Every group that grows past two trading entities eventually holds the same meeting. Someone asks whether the group should run a separate CRM for each company or one system across the whole thing, and the room splits along predictable lines. The managing director of the largest entity wants his own instance, because his team's data is his team's data. The finance director wants one system, because she is tired of assembling a board pack from four exports. Both are right about their own problem, which is why the argument rarely resolves on its merits.
The decision looks like procurement. It is a decision about how the group intends to behave commercially, and the software choice simply makes that intention easy or expensive to act on.
Executive Summary
Running a separate CRM for each company in a group buys clean boundaries and pays for them with the group number: nobody can see total exposure to a shared customer, and every consolidated report becomes a commissioned build rather than a query. Running one generic single-company CRM across several brands buys the group number and pays for it in leakage, because most platforms offer all-or-nothing visibility once records sit in the same instance. The deciding question is whether customers, prospects and people are genuinely shared across the entities; where they are, separate instances create a permanent reconciliation tax, and where they are not, one instance creates a permanent confidentiality problem. The architecture that actually fits a group is a single system with record-level permissions that work between brands as well as within them, which mainstream platforms treat as a configuration project rather than a foundation. Groups should decide on the basis of customer overlap and reporting cadence, not on the basis of which entity shouts loudest.
Why do groups ask whether they need a separate CRM for each company?
Most multi-entity groups did not design themselves. Entities arrive through acquisition, through a founder spinning out a second brand, or through a deliberate decision to keep a specialist trading name separate from the parent. Each one arrives with its own stack, its own spreadsheet habits and often its own CRM subscription, and the question of consolidation gets deferred until a board asks for a single view of revenue.
UK statistics recognise this shape perfectly well. The Office for National Statistics counts 2.73 million VAT and/or PAYE-registered businesses as of March 2025, and its business register is explicitly built around the idea of an enterprise group, a set of legal units under common ownership within which individual enterprises retain some autonomy. Company law, tax and national accounting all handle the concept comfortably. CRM software mostly does not, because it assumes one company, one customer base, one sales team and one set of numbers.
The reason is economic rather than technical. Mainstream platforms are built to be generic and broadly applicable, which is a rational way to address the widest possible market and a guaranteed way to fit no particular audience precisely. Groups of trading entities are a smaller addressable market with a slower, more consultative buying process, so the primitives they need never justify the roadmap space. What gets shipped instead is a configuration surface, and the configuration work lands on an implementation partner's timesheet.
A separate CRM for each company buys clean boundaries and pays for them with the group number.
What does a separate CRM for each company actually cost?
The cost of separate instances shows up in the gap between a question being asked and a number arriving. When each entity runs its own system, there is no query that answers "what is our total pipeline with this customer across the group", because no system holds the group. Someone has to export, match and reconcile, and the matching is the hard part: the same buyer appears as three unrelated records with three different spellings and three different owners.
In practice, this is what turns group reporting into a purchasing decision. Senior management ask for a view on the Monday, the work is scoped at around £2,000, and it lands a fortnight later, by which point the meeting has happened and the need has faded. Do that four times a year across four entities and the group has bought a filing cabinet of stale reports instead of a capability. The true cost of custom CRM reports is rarely the invoice; it is the decisions taken without the number.
Fragmentation also degrades the forecast itself. Salesforce's own forecasting guidance makes the mechanical point plainly: when data sits across disconnected systems, fragmented data creates blind spots, slows decisions and erodes trust in the numbers that do get produced. Gartner's July 2026 research puts a sharper edge on it for anyone hoping AI in sales will paper over the gap, warning that agents are only as effective as the systems they sit in and that fragmented systems simply scale the fragmentation.
Figure
Fragmented systems scale the problem rather than solve it
60%
of chief sales officers say their revenue number is largely driven by factors outside their control
Gartner survey of 210 CSOs and senior sales executives, early 2026
5x
more likely to gain ROI from AI by 2028 among CSOs who overhaul data, automation and user experience
versus those choosing quick fixes
10:1
predicted ratio of AI agents to sellers by 2028
yet fewer than 40% of sellers are expected to say agents improved productivity
Gartner's caution is that agents inherit the shape of the systems they run inside.
What does one shared CRM across several brands actually cost?
The opposite choice has its own bill. Put every entity into one generic instance and the group number becomes trivial, while confidentiality becomes a daily negotiation. Most systems were designed to manage visibility inside a single brand, and they do that adequately at best. Between brands, the usual reality is all-or-nothing: a user can see everything or nothing, and the middle ground gets built with custom roles, sharing rules and automations that break the first time an entity is restructured.
That matters commercially, not just legally. Two group companies bidding for the same framework need to keep their commercial positions apart. An entity being prepared for sale needs its pipeline ring-fenced from siblings who might inherit the relationships. A joint venture brand needs partner-side users who can see one account and nothing adjacent to it. And an account manager who can see every relationship in the group will, sooner or later, help himself to one. Sensible CRM permissions across group companies exist to protect deals from colleagues as much as from outsiders.
There is a data protection dimension too, though it is often overstated in these conversations. As of 10 September 2026, UK data protection law does not require separate systems per legal entity. The Information Commissioner's Office defines a controller by who determines the purposes and means of processing, and notes that organisations processing the same data for different purposes are not joint controllers. Where group companies do decide purposes together, Article 26 of the UK GDPR requires a formal arrangement setting out responsibilities, and the ICO's data sharing code recommends a documented data sharing agreement for routine sharing, including attaching permissions so that only particular roles can reach particular items. Guidance in this area is under review following the Data (Use and Access) Act, so confirm the current position before relying on it.
