A family-owned conglomerate is not a company. It is a set of companies held together by a family, a holding board and a shared name. Its intelligence function is usually designed as if it served one of those three things. It has to serve all of them, and the design that does so is different from the one a single-industry firm would build.
The position of this article is simple. The function belongs at the centre, with a shared taxonomy and a shared standard of evidence. Its analysts sit in the operating companies. Its output has two forms, a board-grade brief and an operating dashboard, and neither is allowed to pretend it is the other. Governance decides who commissions, who reads and what is confidential to whom, and it is written down before the first memo. Most of the rest is discipline.
The information environment
Consider what the group actually is as a place where information moves. Six industries, perhaps: a bank, a cement business, a hospital group, a distribution arm, a property company, a telecoms stake. Each has a chief executive with a profit-and-loss account, a management team and a market that behaves nothing like the others. Above them sits a holding board, usually mixed: family members, a few independent directors, the group chief executive. Beside it, or above it depending on the family, sits a family council, which is not a corporate body and does not pretend to be one. And there is the Sunday board, the meeting after lunch, where the decisions the formal board will ratify on Tuesday are actually taken.
Information flows through this structure along two channels. The formal channel is the board pack: quarterly, long, prepared by the operating companies, edited by the holding. The informal channel is the family. A cousin heard something at a wedding. A brother-in-law's supplier is in trouble. The minister's adviser hinted at a tender. This channel is fast, trusted and unevidenced, and in most family groups it outranks the formal one.
Between the operating companies runs a third condition: silence. The bank may not know what the property company is planning, and often should not. Sister companies compete for capital at the holding and occasionally for the same customer in the market. Confidentiality between them is a real requirement, not a cultural quirk, and in regulated subsidiaries it is a legal one.
Why the function fails
Intelligence functions in family groups fail in a small number of repeatable ways.
The first failure is to serve the board only. A group strategy team at the holding writes elegant briefs on macro conditions, sovereign programmes and adjacent sectors. The operating companies never read them, because the briefs answer no question a chief executive has this quarter. The function becomes a secretariat for the chairman's curiosity.
The second is the mirror image: serving the companies only. Each operating company hires its own analyst, subscribes to its own databases and builds its own dashboards. The holding receives six reports in six formats, on six definitions of market share, and cannot compare its own businesses. Capital allocation, the one decision only the holding can take, is made without the evidence the group has already paid for six times.
The third is the dinner table. A well-evidenced memo says a market is contracting. A family member says a friend in that market is doing fine. The memo loses. It loses not because the family is irrational but because the memo was not designed to survive the encounter: it does not say how confident it is, what source it rests on, or what evidence would change its position. An anecdote with a face beats a paragraph without a footnote.
The fourth is the absence of a shared taxonomy. Without agreed definitions of a market, a segment, a competitor and a customer, the group cannot aggregate anything. Market share in the cement business means tonnes; in the bank it means deposits; in the hospital group it means beds or admissions depending on who is asked. Nothing rolls up, so the board reads six dialects.
The fifth is confidentiality mishandled in either direction. Either the function shares everything, and the regulated bank's deposit data appears in a group deck the property company's chief executive can read, or the function shares nothing, and the intelligence about a common supplier, customer or regulator never reaches the sister company that needs it.
An anecdote with a face beats a paragraph without a footnote. The function has to be designed to survive that encounter.
A design that serves both
The design that works has four parts.
A central function at the holding owns the method: the taxonomy, the source register, the standard of evidence, the memo form and the confidentiality rules. It does not own the questions. It is small, and its authority is over how intelligence is produced, not over what the companies are allowed to ask.
Embedded analysts sit inside each operating company, report day to day to that company's chief executive or strategy head, and report on method to the centre. They know the industry. They use the group taxonomy and the group standard, so what they produce can be read at the holding without translation. They are the reason the operating companies read the function's output at all: it answers their questions, in their building, on their calendar.
A cadence with two registers. The board-grade brief is written for the holding board: the question, the position, the confidence, what would change it, the recommendation and the owner, at a length that is read before the meeting. The operating dashboard is for company management: current indicators against agreed definitions, refreshed on the operating cycle, with alerts when a held position moves. The brief is not a summary of the dashboard. The dashboard is not a longer brief. They answer different questions for different readers.
A clear line between three kinds of work. Market intelligence is about markets: size, structure, growth, regulation, the forces that move them. Competitive intelligence is about named rivals and what they are doing, and is bounded by competition law from the first day. Due diligence is about a specific counterparty in a specific transaction, and is commissioned, scoped and closed. Family groups habitually confuse the three. The same analyst is asked to size a market on Monday, track a rival on Tuesday and vet an acquisition target on Wednesday, with no change in method, legal basis or confidentiality regime. The design separates them so the group can do each properly.
Sourcing discipline and stated confidence
The defence against the dinner table is provenance. Every conclusion carries its sources, the date each was retrieved, an assessed reliability, the corroboration found or not found, and the analyst who signed it. A director who disagrees can ask for the chain and receive it. The anecdote cannot offer the same, and the difference becomes visible to the whole table.
