Engagement models: project, retained or rhythm

By the partners · Engagement patterns · 12 min read

Boards spend more time choosing an adviser than choosing how the adviser will work. It is the wrong order. The engagement model determines who answers the phone, what the advice costs to obtain, how fast it arrives, and — quietly but decisively — what the adviser's incentives look like in month four. Get the model wrong and even excellent advice arrives uselessly: a project report nobody is left to act on, or a retainer that quietly becomes a subscription to someone's overhead.

This piece sets out the three models we offer, the situations each fits, and the failure modes we have watched boards walk into. It is written from practice: the examples are composites of real mandates, anonymised as always.

The project model

A project is a bounded question with a beginning, an end and a deliverable the board can hold. Strategy resets, market-entry decisions, pre-sale preparation, governance rebuilds — all naturally project-shaped. The virtues are clarity of scope and finality of cost. The board knows what it will receive, when, and within which budget envelope; the adviser knows the same. Nothing about the arrangement survives the final report, which is precisely the point.

Projects fail in two characteristic ways. The first is scope drift: the question was "should we enter the DACH region" and by week six it has become "how should the group be organised", which is a different and larger question that was never priced. Good engagement letters name the question so precisely that drift is visible to both sides. The second is the cliff edge: the project ends, the report is presented, and the organisation discovers that deciding is different from doing. Boards should budget for a short landing strip — a monthly session for two or three quarters — from the outset, rather than negotiating it awkwardly at the end.

The retainer model

A retainer is capacity, not content: a standing right to the adviser's time and judgement, priced by the month or quarter. It suits situations where questions arrive unpredictably and speed matters more than structure — a chief executive navigating a shareholder dispute, a chair managing director dynamics, a finance director who wants a second read on anything the bankers put in front of her.

The failure mode is subscription rot. A retainer bought for access can decay into a habit that nobody audits: the adviser is in the diary, the invoices are paid, and no one can say what changed because of it. We guard against this the way good audit committees guard everything: an annual written review of what the retainer was used for, what it influenced, and whether it should be renewed at all. Advisers who resist that conversation have answered it.

The rhythm model

The rhythm model — our most common and, we think, most under-used — places an adviser inside the board's existing calendar. Monthly board presence, quarterly strategy reviews, an annual agenda redesign and effectiveness review. Nothing is project-shaped because the work is not a project; it is participation. The adviser reads the same papers as the directors, watches decisions form, and intervenes earlier than any external report ever could.

Rhythm work is priced by the year because its value compounds: the third quarterly review is worth more than the first, because the adviser arrives knowing what the last one settled. Its failure mode is different — the adviser becomes furniture. The test is whether the adviser changed a decision in the last two quarters. An adviser who only ever agrees has stopped earning the seat.

Choosing between them

Three questions separate the models cleanly. First: is the question bounded? If the board can state what "done" looks like, it is a project. Second: does value depend on continuity? If the adviser needs to watch decisions form over quarters, it is a rhythm. Third: is the need really for availability rather than output? If the honest answer is "we want someone to call", it is a retainer.

Misfit shows up early and predictably. A project model applied to a continuity need produces a sequence of one-off reports with no memory between them. A retainer applied to a bounded question produces a slow, expensive project. A rhythm applied where independence is needed produces comfort, not counsel — the adviser in the room is a participant, and participants are the one thing a board cannot pay someone else to be.

What we tell our own clients

Start smaller than instinct suggests. A two-week structured second opinion is the cheapest way to test whether a firm's judgement is worth a larger mandate — and the fee should be credited against that mandate if it follows. Insist on a named individual rather than a firm brand: the model only works if the person in the room is the person whose judgement you bought. And ask each candidate adviser which model they would decline for your situation; the willingness to rule something out is the single best predictor of candour later.

Finally, price the landing strip into the project from day one. The most expensive sentence in advisory is "we'll work out the implementation plan once the report is done" — because by then the momentum, the access and the adviser's context have all been switched off.

Pricing the models

Projects should be priced as scope, not time. A day rate quoted against a bounded question is an incentive to let the question breathe; a fixed fee against a written problem statement is an incentive to solve it. We quote projects in two numbers: the fee for each movement of the mandate, and the cost of the optional landing strip afterwards. Boards should be suspicious of any firm unwilling to split those figures — bundling them is how landing strips get padded into every proposal whether needed or not.

Retainers are priced on expected access: how many sessions a month, what response time between them, who besides the named adviser may be drawn on. The honest pricing unit is the hour-band, not the hour — £4,000 a month for up to three sessions and ad-hoc counsel between them, for example — because metering every quarter-hour turns a trust relationship into a timesheet dispute. Rhythms are priced annually with a mid-year checkpoint, since the value compounds and the mid-year review is where scope should flex if it flexes at all.

Across all three, the same disclosure rule applies: every pound the adviser receives from anyone else in connection with your work must be shown on the invoice or declined. The day that rule feels negotiable is the day the advice stops being yours.

Governing the engagement itself

Whichever model a board chooses, the engagement deserves its own governance, proportionate to its size. Three mechanisms do most of the work. A one-page engagement letter that names the question, the person, the fee and the model — no appendices of standard terms that quietly reverse the headline promises. A written halfway review on anything longer than a quarter: is the original question still the right one, and is the named person still the right one. And an exit that costs nothing: our letters provide for termination at a month's notice with fees accrued only to date, because an adviser who is expensive to dismiss has an incentive no letter of principle can fix.

The strangest thing about engagement governance is how rarely it is asked for. Boards that would never approve a capital project without milestones approve advisory spend in the tens of thousands on the strength of a proposal document and a handshake. The remedy costs one page and one diary entry per year, and it changes behaviour on both sides of the table.

A closing note on incentives

Whatever the model, ask how the adviser is paid besides your fee. Commissions on software, referral fees from search firms, success fees aligned to deal completion — each bends the advice, sometimes imperceptibly, always in a direction you did not choose. Our own rule is absolute: no commissions, no referral fees, no success fees on transactions. It costs us revenue and earns us the ability to say what we actually think, which is the product.