The case study: a compliance consultancy, from close through conversion

The firm at close

Call it Northgate. Forty-five people in two offices, advising manufacturers and brands on product, chemical, supply-chain and sustainability compliance. Founded by two partners nineteen years ago; six partners now, four of whom hold most of the client relationships. The founders want to retire within five years and have no internal succession.

The P&L at close, composite and illustrative:

Entry at an assumed 7x EBITDA, $8.4M enterprise value, in a structure of 55 percent cash at close, 20 percent deferred over three years against scorecard metrics (P7), and 25 percent rollover into the platform holding company. The structure is illustrative; the market data in the risk chapter says the sector's multiple is lower than it was and the private market has held better than the public one, and the deal team should price from current comps, not from this page.

What diligence found

P1, work decomposition. Trailing-twelve-month time records exist because the retainers are managed to hours. Mapped to units across the three largest service lines (supplier compliance programs, substance and regulatory screening, sustainability reporting), the delivery cost splits: 58 percent codifiable (supplier data requests and chasing, substance screening against lists, declaration and report drafting, regulatory change monitoring), 37 percent judgment (gap analysis, remediation advice, client review meetings, new-regulation interpretation), 5 percent reserved (a small amount of accredited test coordination). The ceiling is 58 percent of $5.6M, or about $3.2M of delivery cost that can in principle move. Assumption: the split is illustrative of the archetype; a real firm's audit will differ.

P2, revenue quality. Rebuilt from invoices: retainers renew at 91 percent by count, but a third of them were repriced downward at the last renewal when clients saw hours fall. Realization is 84 percent: the firm records hours it does not bill. Top client is 9 percent of revenue; top ten are 46 percent. The "what would they pay for the outcome" exercise, run with the partners, lands at roughly current retainer values for the standard program, which means the fixed price recovers the 16 points of unbilled realization and nothing more. Assumption.

P3, relationships. Four partners hold 71 percent of revenue between them. Two are the founders. Non-competes are enforceable in one office's jurisdiction and doubtful in the other. All four meet; three commit to roll; the fourth, holding 14 percent of revenue, wants cash and a two-year transition. That is a structure problem solved with a longer earn-out for him and a named successor from the firm's senior managers.

P4, contracts. All 212 agreements read. Fourteen percent of revenue is under client master agreements that require named personnel or restrict subcontracting in ways that catch a platform; all renew inside eighteen months. Six percent of revenue is under agreements whose data-protection terms prohibit third-party processing without consent; consent is obtainable. No agreement grants derivative data rights; every one must be amended for R10 to exist.

P5, regulation. No ownership rule applies. The reserved 5 percent is accredited test coordination that stays with an accredited partner lab.

P6, platform fit. The platform team runs twenty sampled engagements. Of the codifiable units, 45 percent are "ready" (supplier data collection, list screening, change monitoring), 35 percent "configure" (the firm's declaration formats, its client-specific report templates), 20 percent "build" (a sustainability-reporting calculation the platform does not yet do; six months). Error rate on ready units against the firm's own output: 4.1 percent on a sample of 400, versus a manual rework rate the firm estimates at 3 to 5 percent but has never measured (D6 will). Assumptions throughout.

P7, the model. Base case: P9 and R1 on the retainer book from realization data; D1 on ready and configure units; D5 (the firm spends $340K a year on subcontractors); D6 as a baseline exercise; O4, O5, O6 from carried evidence. R2 in the base case, because the build is six months. Upside, priced at zero: R6, R9, R10, and the re-rating. Exit modelled at the entry multiple as the base.

Year one

Day one to one hundred (P8, P9, D4, D6, D8). Client files migrate; the gateway goes live with the firm's confidentiality rules; the scorecard baselines publish: recurring share 55 percent (by habit; 38 percent by contract), gross margin 38 percent, revenue per delivery FTE $265K, on-platform share zero, book $9.0M, NRR not yet measurable, named-person retention 100 percent, manual rework rate (measured for the first time, D6) 4.4 percent. The utilization report leaves the management pack in week six, replaced by units and error rate. Two senior managers spend half their time for the quarter writing the firm's screening and declaration method into frameworks (D4), paid for explicitly (O7).

P9 runs in the first renewal cycle: 60 percent of renewing retainer revenue converts to a fixed annual subscription at a price recovering most of the realization gap, delivered by the same people. Two clients decline and keep hourly terms at a premium. One client, 2 percent of revenue, leaves, citing the ownership change rather than the price. Assumptions.