Confidence is stated in words, not implied by tone. A workable scale has three or four levels, each defined by the evidence behind it: confirmed by several independent sources of good reliability; supported by one good source or several weak ones; inferred from indirect evidence; or assumed in the absence of evidence, with the assumption named. The confidence line sits at the top of the memo, not in the prose. A low-confidence position is still useful if it is honest. A high-confidence position built on a single unverified source is the failure the whole apparatus exists to prevent.
The source register is governed. Every source the function may use is entered with its type, licence terms, reliability rating, refresh frequency and legal basis. Legal reviews the register before the function starts and at least once a year afterwards. Sources are added and withdrawn on the record. Nothing is used because an analyst found it; it is used because it was admitted.
Collection is lawful by construction. Public records, regulatory filings, tender portals, licensed databases, trade press, company publications and expert conversations under disclosed terms. No pretexting, no approach to a rival's employees for confidential material, no purchase of documents a seller had no right to sell. In a family group this matters more than elsewhere. The family's name is on every business, and a collection scandal in one company is a reputational event in all of them.
What is different in the region
Family conglomerates in the Middle East and North Africa work in an environment with specific properties, and a function designed for London or Zurich will miss them.
Arabic-language sources carry the signal. Regulatory gazettes, ministerial decisions, court records, licensing announcements and much of the business press are published in Arabic first and often only. A function that reads English coverage is reading a translation, late and filtered. Analysts who read Arabic are not an addition to the collection capability. They are the collection capability.
The state is a market participant. Sovereign programmes, national visions, public investment funds and state-owned enterprises shape demand in most of the sectors the group operates in. Their published plans, budgets and procurement pipelines are primary sources. Tender portals, national and sector-specific, are where demand becomes visible before it becomes revenue, and monitoring them is a discipline in itself: registration, alerts, classification against the taxonomy, and a record of what was tendered, what was bid and by whom.
Regulatory gazettes are the earliest warning. Draft regulation, consultations and gazette publications announce obligations months before enforcement. In a group with a bank, a hospital and a telecoms stake, three regulators are publishing at once, and the function's job is to read each for the obligation it creates and the date by which it bites.
Personal data has a legal regime. Where the function's work touches personal data (a customer list from an operating company, an expert's contact details, a rival executive's public profile), Jordan's Personal Data Protection Law applies to processing within its scope. It requires a lawful basis for the processing, limits the use of the data to the purpose it was collected for, and gives the data subject rights the group must be able to honour. Comparable obligations under the GDPR apply where the group processes the data of people in the European Union. The practical consequence is that the source register records the legal basis for any source that contains personal data, and the runbook contains a procedure for a data-subject request, because one will arrive.
Intelligence and the group brand
A family group has a brand architecture whether or not it has designed one. The family name sits on some companies and not others. Some subsidiaries trade under their own marks; some carry an endorsement; some are deliberately unbranded. Which arrangement is right is a strategic question, and it is answerable with intelligence rather than taste.
The function can tell the board how each market perceives the group name: as a guarantee of standing, as a signal of domestic or foreign ownership, as a liability where a sister company has had a difficult year. It can tell the board where the name carries a premium and where a subsidiary would do better under its own mark. It can say which regulators and which sovereign programmes favour a visibly local group and which do not care. A group brand architecture built on that evidence is defensible. One built on family sentiment is a decision waiting to be reversed by the next generation.
The reverse holds as well. A brand decision creates an intelligence requirement. If the group chooses to put the family name on the hospital, it has chosen to monitor the reputation of the hospital as a group matter, and the function's scope changes on the day the sign goes up.
Governance
Governance answers three questions, and it answers them in writing before the function starts.
Who commissions. The holding board, through the group chief executive or a nominated committee, commissions board-grade work. Operating company chief executives commission work for their companies through their embedded analyst. The family council does not commission directly; its questions go through the holding board, which decides whether they are questions the function should answer. This is the rule most often resisted and the one most necessary.
Who reads. Board briefs go to the holding board and to the chief executive of the operating company they concern. Operating dashboards stay in the operating company. Cross-company intelligence, on a shared supplier, a shared regulator or a shared customer, is released to sister companies only under a rule the general counsel has approved, and never where a regulated subsidiary's confidential data would travel with it.
What is confidential to whom. The most sensitive material the function holds is the list of questions the board is asking, because the questions reveal intent. That list is held at the centre and seen by the fewest people. Below it sits a tiered scheme: group-confidential, company-confidential, and shareable across the group. Every memo carries its tier on the first page. Competition law adds a fourth boundary. What may be collected about a rival, what may be shared inside the group, and what must never be exchanged with the rival itself, is written into the collection rules before the first competitor file is opened.
What to do on Monday
- List the decisions the holding board must take in the next twelve months, with an owner and a date for each. That list, not a subscription catalogue, is the function's first requirement.
- Write the taxonomy: one definition each of market, segment, competitor and customer, agreed with every operating company chief executive. Until it exists, nothing rolls up.
- Ask the general counsel to draft the confidentiality tiers and the competition-law boundaries, and to review the source register before any analyst uses it.
- Fix the confidence scale and put the confidence line at the top of every memo, from the first one.
- Assign one Arabic-reading analyst to the operating company with the most regulatory exposure, and have them monitor the gazette and the tender portals for a full quarter before anything else is asked of them.
- Decide the family council's route to the function, and write it down.