14 January · ten weeks before closing
The committee said yes. Financial due diligence is green, technical due diligence too. As the meeting closes, the partner in charge asks a question that has no box in any of the reports on the table: “are those two going to work together?” No one in the room is convinced by their own answer.
An illustrative situation, drawn from real ones. The companies, the people, the figures and the facts presented here are fictional and composite: they designate no existing organisation and no existing person. The method is the one we apply. This dossier is a demonstration — it reports no engagement result.
Declared door — Investors · Assessment of a portfolio company's leadership team
An empty chair at the end of a long committee table, the files closed and stacked, a lamp still lit above the unoccupied place.
What the fund bought, what the two companies really are, and why the question asked in committee was aimed at the wrong object.
Leadership team assessment — proposed combination of Cryotem and Montval Froid
What the combined group will be able to carry, what it will not yet carry, and where to act before closing.
- Commissioned by
- Vallonis Capital — Claire Mongin, partner, and Étienne Vasseur, investment director.
- Sequence
- Build-up no. 2 on the Cryotem portfolio company. Closing targeted for 25 March.
- What this is
- A reading of the leadership teams of both companies as a single system that has to work together.
- Why now
- Six weeks before closing: this is still the window in which a clause, a mandate letter or an integration timetable can be rewritten without reopening the price.
- What you do with it
- Settle four decisions, three of them before signing.
- Time
- Reading: 20 minutes · the essentials: 1 minute · the exhibits: on request.
The essentials
- The fit between the two leaders is not the risk in this transaction. The eleven interviews converge: Vincent Delmas and Karim Belhadj read each other well, divide the ground between them without ambiguity, and both have already lived through an integration. The doubt voiced in the committee of 14 January is sound in its instinct and mistaken in its object.
- The risk sits elsewhere, and it is measurable: the capacity to price complex jobs. At Cryotem, ninety jobs a year exceed €250k; they carry 61% of gross margin; two people price them, and the organisation chart marks neither of the two as critical.
- The transaction multiplies the workload at the moment the resource shrinks. The combined group will have roughly one hundred and twenty-five complex jobs to price each year. Available capacity today stands at 2.4 full-time equivalents; it falls to 1.4 within eighteen months; the requirement is 3.2.
- Bid turnaround is directly tied to the conversion rate. Across the one hundred and eighty-seven complex jobs of the past two financial years, those bid within eight days convert at 44%, those bid beyond fifteen days at 21%. A loss of capacity shows up in the jobs that go out late, rather than in the income statement.
- Cryotem is a company that performs and drifts away from its thesis. The thesis rests on regulatory retrofit; actual activity remains overwhelmingly conventional replacement work; no body measures that gap, so no one has yet had to decide on it.
- Four decisions are proposed, three of them before closing. None of them touches the price. All of them fit inside the transaction documents or the hundred-day plan.
How to read this dossier
Three parts, then the frame. Understand — what the fund bought and what the two companies really are. Decide — the reading of the system, the point where to act, what inaction costs. Act — four decisions with their owners and their deadlines, and what the investigation has not established. A fourth part carries the frame.
Every statement in this dossier carries its register. Four grades, marked line by line:
- établi
- A dated primary source founds it — an exhibit, a data extract, a set of minutes.
- déclaré
- Someone said it or published it. We have not verified it, and we do not hold it as given.
- réserve
- An estimate, a calculation, a secondary source. Robust, unmeasured.
- non établi
- We did not find it. It is written down here, and it is never filled in.
The rule that governs the other three: what is not established is never completed by a plausible inference. What is missing appears in § 12, named, together with what would settle it and what it costs to go and find it.
The reference point: what this team has to be read against
Nothing is judged in the absolute. A leader who systematically takes back control of the large files is an asset in a company of twenty people and a risk in a group of three hundred and thirty-five. A close-knit team is a strength on paper and a fragility if its cohesion rests on two people whom the integration plan separates.
This is why every engagement opens here, and never with the team. Until the project has been reformulated, no reading is possible — there is no scale to read against. And it has to be reformulated rather than taken as given: the subject presented is rarely the real subject.
