What an Agency Pays Per Client to Run Its Stack

Agencies price retainers against salaries. Someone works out what a strategist and a designer cost for the hours a client will take, adds a margin, and quotes. Tooling sits in a different mental box marked overheads, reviewed once a year, and treated as a fixed cost of being in business. It is not a fixed cost. It is a per-client cost wearing a monthly subscription, and it goes up every time you win work.

A fixed cost is one that does not move when the business gets busier. Rent is a fixed cost. Your software is not, and the reason is structural rather than a matter of which vendors you picked.

Win a client and you often open a sub-account, a workspace, or a project. Hire someone to service them and you add a seat on every per-seat tool at once. Neither of those is a decision anyone makes deliberately about software. They are consequences of doing the thing the agency exists to do, which is why they never get reviewed at the moment they happen.

6 people, 12 clients, 3 per-seat tools and 3 flat ones, priced at the medians across the 68 go-to-market tools we track. Entry tiers, so this is a floor rather than a forecast.

$633 a month in total, which is roughly $53 per client. On a small retainer that is a visible slice of the margin. On a large one it is noise. The point is not the number, it is that most agencies have never produced it at all, so it never enters the pricing conversation.

Notice what happens if the agency doubles its client count without hiring: the same $633 spreads to about $26 per client. Utilisation is the lever, not the vendor list. Every conversation about switching tools to save money is smaller than the conversation about how many clients the current tools are serving.

Headcount. Of the 68 tools we track, 25 are priced per seat. Hire one person and you do not add one line item, you add one per per-seat tool you run, all on the same day. At the median of $19 per seat, three per-seat tools means every hire costs about $57 a month in software before they have done any work.

Client count. The second axis is the one agencies share with almost nobody else. Sub-accounts, client workspaces, extra brand slots, additional connected inboxes and social profiles: many tools meter on exactly the thing that grows when the agency succeeds.

Every agency has a second, invisible stack: the licences bought on a client's behalf, on the agency card, because it was faster than waiting for their procurement.

These are worth auditing before anything else, for three reasons. They are rarely on anyone's renewal calendar. They frequently outlive the engagement, because nobody cancels a tool when a client leaves. And they are almost never billed back, because the amount felt too small to invoice at the time and then became awkward to raise.

The audit is unglamorous and usually pays for itself the first time anyone runs it: list every subscription, name the client it exists for, and check that client is still a client.

Not the most expensive line. The one that appears on both axes, which is where consolidation compounds instead of just saving a subscription.

First, anything priced per seat that only some people use. Per-seat tools with partial adoption are the purest waste in an agency stack: you are paying full price for a licence somebody logs into twice a month. This is usually the fastest real saving available.

Second, the tools that meter on clients. These are the ones that punish you for winning work. A flat-priced equivalent costs the same at twelve clients and at twenty, which turns a variable cost into an actual fixed one.

Third, the seams. Where a person moves data from one tool to another on client work, that time is already inside a priced retainer, so it comes straight out of margin. It never appears on a bill, which is exactly why it survives every cost review.

The general sequence, with the migration plan, is in the GTM stack consolidation playbook. What is specific to agencies is the second axis: you are not only consolidating to spend less, you are consolidating so that winning your next five clients does not cost you five more line items.

Once you have a per-client number, you have three honest options, and the only wrong one is the fourth.

Bill it through. Transparent, easy to defend, and it makes the client an ally when you want to remove a tool. It also makes your own stack visible to them, which some agencies dislike.

Price it in. Add the per-client figure to the retainer as part of the cost base. Simplest, and it survives client scrutiny because the number came from arithmetic.

Absorb it deliberately as a cost of the service level, having actually looked at it. Fine, as long as it was a decision.

The fourth option is absorbing it without knowing the number, which is where most agencies are. It is not a pricing strategy; it is a rounding error that grows on two axes while nobody is watching. For how the wider workflow fits together, the agency overview covers the delivery side.