Retainers went up. Margin didn't.

Because you added revenue without repricing it. Retainer fees stayed where they were set while salaries climbed, delivered hours drifted past scoped hours, and rework and founder time got absorbed instead of billed. Promethean Research put average agency net margin at 13 percent in 2022, down from 17 percent the year before, on 15 percent revenue growth.^[4]

That is the whole mechanism. More revenue moving through a cost base that grew faster than the price did. What follows is where each dollar leaves, and the three numbers you can pull this week that show you exactly which accounts are doing it.

What's actually going on

Nobody chose this. It arrived one renewal at a time, in decisions that were each defensible on the day they were made.

Revenue grew. The cost base grew faster.

Promethean's survey of digital agencies found net margins fell to an average of 13 percent in 2022, down from 17 percent in 2021, and named the causes plainly: salary growth, poor utilization rates, and high turnover.^[4] Salary growth for key functional roles averaged around 20 percent between 2019 and 2022, and in that same year only 38 percent of shops raised prices.^[4]

Read that as roughly six in ten agencies absorbing a 20 percent cost increase at the old price. That is not a delivery failure or a talent failure. It is arithmetic. And adding headcount does not correct it, it scales it, because every new person enters at the new salary level and gets sold at the old rate.

Utilization is a ratio, not a paycheck

Utilization is the metric most agencies watch, and it is the one most likely to reassure you while you lose money. It measures billable hours against available hours. It says nothing about whether those billable hours were priced correctly.

Deltek's summary of the SPI Research Professional Services Maturity Benchmark put average billable utilization at 66.4 percent in 2025, the lowest in that survey's history, while the most mature firms held above 80 percent.^[1] Promethean puts healthy agency ranges at 70 to 90 percent for production staff and 60 to 80 percent for account management, and notes that partners and senior positions run much lower by design.^[2]

Here is the part that matters. A person can be 85 percent utilized on an account that loses money. Utilization tells you the seat is full. Margin per account tells you the seat is earning. Most agencies track the first and estimate the second, and the estimate is generous. Promethean's own benchmark for average project margin, 35 percent, is drawn from the 59 percent of surveyed agencies that actually track project margins at all.^[3] Four in ten do not measure it.

The price was set once and never revisited

Retainers are the default model. Promethean's surveys put retainer use at 7.5 out of 10 digital agencies, with the most common retainer under $5,000 a month and almost half under $10,000.^[3]The same research found 42 percent of agencies report average retainer tenure above two years, including 30 percent above three.^[3]

Put those two facts side by side. A large share of your revenue is sitting on prices agreed two or three years ago, against a cost base that moved every year since. Nobody decided to give a discount. The discount happened by standing still.

The move that usually makes it worse

You sign another retainer. Or, because payroll is already heavier than you planned, you hire a cheaper person to deliver the retainers you have.

Both feel like margin moves. Both usually cost you.

Another retainer adds revenue at the same unfixed price, which means it adds a proportional slice of the same unfixed margin, plus a new account's worth of onboarding, calls, and founder attention that nobody scoped. If the account you already have runs at 43 percent once everything is counted, the next one runs at 43 percent too. You have bought yourself more work at the same rate.

The cheaper hire is the worse of the two, and the math is not close.

Promethean's method for loading a salary is simple and worth copying: annual salary times 1.30 for employment taxes, benefits, and overhead, divided by 2,080 working hours a year.^[3] Run two people through it. A junior at $45,000 loads to about $28 an hour. A senior at $95,000 loads to about $59. The senior looks twice as expensive.

Now price the actual output instead of the seat. Say the senior produces the deliverable in 5 hours and it ships. The junior takes 12 hours and needs 2 hours of your review before it can go out. Your own time, loaded the same way against a $180,000 comp level, is about $113 an hour.

Senior path: 5 hours at $59 equals $297.
Junior path: 12 hours at $28 plus 2 hours at $113 equals $563.

The cheap hire cost 90 percent more to ship the same thing. And that assumes it ships on the first pass. Add one round of rework and the gap roughly doubles. You bought a lower hourly rate and paid for it with the most expensive hour in the building.

Gallup closes the loop on the other end. Replacing a single employee costs between one-half and two times that employee's annual salary, and Gallup calls that a conservative estimate.^[6] Hiring cheap and churning cheap is not a cost saving. It is a cost deferral with interest.

The break in an agency

Five places the money leaves. Most agencies have three of them running at the same time, and none of the three appear on the P and L as a line you could point at.

Scope creep is a pricing event that never gets priced

PMI's Pulse of the Profession found 52 percent of projects completed in the prior 12 months experienced scope creep, up from 43 percent five years earlier, and defines it precisely: uncontrolled expansion of scope without adjustments to time, cost, and resources.^[5] The same research put waste from poor project performance at 9.9 percent of every dollar invested.^[5]

In an agency that expansion has familiar names. The extra revision round. The quick landing page. The added reporting call. The strategy session that was never in the scope. Each one is small enough that raising price over it feels petty. Nine months later the retainer is delivering a different scope than the one it was priced against, and the price is the only variable that did not change.

Promethean names this as a standing limitation of the retainer model itself: clients enter retainers expecting every request fulfilled, and without detailed monthly time reporting you cannot see it happening.^[3]

Rework is delivered twice and billed once

Rework is the version of the deliverable that exists because the first one was not right. Wrong brief, wrong reviewer, wrong standard, wrong feedback loop. The cause does not change the accounting. Every rework hour is capacity you sold once and spent twice.

Deltek's summary of SPI's benchmark is useful in two places here. Firms lost an average of 4.5 percent of revenue to leakage in 2025 while the top performers held it below 3 percent, and projects that overrun by more than 10 percent damage project margin, client satisfaction, and future bookings, with the most mature firms holding overruns to single digits.^[1] Rework is how overrun actually happens inside an agency. It rarely shows up anywhere, because it is not tracked separately from the work it repeats.