The thesis, as the March 2024 memorandum states it established: the scheduled phase-out of refrigerants with a high global warming potential forces an entire installed base to retrofit inside a dated window. The demand is regulatory, and therefore barely elastic to the cycle. Whoever holds geographic density together with the ability to price quickly will take a disproportionate share of that wave, because on jobs of this kind the first credible price sets the price.
That thesis carries a consequence the memorandum writes down without dwelling on it, and it commands this whole dossier: it makes pricing speed, rather than installation capacity, the scarce resource.
The project, qualified
We do not score a project — a yardstick cannot be measured against itself. We qualify it on four attributes, and here it is the fourth that is missing.
yes
yes
yes
no
Over the last financial year, 78% of new orders are conventional replacement work and 22% regulatory retrofit; the chairman devotes the bulk of his selling time to the long-standing accounts. established 2025 job register, 1,412 lines · declared commercial diary, in interview.
A project that is formulated, shared and communicated but does not reach the diaries remains a solid project: it simply shows where it fails to travel. Here, it does not travel through the chairman's selling time. That is management information, to be handled as such.
Two very different companies that the organisation chart makes look alike
Cryotem — 240 people, €68.4M of 2025 revenue, two sites, a legible structure: a founder-chairman who sold the majority and reinvested, a chief executive recruited by the fund eighteen months later, an engineering department, a site management department, a finance and administration department. The organisation chart is up to date, the processes are written down, and — rarer — they are followed. Decisions taken in the executive committee are applied; people say so without being asked about it.
Montval Froid — 95 people, €21.3M, one site, a founder who holds the entire share capital. The organisation chart has existed since a finance director was recruited in 2023. It describes an organisation that works differently: jobs, hiring, workload trade-offs and customer disputes all pass back through the founder, and everyone accommodates that because it works. The processes are written down; going around the process is the norm, and no one experiences it as an anomaly.
These two companies have different problems, and the classic integration error is to apply the same remedy to both.
The question of 14 January, and why it fell beside its object
In committee, the partner in charge asked the question everyone had in mind: are those two going to work together? The instinct is sound. The object was beside the point.
What the interviews establish on that point. Delmas and Belhadj have each led an integration — one on the acquirer's side in 2019, the other on the acquired side in 2016. They describe the post-closing division of ground in the same terms, neither of them having had knowledge of what the other said. None of the nine other people interviewed reports tension between them, and two of them volunteer the opposite. On the personality inventory taken by the six members of the two management teams — an instrument from the five-factor family, with documented validity, debriefed individually to each of them — the two men differ on openness to experience and meet on conscientiousness and on tolerance for open conflict. That is a configuration of complementarity rather than collision.
What this does not say. That the transaction is safe. Only that the risk has lodged elsewhere — and that a fund which had bought nothing but the answer to the question asked would have left reassured on the wrong subject.
The grid, its application, the fulcrum, and what inaction costs.
The grid: how we read a leadership team
What follows applies a proprietary grid, built for this type of question and tested on earlier cases before being applied to this one. Setting it out here rather than in an appendix spares you reading a reading without knowing where it comes from.
It reads a leadership team as a system on three levels, rather than as a list of people.
The project, the investment thesis. That is § 03, and it is the yardstick for everything else.
What the organisation holds — the stock of expertise (§ 07). How it allocates that stock — which functions really carry performance, and whether they are held (§ 08). What people do together, and what keeps them there: in this dossier the two make sense only together, and they are read in § 10.
The bodies that govern, distinct from those that manage (§ 09). That demarcation changes the remedy, and § 09 shows it on a single fact read twice.
Two movements run through this system, and that is where the diagnosis is made.
The first ties the official organisation to the real one, and it amplifies in whichever direction it turns. Three cases, which every leader has already seen. They reinforce each other — a decision taken on Monday is applied on Tuesday. They ignore each other — the organisation chart says one thing and everything passes back through the founder's office. They destroy each other — every departure calls for a turn of the screw, and every turn of the screw calls for the next departure.
