to happen while the team is physically in the building. On-site hours with local leadership are the only genuinely scarce input in the whole design. Everything that does not require them can be pushed to one side of the visit or the other, and a design that has not decided which is which will spend its scarcest resource on writing.
A proposal put to a European composite insurance group answered that question in a diagram. It designed one standardised visit per operating company, drew it as four lettered stages, and mapped those stages onto the three phases of the wider engagement, so that each country visit ran the whole method in miniature. Only the second stage happens with the team in the room. Everything below is a design inside a competitive bid. No visit in it was ever run.
The challenge
Six operating companies plus the group centre, inside a three-month span, with all six country visits compressed into a single month of the plan. That arithmetic removes the second trip as an option. Whatever a visit fails to settle it fails to settle permanently, because the team is in a different market the following week and the aggregation phase begins before anyone could return. It also forces standardisation for a reason that has nothing to do with elegance: six visits run six different ways produce six outputs that cannot be ranked against each other, and ranking them was the point of the third phase.
The commercial sheet made the same constraint visible from the other side. Exploration was offered as one week per operating company, and the method spends three of those days on site, a gap set out in Price a Multi-Country Analytics Rollout Per Country, Not as One Fee, which accounts for the difference as travel plus the work sitting either side of the visit. That second half is the part to hold on to, because the work either side is not overhead. It is two of the four numbered stages, moved out of the room on purpose.
The approach
Stage A happens before the team lands, and it holds two things. An intake conversation with the chief executive of the operating company, and the distribution of an online maturity and value scan to the people who will sit in the sessions. Both sit outside the three days, both cost a fraction of an on-site hour, and the order inside the stage is the part worth copying.
The chief executive is the first conversation of the entire visit, and it happens while there is nothing yet to argue with. No candidate list, no scores, no draft roadmap. A local chief executive handed a finished analysis will respond to the analysis, which is a conversation about a document. Asked earlier, with nothing on the table, what the market is doing to the book and where the next two years of margin has to come from, they answer a different question, and that answer decides what the sessions are permitted to be about. The intake is not a courtesy call before the real work. It is the scoping instrument, and putting it anywhere later in the sequence turns it into a review.
The scan is the same move applied to the room rather than to the sponsor. A published sibling from this same document, Your Published Strategy Already Promised the Analytics Function, records the reason in the plainest possible terms: the capability review went out a week ahead so the session could argue about results instead of collecting them. Data collection is not a use of on-site hours either.
Stage B is the visit itself, and it is two sessions in a fixed order. An ideation session that opens the field, then a maturity and roadmap session that closes it. Same team, same trip, largely the same people. In most discovery designs the gap between those two sessions is measured in weeks, and that gap is where the output quietly dies. An idea that leaves a workshop without a value and effort estimate on it arrives at the next meeting as a memory, carried by whoever liked it most, with no number attached that anyone in the original room actually made. Holding the scoring session on the same visit means every candidate gets priced while the people who proposed it are still present and can be argued with, and while the analyst who wrote the estimate can be challenged on it by the person whose budget it lands in.
An idea that leaves the workshop unpriced does not survive the trip home. It survives as a memory, defended by whoever liked it most.
Stage C happens after the team leaves: the artefacts and the value map report for that operating company. Writing has no claim on on-site hours.
Stage D is the stage most workshop designs do not have at all. The local report feeds into the group-level roadmap, which is the artefact that becomes the investment plan put in front of the board. So the visit is not finished when the local report is signed off locally. It is finished when its rows are in the group plan. That is a scoping decision with real consequences, because it puts the merge inside the visit team's job rather than the receiving group function's backlog, and the merge is where a country's optimism gets normalised against five other countries' optimism.
Two further constraints hold the three days together. The visit team is three people with three distinct roles, a senior insurance and digital practitioner, a data scientist and an analyst, a composition the design names on the same slide as the stages rather than leaving to staffing. And the scope per operating company is deliberately narrow, one small segment rather than the whole book. A three-day visit into an entire multi-line insurer is a diary entry. A three-day visit into one segment is a schedule.
- 4
- Stages per visit
- 1
- Stage on site
- 3 days
- On-site time
- 3
- People in the visit team
The outcome
This was a design in a competitive proposal, and the tense matters. No operating company was visited, no chief executive was interviewed, no scan was distributed and no roadmap was merged into anything. Every figure above is a planned figure. What the document produced was a repeatable unit, and the unit is the part worth carrying.
Two things about it have held up. The first is that a visit designed backwards from its scarcest input beats a visit designed forwards from an agenda. Once the three on-site days are reserved for decisions only, everything else has to find a home either side, and the design gets an honest answer to a question most programmes never ask out loud, which is what the trip is actually for.
The second is Stage D. A local list that never becomes a row in a funded group plan is a workshop, and the industry has produced an enormous number of them. Making the merge a numbered stage of the visit rather than a separate group-level activity is what stops six good country outputs from turning into six orphaned decks.
One part of the design would now be built differently. Stage C, the writing, was a real cost in a plan of this shape, and it is close to free today: a structured session output, a scored candidate list and a per-country report are exactly what a language model assembles well from notes that already have a schema. The instinct that follows is to spend the freed time on a longer visit, and that is the wrong reading. Cheap writing is an argument for more markets covered in the same month, not for more hours in the same room, because the room was never the bottleneck the writing was starving.
The maturity half of the scan would change too. Self-reported readiness was the only instrument available in a plan like this one, and it is the weakest part of the whole sequence. Whether an operating company can complete a policy change or settle a claim without a person touching it is answerable from workflow events rather than from a questionnaire, and the same holds for the agentic pilots insurers are funding now. Whether a drafting or triage agent is worth paying for in a given market depends on whether its output has anywhere to go, and that is a fact about the estate, not an opinion anyone in the session holds.
What none of that touches is the sequence. The senior conversation still has to come before there is anything to defend, the pricing session still has to sit on the same visit as the ideation, and the visit still has to end in the group plan rather than in a local folder.
Source: a competitive proposal written for the founding engagement of a European composite insurance group's group-level data analytics function, and specifically the slide setting out the anatomy of the standardised operating company visit. It is a proposal and a designed method, not a delivered engagement: the stages, the team, the durations and the country count are all planned figures, and the copy we hold carries no evidence that any visit was run. The reading that only one of the four stages happens on site, and that this is the design's central choice rather than an accident of drafting, is ours.