MRR

A single MRR figure tells you almost nothing about the business that produced it. The same total can come from a healthy agency compounding its existing relationships or from one replacing losses as fast as it makes them, and those require completely different decisions.

The four components separate them. New is revenue from clients who were not there last month. Expansion is existing clients paying more. Contraction is existing clients paying less. Churned is revenue gone entirely.

Read together they answer the only question that matters: is the base growing or is the sales team outrunning a leak. An agency with strong new business and heavy churn is buying the same revenue repeatedly, and paying acquisition costs each time.

It also changes where effort goes. If contraction and churn are the problem, hiring another salesperson makes the numbers look better for two quarters and makes the business worse.

Winning revenue from an existing client costs a fraction of winning it from a new one. There is no pitch, no procurement, no onboarding, no learning the business, and the trust required has already been earned.

It is also the most reliably neglected, for a structural reason: nobody owns it. New business belongs to whoever sells, delivery belongs to the account team, and the question of what else this client needs sits between the two and therefore with nobody.

The answer is usually not a new service. It is a service the agency already offers and the client did not know about, or knew about and assumed was somebody else remit. Agencies consistently overestimate how clearly their clients understand what they do.

The QBR is the natural place for this, because the conversation is already forward-looking and the person who can authorise it is in the room. What the next quarter needs, said out loud, produces expansion revenue more reliably than any campaign aimed at strangers.

An agency with forty per cent of its revenue in one account does not have a client, it has a shareholder with no obligations. That account determines hiring, cashflow, and what the agency is willing to say in a meeting.

The distortion is behavioural before it is financial. Scope creep on a dominant account is absorbed rather than raised, because the cost of the conversation feels larger than the cost of the work. Bad news is softened. Pricing is never revisited. Each of those is rational in isolation and corrosive together.

There is no threshold that makes it safe, but the useful test is a question rather than a number: if this client left at the end of the quarter, what would have to happen. If the honest answer involves redundancies, concentration is a live risk rather than a theoretical one.

Reporting it alongside MRR is what keeps it visible. A total that grew because one client doubled is a total that became more fragile, and the headline says the opposite.

MRR is only useful if it is genuinely predictable, which means being strict about what goes into it. Retainers with a rolling term count. Projects do not, however regularly they recur, because a client who commissions a project every quarter can stop without anybody breaching anything.

Pass-through media spend must be excluded. An agency managing a hundred thousand a month in ad budget for a five thousand fee has five thousand of MRR, and counting the rest produces a number that describes the client budget rather than the business. It also collapses the moment the client pauses spend, which is the point at which an accurate number would have been most useful.

Variable fees need a judgement. A percentage-of-spend arrangement is recurring in form and not in amount, and the honest handling is to count a conservative baseline rather than last month figure.

The test worth applying: if this revenue disappeared next month, would anybody have had to make a decision to end it. If it can simply stop, it is not recurring, and including it makes the one number the agency plans hiring against less reliable than no number at all.