The second ties control to intent. It corrects, or it has stopped correcting: a committee that tracks the order book by line of business sees the drift in the quarter it begins; a committee that tracks cash alone discovers it at the mid-term review. § 10 says which of the two cases arises here, and why a weak quadrant changes meaning according to the system that carries it.
This reading produces a location rather than a score. A score gets filed, and it gives a committee the feeling of having decided. What a committee needs is a place: the point where a limited action moves the most things at once.
The difference shows on two moves this dossier might have recommended. Recruiting one more person into the engineering department adds hands: the workload falls, and everything else holds as before. Writing down, before signing, the threshold above which a job belongs to the pricing function moves the workload, the bid turnaround, the conversion rate, the operations director's remit and the quarter in which a mistake becomes visible, all at once. The first belongs to management. The second is a move played at the fulcrum.
Go has always had a name for that distinction: an average player plays where the board is large, a good player plays where the board is decided. § 08 says where the fulcrum of this transaction sits; § 11 says the move played there.
A real stock of expertise, and an undocumented one
The trade expertise of the two companies runs deep and is rare. In its most decisive part it is also embodied rather than written — that is, lodged in people rather than in procedures, tools or databases.
The clearest case is the pricing of complex jobs. Pricing an industrial refrigeration installation on a site that keeps running means arbitrating between technical solutions, sizing the risk of working alongside other trades, anticipating the customer's production shutdown constraints, and setting a price that holds for eighteen months. At Cryotem the price library exists and it is well kept; the judgment that puts it to work exists nowhere but in two heads.
Of the 90 jobs above €250k bid in 2025, 74 were priced by Michel Sarda and 16 by Serge Bonnaud. None by anyone else.
established extract from the 2025 job register (1,412 lines), “pricing owner” field, cross-checked in interview with both men and with the site management department. — What the exhibit does not establish: the share of simple jobs that benefited from an informal judgment call by one or the other — the field is left empty below the threshold.
At Montval, the same function is held entirely by Vincent Delmas.
The fulcrum: a function the organisation chart does not name
This is where the reading turns. Are the functions that carry performance identified, equipped, bounded, held by the right people? On the visible functions — site management, service operations, administration — the answer is yes, and both organisations are healthy. On the function that decides the margin, the answer is no on all four counts at once.
It is not identified. There is no job description, no title, and no “pricing” budget line. Michel Sarda reports to the engineering department as a senior technician. Serge Bonnaud reports nowhere: he steps in because he has always done so.
It is therefore not equipped. No decision-support tool, no pricing review, no pairing. The price library is updated by Sarda himself, in whatever time he has.
It is not bounded. No one can say at what amount or what degree of complexity a job “goes up”. The rule exists, it is oral, and it varies with the workload of the moment.
It is not being handed on. No handover has been set in motion, although the departure has been announced.
An engraved grid, dense and regular, in which a single cell has been left hollow — and the light rises from that cell.
Gap at eighteen months: 1.8 full-time equivalents — 56% of the requirement.
Solid: observed capacity established 2025 job register, interviews. — Hollow: calculated requirement qualified 125 ÷ 39 jobs per FTE per year, at constant turnaround — mechanism in § 13. — Montval volume (~35 jobs a year) not established — twelve rolling months only. See § 12.
Complex jobs bid within eight days convert at 44%. Those bid beyond fifteen days convert at 21%. A loss of capacity shows up in the jobs that go out late, rather than in the income statement.
established analysis of the 187 complex jobs of the 2024 and 2025 financial years: date the specification came in, date the bid went out, outcome. Two jobs excluded. — What the exhibit does not establish: causality. Jobs bid late may be late because they are harder, and therefore less winnable. Controlling by amount band narrows the gap without closing it (44% / 26% at comparable amounts).
The fulcrum sits there: a function no one has named, held by two people one of whom is leaving, and onto which the transaction loads a workload multiplied by 1.4. The relationship between the two leaders holds — § 05 establishes that.
The move played at that spot fits into three lines of documentation, and it is played before signing. It is written out in § 11.