Months four to twelve (D1, D2, D3, D5, R2, R11). Ready units go live in month four with 100 percent review (D3). By month nine, two consecutive samples show platform error at 3.6 percent against the 4.4 percent manual baseline and sampling falls to one in five. Configure units go live in month seven. By year end, 30 percent of codifiable units are delivered on-platform. Production hiring froze at close; three analysts left and were not replaced; two became reviewers. Subcontractor spend falls to $120K (D5). R2 launches in month ten: continuous monitoring with the annual assessment included, for eight clients from P3's most committed list, on a premium tier that names a partner (R11).

Year-one P&L (illustrative). Revenue $9.3M (price recovery on converted retainers, offset by the lost client and project softness as the firm's attention turns inward). Delivery cost $5.1M (two fewer analysts, subcontractors down, platform cost up $180K). Gross margin 45 percent. Overhead $2.3M (foundation costs expensed; one-time). EBITDA $1.9M, 20 percent. Scorecard: recurring share 61 percent by contract; gross margin 45; revenue per delivery FTE $300K; on-platform 30 percent of codifiable units; book retention 97 percent; NRR 103 percent; named-person retention 100 percent; error rate 3.6 versus 4.4; R2 live with eight clients.

Years two and three

Year two (R1, R8, R3, D1, D2, O-group). The second renewal cycle converts the rest of the retainer book; hourly revenue falls to 6 percent. R8 converts half the episodic assessment clients to monitoring subscriptions. R3 sells scope: more jurisdictions and regulations per client, configured rather than staffed; NRR reaches 108 percent. On-platform share reaches 55 percent of codifiable units as the build units land. Delivery headcount falls to 26 through attrition; eight are now reviewers; span of review is published. The finance system is rebuilt on units and subscriptions (O4); the time system is retired (O6). Partner compensation moved to the scorecard formula at the first cycle (O7); one partner, not a relationship holder, leaves over it.

Year three (R6, D7, X3, X6). A base tier opens to smaller manufacturers at platform cost (R6); 31 new logos at a tenth of the average client's revenue. Platform cost per unit is managed down (D7) as the firm routes screening to smaller models. Data-consent amendments cover 58 percent of revenue (X3), enough to give a supplier-risk benchmark to premium-tier clients, not yet enough to sell it. The exit scorecard is produced quarterly.

Year-three P&L (illustrative). Revenue $11.5M (NRR 108 percent on a retained book, down-market logos, R8 subscriptions). Delivery cost $5.1M (26 people, platform cost $520K, subcontractors near zero). Gross margin 56 percent. Overhead $3.1M (marketing and inside sales built from near zero; shared platform cost allocated; partner compensation reset). EBITDA $3.3M, 29 percent.

The scorecard, three dates

Every figure above is an assumption chosen to be plausible for the archetype and consistent with the lever cards' mechanisms; none is evidence, and a real firm's register will replace all of them.

Three exits, and the honest finding

At year three, EBITDA is $3.3M on $11.5M of revenue, 74 percent of it recurring by contract, at a 56 percent gross margin. What is it worth?

Three things follow, and they are the reasons this chapter exists.

First, the delivery margin does most of the work. The move from 13 to 29 percent EBITDA on a growing book is worth 2.7x the entry value at the entry multiple, with no re-rating at all. That is the return a sponsor should underwrite, and it is earned by P9, D1, D2 and D5 in that order.

Second, at 2026 software multiples the re-rating is smaller than the thesis's critics and its promoters both assume. Public software trades near 3x revenue, not the 20x of the last cycle, and a converted services firm priced as software at today's multiples is worth about what a sponsor would pay for it on EBITDA with a premium.¹ The category-error argument is right that a services firm will not trade like 2021 software; it is also true that 2026 software does not trade like 2021 software. The re-rating case is real but it is a few turns, not a different asset class, and it should be modelled as upside.

Third, the exit cases converge, which is good news for the thesis and bad news for anyone who pitched it as arbitrage. A firm that is worth roughly the same to a services buyer with a premium, a sponsor and a software strategic is a firm with three exits instead of one, and that is worth more than any single multiple.

What killed what, on the way

For the register's record, the levers that hit kill criteria in this composite and what was done:

• D1, sustainability-reporting build unit: on-platform share stalled below 50 percent two quarters after go-live because the calculation needed client data the firm did not hold. Re-tagged to judgment pending a data-collection change; the build was not wasted but the unit's year-one contribution was zero.
• R6, first quarter: cost to serve the down-market tier exceeded revenue because onboarding was manual. Paused one quarter while onboarding moved to the platform; relaunched.
• O7: one partner left after the formula change. Not a relationship holder, so not a book event; recorded as the cost of the lever.

Nothing else was killed. A register where nothing is ever killed is a wish list; one where a build unit, a growth tier and a compensation change each hit their criteria and were acted on is the system working.