How do the three options compare for a group of companies?
| Dimension | Separate CRM per company | One generic shared instance | One instance, entity-aware permissions |
|---|---|---|---|
| Group pipeline number | Manual consolidation, usually a commissioned build | Available immediately | Available immediately, filtered by entitlement |
| Sharing a lead across brands | Email, spreadsheet or re-keying | Trivial, and often uncontrolled | Explicit handover with an audit trail |
| Account confidentiality | Strong by default | Weak; typically all-or-nothing visibility | Configurable per record, per role, per brand |
| Licence and admin overhead | Multiplied by entity count | Single subscription, heavy configuration | Single subscription, group-shaped setup |
| Reporting effort | Rises with each entity added | Falls, until brand-level reporting is needed | Rolls up and drills down natively |
| Typical failure mode | Group blind to shared customers | Reps see deals they should not | Requires the vendor to have built it in the foundations |
The third column is the one the market underserves. Fine-grained visibility control within a brand and between brands is a foundational primitive, close to impossible to retrofit convincingly, which is why groups so often end up paying for millions of implementation hours to approximate it. Building the group case into the foundations rather than the configuration layer is the reason Atrium exists.
Which question actually settles it?
Customer overlap settles it. If the entities sell to genuinely separate populations, with no shared buyers, no referrals and no bundled propositions, then separate systems are defensible and the only real cost is reporting. The moment the same buyer, procurement function or parent organisation appears in two entities' pipelines, separate instances start destroying value that the group has already paid for.
McKinsey's M&A research quantifies how fragile that value is. Fewer than one in five organisations it polled in the M&A space achieved their cross-selling goals, despite cross-selling accounting for roughly a fifth of the value companies derive from revenue synergies, with the majority of synergies taking three to five years to capture. In a separate survey of 200 experienced M&A executives, McKinsey found an average shortfall of 23% between revenue-synergy goal and attainment. Groups that keep buying companies on a cross-sell thesis and then keep their customer data in four unconnected systems have picked the architecture that makes the thesis hardest to deliver.
Figure
Cross-selling is where group value leaks
<20%
of organisations polled in the M&A space achieved their cross-selling goals
~20%
of the value companies derive from revenue synergies comes from cross-selling
the largest single lever in McKinsey's sample
3-5 yrs
to capture the majority of revenue synergies
average gap between goal and result around 20%
McKinsey surveyed more than 75 M&A executives across 12 industries on cross-selling experience.
Take a group of four trading entities: a fit-out contractor, a maintenance business, a plant hire company and a small consultancy, with 26 sales and account staff between them. Three of the four sell to the same national retail estate, and the retailer's facilities director sits in three separate CRMs as three unlinked contacts. On separate instances, nobody in the group knows that the retailer represents a mid-six-figure relationship across three brands, so nobody treats it like one. The maintenance business discounts to win a renewal that the fit-out arm could have protected on relationship strength. The consultancy pitches cold to a buyer who has been a customer for six years. Each entity's system is accurate; the group's picture of itself does not exist. That pattern is exactly why sharing leads between companies in the same group is worth more than most groups assume.
Who is this decision actually for?
The buyer and the user want different things here, which is why the meeting stalls. A group CEO or FD may never open the CRM except to look at reports, so simple, flexible and reliable group reporting is the whole of their interest. Sales and account teams live in the system all day and care about whether they can pass a lead to a sibling brand without losing credit, and whether the account they have nurtured for four years is visible to a rival entity's new business team.
A group that decides this question on the FD's requirements alone will consolidate and then spend two years fighting leakage. A group that decides on the entity MDs' requirements alone will keep four clean systems and never see itself whole. The decision belongs to whoever is accountable for the group behaving as one commercial organisation, and the honest test of any option on the table is whether it serves both readers of it. Groups weighing this alongside cost should read the wider case for a multi-entity CRM before committing to either extreme.
Frequently Asked Questions
Should each company in a group have its own CRM?
Only where the entities share no customers, no referrals and no salespeople. Where buyers, prospects or account teams overlap across brands, separate systems impose a permanent reconciliation cost and make cross-selling harder, which matters given McKinsey's finding that fewer than one in five organisations in the M&A space achieved their cross-selling goals.
Can one CRM keep customer data separate between group companies?
Technically yes, practically it depends on the platform's foundations. Systems designed for a single company usually offer visibility that is close to all-or-nothing once records share an instance, and the middle ground is built through custom roles and sharing rules. Record-level permissions that operate between brands as well as within them are the primitive that makes genuine separation inside one system possible.
Does UK data protection law require a separate CRM for each legal entity?
No. The Information Commissioner's Office determines controllership by who decides the purposes and means of processing, and group companies may be separate controllers, joint controllers, or neither, depending on the arrangement. Where controllers jointly determine purposes and means, Article 26 of the UK GDPR requires a documented arrangement; ICO guidance in this area is under review following the Data (Use and Access) Act, so the current position should be confirmed before acting.
How can a group get one consolidated pipeline view without merging its brands?
By holding the data in one place and controlling visibility at record level rather than at system level. Consolidation of reporting and separation of access are independent design choices, and platforms that conflate them force groups into either commissioned reports or uncontrolled sharing. Gartner's 2026 research warns that fragmented systems propagate fragmentation into any AI or analytics layer placed on top.
What is the hidden cost of running several CRM instances across a group?
Beyond duplicated licences and administration, the recurring cost is bespoke reporting. Consolidated group views typically have to be commissioned rather than queried, and the commissioned build arrives after the decision window it was meant to inform. Multiplying that across entities produces a reporting bill that grows with every acquisition.