The same fact, read a second time: what governs, and what makes things run
To manage and to govern are two verbs that common usage runs together. Separating them here changes the remedy — and this is the demarcation announced in § 06.
What the previous paragraph describes is an operational shortcoming: a function that is under-resourced. It is treated by a recruitment, a pairing and a written boundary.
The same fact reads a second time at the level of control, and there it becomes a shortcoming of another nature. Cryotem's strategy committee meets each quarter. Across the last six meetings its agenda devotes on average 71% of the time to financial subjects — revenue, margin, cash, covenants — and 14% to the order book established. No meeting has examined the capacity to produce bids. The explanation is simple: a committee tracks what it is given to track, and that figure has never been produced.
The remedy here is an agenda item and a figure to track, rather than a recruitment. Running the two readings together means treating one and leaving the other — and that is the most frequent scenario: an estimator is recruited, no one watches whether it is enough, and the problem is rediscovered in the twelfth month.
What the system does to itself, and what the transaction does to it
Here are the two movements announced in § 06, applied.
Two companies live inside each company: the one of the organisation chart and the processes, and the one that actually plays out — who talks to whom, who follows whom, who believes in it. The health of a team reads in what the two do to each other, rather than in either one alone.
At Cryotem, they reinforce each other. A decision taken in committee is applied within a fortnight; the processes help rather than hinder; attrition in the technical functions runs at 4.1% over three years, well below the sector; and people can say why they stay. This is a healthy system, and it deserves to be said to an investment committee, because it is what makes the transaction feasible.
And this healthy system is drifting away from what it was financed to do. Activity drifts towards conventional replacement work, the chairman's diary follows, and nothing in the control apparatus measures the gap with the thesis. An engine running well, under a pilot that has stopped correcting: this is success adrift, and it is the hardest state to see from the inside, precisely because every performance light is green. It is also, in our grid, one of the six combinatorial states — and the only one that produces no symptom before the mid-term review.
At Montval, they ignore each other. The organisation chart says one thing, practice does another; everything travels up to the founder; the processes exist and everyone works around them. The coupling has stopped — there is a disconnection rather than destruction or reinforcement, and value leaks out of it in silence: two people interviewed describe trade-offs made twice, once by the process and once by the founder's office. In this state the remedy is never to reorganise — reorganising a disconnection only moves its lines.
The leader who single-handedly holds Montval's critical function is asked to become operations director of a group of three hundred and thirty-five people; his pricing time is therefore taken away, with no one designated to take it over. On the other side, the volume of jobs to be handled goes up.
The predictable sequence follows from the facts: bid turnaround lengthens, the conversion rate falls, management tightens control over a system that ran on trust, and the rarest person — the one who knows how to price — is the first to draw the consequences. At that moment the two companies stop ignoring each other: they destroy each other — every turn of the screw calls a departure, every departure calls a turn of the screw, and the mechanism that made the system strong sets the speed of its fall.
That tipping point, rather than the present state, is the risk in this transaction. It is avoidable, and it is avoidable now: at this stage it is handled with three lines in the documentation and one mandate letter. After closing, it is handled with an emergency recruitment in a market where these profiles are unavailable.
The decisions, their owners, their deadlines — and what this dossier has not established.
Four decisions, three of them before signing
Here is the move played at the fulcrum of § 08. Four decisions: we investigate them, the committee takes them.
Four seals in a row: three carry a clean impression, the fourth has been left blank.
Create a title, a job description and a budget line for responsibility for the pricing of complex jobs; attach Michel Sarda to it explicitly until he leaves; write down the threshold above which a job belongs to that function. Cost: nil before closing, beyond the time to draft it. What is lost by leaving it until later: after closing, this creation becomes a post-acquisition reorganisation — the teams of both companies read it as one, and it then costs the capital of trust that will be needed elsewhere.
Michel Sarda has announced his departure on an eighteen-month horizon. He is willing to hand over — he says so in interview, unprompted. An intention is one thing, an arrangement another. What makes a handover effective: a designated and recruited counterpart, a volume of jobs priced in tandem against a stated target, a timetable, and a contractual consideration that recognises handing over as work. The market constraint is the hard point: the pool of experienced industrial refrigeration estimators is narrow and regional, and a realistic recruitment takes five to nine months. Set against the eighteen months available, the search has to open now.
