Why FDD on a multi-entity group starts with structure, not EBITDA: perimeter risk, intercompany eliminations, entity-level P&Ls, FX, and dormant-entity traps.
The target that presents itself to a buyer as "one business" is, on the legal papers, almost always a cluster of entities - operating subsidiaries across several countries, a holding company on top, and usually a dormant shell or two left over from a reorganisation nobody remembers. Before your EBITDA bridge means anything, before your net debt schedule ties, you have to answer a deceptively simple question: what, precisely, is being bought, and how do the pieces fit together? Get the structure wrong and every number downstream is a confident answer to the wrong question.
The first genuinely important question on a multi-entity target is not "what's the EBITDA" - it is "what is the perimeter?" Buyers sometimes acquire the whole group; sometimes only named entities, leaving others (and their liabilities) with the seller. This is not a technicality. It determines which P&Ls you consolidate, which debt is assumed, and which working capital sits inside the deal.
Perimeter risk here is a starting condition, not a mid-process surprise - closer in spirit to a carve-out than to a normal single-entity review. If the deal excludes two subsidiaries you have folded into every schedule, you do not adjust a number; you rebuild the analysis. The cheapest hour on any group engagement is the one you spend confirming the perimeter against the SPA draft before you touch the data.
On a group, "what's the EBITDA?" is meaningless until you can answer "the EBITDA of what?" Nail the perimeter first, or nail nothing.
Perimeter is not always a clean list of entities, either. A deal can carve a single trading division out of an entity that also houses activities the buyer does not want, meaning the legal-entity boundary and the economic boundary do not coincide. In those cases you are effectively running a mini carve-out inside the consolidation exercise, and the perimeter question becomes "which contracts, employees, assets and liabilities travel with the deal, and which stay behind?" That is a legal and commercial question as much as a financial one, and it is why perimeter has to be confirmed with the deal team and counsel rather than assumed from an organigram. The most expensive perimeter errors are the ones where an analyst quietly decided the boundary themselves and nobody checked.
Consolidated group accounts strip out transactions between group entities: a sale from Subsidiary A to Subsidiary B is not a sale to the outside world, so it should not inflate consolidated revenue. Mechanically simple, frequently botched. Two failure modes recur:
A worked example makes the revenue distortion concrete. Suppose the group reports as follows before you check the eliminations:
| Line | Sub A | Sub B | Naive sum | Correct consolidated |
|---|---|---|---|---|
| External revenue | 40.0 | 25.0 | 65.0 | 65.0 |
| Intercompany sales (A→B) | 10.0 | - | 10.0 | eliminated |
| Total revenue booked | 50.0 | 25.0 | 75.0 | 65.0 |
| Cost of sales | (35.0) | (20.0) | (55.0) | (45.0) |
| Gross profit | 15.0 | 5.0 | 20.0 | 20.0 |
| Gross margin % | 30% | 20% | 26.7% | 30.8% |
Note the trap: gross profit is £20.0m either way, so a quick EBITDA check looks fine. But the naive margin of 26.7% is nonsense - the real figure is 30.8%, and any growth or benchmarking analysis you run on the inflated £75.0m revenue base is wrong. Incomplete eliminations do not always change the profit; they poison the ratios.
The elimination problem also has a timing dimension that is easy to miss. Intercompany balances have to reconcile on both sides - Sub A's receivable from Sub B must equal Sub B's payable to Sub A. When they do not, someone has booked a transaction in a different period, at a different value, or not at all. Those mismatches are not merely tidy-up items: an unreconciled intercompany difference can be masking a real revenue-recognition problem, an FX translation error, or a balance that one entity has quietly written off while the other still carries it as an asset. Part of your standard group procedure should be to obtain the intercompany reconciliation and confirm it actually nets to zero across the group before you rely on any consolidated line.
There is a cash dimension too. Consolidation eliminates the profit and loss effect of intragroup trading, but it does not tell you whether the group's cash is trapped. If one subsidiary sits on the cash and another needs it, dividends, thin-capitalisation rules or local restrictions may prevent it moving freely - which matters enormously for the net debt picture a buyer is pricing. Cash in a subsidiary you cannot easily upstream is not the same as cash at the top of the group, and a naive consolidated cash figure treats them as identical.
If the deal acquires only certain entities, consolidated group EBITDA is close to useless for pricing. You need entity-level (or carve-out) numbers, which means unwinding shared costs, intercompany charges, and central overhead allocations that were never designed to be pulled apart. There is rarely a clean, pre-built entity-level P&L sitting in the data room, and reconstructing one is some of the most painstaking work in the discipline.
The core problem is that head-office costs - group finance, IT, insurance, the CEO's time - are allocated by conventions built for internal reporting, not for a standalone sale. A subsidiary that "pays" a modest management charge to the holdco may in reality consume far more central resource; strip it out and the entity's true standalone cost base is understated. This is exactly the standalone-cost analysis that dominates carve-out FDD, and it feeds directly into the EBITDA adjustments you can defend. It also cascades into the EV to equity bridge: entity-level net debt and working capital have to be reconstructed on the same perimeter as entity-level EBITDA, or the bridge does not reconcile.