He is today, alone, Montval's pricing function, and the organisation plan makes him the group's operations director. The two do not hold together over the first twelve months. Three routes: he holds both, and it is said out loud — the mandate letter bounds his group remit to the first year; he lets the pricing go, and his replacement comes first; the pricing perimeters stay separate for eighteen months. All three routes are sound. Leaving the choice open is a fourth, and it is the only one that fails for certain.
Put two figures on the standing agenda, at the same rank as cash: the average bid turnaround on complex jobs, and the number of jobs awaiting pricing. These are the only two indicators that would have shown in advance what this dossier took eleven interviews to establish. This item also handles the gap with the thesis noted in § 03: a committee that tracks the order book by line of business sees the drift in the quarter it begins.
What this dossier has not established
What is missing appears here, named. On the two points below, a conclusion would have been comfortable and wrong.
The estimate of thirty-five jobs a year rests on the last twelve months, the only window open to the investigation. The trade is seasonal — the production shutdowns of food-industry customers concentrate enquiries into two periods of the year — and twelve rolling months can depart appreciably from a multi-year average.
What would settle it: the 2023 and 2024 job registers, “amount” field and “outcome” field. Where: in the target's sales management system, already in place. How: a formal request under the pre-contractual information clause, ten working days to respond. What it costs: a letter. What it changes: if the real volume is 45 rather than 35, the requirement moves from 3.2 to 3.6 FTE and decision 2 becomes a double recruitment.
The eighteen-month horizon is the one he gave in interview. No instrument, no notification and no end-of-career agreement documents it. A date declared to a third party in the course of an assessment binds no one, and it can move closer as easily as further away.
What would settle it: a conversation held by the chief executive on end-of-career terms, which is due in any case and which decision 2 makes necessary.
A third point deserves to be flagged as an uncertainty rather than a gap: the analysis in § 08 establishes a correlation between bid turnaround and conversion, and not a causal link. Controlling by amount band narrows it without closing it. We hold it as a robust regularity; we present it as a law nowhere in this dossier.
What each person received, what we commit to, how this dossier was produced, and where the exhibits come from.
The frame
What each person interviewed received
Eleven people were heard. Each signed, before the interview, a consent form setting out the purpose of the engagement, what would be done with what they said, who would read the dossier, and how long the records would be kept.
Each receives, personally, a debrief that the client does not receive. It covers what concerns them: the reading of their own personality inventory, given back in a forty-minute conversation; what their answers made it possible to establish about the organisation; and which parts of the dossier delivered to the fund come from their perimeter. No individual debrief contains an appraisal of another person, and none of them travels up to the client.
That is the consideration given in return, and it is also what makes the process workable: a team that knows what it receives answers differently from a team that is put through an audit.
What we guarantee
This dossier investigates an investment decision; the committee takes it. Contractually the purpose of the engagement is stated in those terms, and it has not varied: we investigate the decision; we assess no one.
Four commitments follow from it, and they appear in the engagement letter. Every conclusion in this dossier is carried and signed by a person, by name — no automatic chain produces any of them. No decision is automated: the dossier prepares four decisions and takes none of them, and it produces no score on a person, no ranking, and no individual recommendation to retain or replace. Consent first, individual debrief after. Every statement carries its register — the key opens this dossier; a statement that is not established is never completed by a plausible inference.
On neutrality. Frontière has received no remuneration tied to the completion of the transaction, receives no introduction commission, and will receive no fee proportionate to any recruitment or replacement that follows from this dossier — including the post opened by decision 1, whose execution, if it were entrusted to us, would be billed as a fixed fee. A pair of eyes that neither bought the deal nor sold the team.
How this dossier was produced
The grid is in § 06. What remains is what fed it, and what settled it. Three weeks, three links.