Multi-jurisdictional groups often consolidate entities reporting in different local currencies, translated into the group's presentation currency at rates that shift year to year. That alone means an entity-level figure viewed in isolation may not be comparable to the consolidated line, and year-on-year "growth" can be partly an FX artefact. Before you attribute a movement to trading, confirm whether it is real or a translation effect.
The subtler trap is accounting-standard mismatch. Individual entities may report under local GAAP before conversion to the group standard (IFRS, say). A number you lift straight from an entity's statutory accounts may be built on a different basis for revenue recognition, leases, or provisions than the consolidated figures you have been working with. Two areas bite most often:
| Area | Local-GAAP risk | Why it matters for FDD |
|---|---|---|
| Leases | Off-balance-sheet under old local rules | Understated debt-like items; see IFRS 16 |
| Revenue recognition | Different cut-off / point-in-time vs. over-time | Distorts entity-level growth and margin |
| Provisions | Recognised earlier/later than group basis | Skews normalised EBITDA and net working capital |
Pull an entity number without checking its underlying basis and you can import a leverage or timing distortion into your model without realising it - the kind of quiet error that surfaces only when the net debt schedule refuses to tie.
Groups accumulate dormant and non-trading entities - remnants of old reorganisations, discontinued ventures, holding structures set up for tax or legal reasons that never generated real activity. Individually they rarely move the headline numbers, so the temptation is to ignore them. Resist it, at least for one pass. A "dormant" entity can still carry a residual liability: an old lease guarantee, an unresolved dispute, a tax exposure from a prior year, a warranty that never expired. "Dormant" describes trading activity, not risk. A quick confirmation that each non-trading entity is genuinely clean is far cheaper than the alternative.
The practical discipline that prevents most of these problems is boring and non-negotiable: draw the group structure chart before you build a single schedule. Every legal entity, ownership percentages, reporting currency, accounting basis, and - critically - a clear line around the acquisition perimeter. Overlay where the intercompany flows run and where central costs are borne. Once that chart is right, the eliminations, the entity-level reconstructions, and the FX and GAAP adjustments all have a home. Without it, you are consolidating in the dark and hoping the numbers happen to tie.
This structure work is also what makes your FDD report legible to a deal team that has never seen the legal architecture. A one-page perimeter chart at the front of the report does more to prevent mispricing than another ten pages of schedules.
A practical way to keep the chart honest is to reconcile it back to two independent sources: the statutory filings for each entity (which confirm the legal existence, ownership and reporting basis) and management's own group reporting pack (which shows how they actually consolidate in practice). Where the two disagree - an entity in the legal structure that never appears in the reporting pack, or a reporting unit that maps to no single legal entity - you have found either a dormant shell worth a look or a consolidation shortcut worth understanding. Neither is necessarily a problem, but both are exactly the kind of loose thread that, left unpulled, becomes a nasty surprise in week three.
Interviewers use group-structure questions to see whether you understand FDD as more than single-company arithmetic. A common one: "You're staffed on a multi-entity group and management hands you consolidated accounts. What's the first thing you do?"
"Before I look at EBITDA, I'd establish the perimeter - what's actually being acquired, because on a group it's often specific entities rather than the whole thing, and that determines which P&Ls, which debt and which working capital are even in scope. So my first move is to build the group structure chart: every legal entity, ownership, reporting currency, accounting basis, and a clear line around what's in the deal. Then I'd check the intercompany eliminations, because incomplete eliminations can inflate revenue and costs together - the EBITDA might still be right by coincidence, but the margins and growth rates are wrong. If the deal is only part of the group, consolidated EBITDA is close to useless and I'd need to reconstruct entity-level numbers, which means unwinding central cost allocations that were never built to be pulled apart. And I'd sanity-check for FX and local-GAAP differences before comparing any entity figure to the consolidated one. The theme is that on a group, structure comes before numbers - get the perimeter wrong and every schedule answers the wrong question."
That answer lands because it leads with perimeter, names the specific consolidation traps, and shows you know entity-level reconstruction is where the real work sits.
Multi-entity consolidation work is unglamorous next to building a headline EBITDA bridge, and it never makes the highlights page of a report. But it is the foundation the whole engagement stands on. Getting the group structure and perimeter right at the outset is what stops you rebuilding every schedule in week three when it turns out the deal excludes two subsidiaries you had baked into everything. On a group, the analyst who draws the chart first is the analyst who finishes on time - and finishes right.
The Transaction Services Interview Programme (€119.99, one-time) includes a dedicated module on group and multi-entity engagements - mapping the acquisition perimeter, spotting broken intercompany eliminations, reconstructing entity-level P&Ls from consolidated accounts, and handling FX and local-GAAP mismatches under interview pressure. Enrol today.
Hundreds of candidates prepared their interviews with this programme. Those who landed the role have one thing in common: they worked the cases before walking into the room.