One — the investigation. Cryotem's 2025 job register, line by line: 1,412 jobs, each tied to its pricing owner, its amount, the date it came in, the date the bid went out and its outcome. The 187 complex jobs of the 2024 and 2025 financial years taken up separately. The last six sets of minutes of the strategy committee, timed by subject. The statutory registers and the filed accounts of both companies. The regulatory retrofit timetable and three years of public consultations in the sector. Nine databases and registers were interrogated; three returned nothing, and that is written down below. This link is the least visible part of a deliverable and it decides what gets found: a reading that had seen nothing but the eleven interviews would have produced the report the committee expected, rather than this one.
Mechanism behind the requirement of 3.2 FTE (§ 08): a target volume of 125 complex jobs a year; an observed capacity of 39 complex jobs per full-time equivalent per year at Cryotem over the 2025 financial year, at constant bid turnaround; 125 ÷ 39 = 3.2. The observed capacity is one year's ratio rather than a trade standard — it is a qualified, and a change in how pricing is organised would alter it.
Two — the matrix. Described in § 06. It is not recycled from another object: a grid is built from whatever really decides the outcome of the game under study, and it is tested on known cases before being applied to a new one. Here, what decides the outcome is the speed at which a credible bid lands on the table — that is what § 03 establishes, and it is why the reading went looking at pricing rather than at the understanding between the two leaders.
Three — the judgment. The fulcrum in § 08 is a judgment, formed on what the matrix brought out and on what the interviews contradicted. No calculation produces it. It is dated, it is open to challenge, and it is signed.
The name of the thing. In the game of Go, the tesuji is the skilful move — the one played at the place that decides the game, often far from where attention has settled. That is the name we give to the last link: the reading locates, the move is played there, and someone signs it.
The instruments. A structured interview and a personality inventory from the five-factor family, both with documented validity. Typologies without established predictive validity are excluded from our protocols — they produce convenient labels and wrong decisions.
The exhibits
Each entry states what it establishes, the mechanism whenever a figure comes from a calculation, and what it does not establish — the third line is the one that counts, because it turns a gap into a question to ask.
| Exhibit | What it establishes | What it does not establish |
|---|---|---|
| P1 Cryotem 2025 job register (extract of 22 January, 1,412 lines) | Who priced what, the amounts, the turnaround times. | The informal judgment calls below the €250k threshold: the field is left empty. |
| P2 Complex jobs 2024-2025 (187 lines) | The relationship between turnaround and conversion. Mechanism: jobs won ÷ jobs bid, by turnaround band in working days; control by amount band on a second pass. | Causality. |
| P3 Minutes of the strategy committee (six meetings, Sept. 2024 – Dec. 2025) | How the time is split by subject. Mechanism: the timings stated on the agendas. | A measurement — absent real time-stamping, this is a structural approximation. |
| P4 Investment memorandum (12 March 2024) | The thesis and its assumptions. | What has since been revised without being rewritten. |
| P5 Eleven structured interviews (20 January – 4 February) | The operating regimes, the declared intentions, the cross-perceptions. | No external fact: every factual element cited in interview was cross-checked against an exhibit or marked as declared. |
| P6 Personality inventories (six administrations, five-factor family) | Dispositions, at the level of the group. | No individual prediction of performance. They appear in the dossier in aggregated form only. |
| P7 Filed accounts and statutory registers (2022 to 2024 financial years) | The financial and legal position of both companies. | — |
What returned nothing
Three registers were interrogated without result, and that belongs in the dossier: the target's register of beneficial owners adds nothing the fund already knows; the search for employment tribunal litigation against both companies is negative over five years; the review of public contracts awarded in the area leaves the target's volume of complex jobs unreconstructed, the private share of its register being preponderant.
A register interrogated without result is information; leaving it out would suggest it had never been opened.
Book 30 minutes on a real situation
A framing conversation, never a production. You describe a situation you actually have on the table; we tell you what an investigation would bring out, and what it would leave alone.
Book 30 minutesThe other door — Leaders: investigating a decision to open the share capital →
An illustrative situation. Facts in this dossier stated as at 6 February.
What is not established is not asserted.