Founder hours are real cost carried at zero

This is the one that hides best. Your hours on an account are not free. They are unpriced, which is a different thing. Promethean notes that utilization for partners and senior positions runs much lower than production staff, which is correct design, because senior time is supposed to go to the work only senior people can do.^[2]When it goes into delivery instead, two things happen at once. The account's true cost rises and nobody records it, and the highest leverage hour in the business gets spent on the lowest leverage work in the business.

Why the work keeps landing on your desk in the first place is a separate problem with a separate fix, covered in the piece on why delivery routes back to the founder. This article is only about what it costs once it does.

Scoped hours and delivered hours are two different numbers

Promethean's retainer formula is clean: Retainer Price equals Team Cost divided by one minus Desired Margin.^[3] It works. It only works on the hours you actually deliver.

Most agencies price against a scoped hour count, then never compare it to the delivered hour count. If you scoped 60 hours and delivered 84, your margin did not slip a little. Your effective hourly rate fell 29 percent, and it fell silently, because the invoice read the same both months. The client did not negotiate you down. The scope did.

Nobody knows margin per account

Ask yourself which account is most profitable. Not biggest. Most profitable. If the answer is a feeling rather than a number, that is the break, and it sits upstream of every other one on this list, because you cannot fix leakage you cannot locate.

The clients who want you personally on every call are usually the ones with the worst margin per account, for reasons covered in the piece on clients who only want the founder. That is worth knowing before the renewal conversation, not after it.

What to check in the next 7 days

Three numbers, one account, one spreadsheet. Pick the account you assume is your best one. It is usually the most instructive. Substitute your own salaries into every figure below.

1. Effective hourly rate per account.Take last month's retainer fee, excluding pass-through media and ad spend. Divide it by every hour actually delivered on that account: production, account management, calls, reporting, revisions, QA, and your own hours. Not the hours you scoped. The hours that happened.

On an $8,000 retainer scoped at 60 hours and delivered in 84, the priced rate was $8,000 divided by 60, or $133 an hour. The effective rate is $8,000 divided by 84, or $95 an hour. That is a 29 percent discount you never agreed to.

2. Rework hours.Go back through those 84 hours and mark every one spent producing something that had already been produced once. The second version of the deck. The re-cut. The report that got rerun because the first pull was wrong. Load each hour using Promethean's method, salary times 1.30 divided by 2,080.^[3]

In this example, 9 team hours at $47 plus 4 of your own hours at $113 comes to $875, or 11 percent of the retainer. Compare that against the 4.5 percent average revenue leakage in SPI's benchmark.^[1] Agencies running this for the first time usually land in double digits.

3. Founder hours per account. Log your own hours on that one account for a month and price them at your loaded rate, not at zero. Nine hours at $113 is $1,017 on an $8,000 retainer. That is 12.7 percent of the fee, and it appears on no report you currently produce.

Then put the three together. Team cost is 75 hours at $47, or $3,525. Founder cost is 9 hours at $113, or $1,017.

Margin the way most agencies report it, counting team hours only: $8,000 minus $3,525, divided by $8,000, equals 55.9 percent.
Margin as actually delivered: $8,000 minus $4,542, divided by $8,000, equals 43.2 percent.

Twelve and a half points of margin, gone into hours nobody charged for. And here is why it survives: against Promethean's stated 25 to 65 percent gross margin range for digital agencies, 43 percent still reads acceptable.^[3] The account looks fine at the gross line while the business runs at a 13 percent net.^[4] Nothing in your reporting flags it, because your reporting was built to watch utilization.

Now run the same three numbers across every account and rank them. Two things usually fall out. The biggest retainer is not the best account. And at least one account is running below your loaded cost, which means you are paying for the privilege of keeping it.

One sequencing note. Do not reprice anything until you have run all three numbers on at least half your book. Repricing off a single bad account produces a policy. Repricing off the ranked list produces a decision.

When a look, diagnostic, or embed is the next step

Most agency owners who actually run those three numbers do not need help with what comes next. If one account is obviously underwater and the rest are fine, go fix that account. Reprice it at renewal, cut the scope back to what was sold, or let it go. You do not need an outside opinion for that.

Outside help earns its place when the ranking comes back and the damage is spread evenly across the book. Even distribution means the break is not in an account, it is in how you price and scope all of them. That is an offer architecture problem, and no amount of delivery discipline downstream fixes a price that was set against the wrong number.

The first step is a free look. Send the agency site and last month's numbers on one account, and you get back the one thing worth fixing first. No proposal attached. Start it from the agencies page, which lays out the margin cycle this article sits inside.

If the constraint is still not obvious after the look, the diagnostic is one working conversation that names it and sequences the fix. And when the answer is that pricing, scope, and delivery all need rebuilding rather than advising on, that is what the embedded operator engagement is: offer and pricing structure, scope definitions with real hour bands, margin per account reported monthly with founder time priced in, and capacity planned against cash, installed rather than recommended.

One note on proof, since it matters. NOiC currently runs a live marketing agency engagement and that client is not public, so there are no agency-side numbers published here. The named results on this site are clinic work: kingdom went from zero to multi-million annual revenue in 12 months, Physio Plus TX tripled monthly revenue in 5 months, and Premier Hormone Health doubled revenue with retention held. Those are healthcare businesses, not agencies. They are here as evidence that the method has been run inside real operating companies with real books, not as a claim about agency results.

The last word belongs to the oldest finding in pricing research. Marn and Rosiello wrote in Harvard Business Review that getting the price right is the fastest and most effective way for a company to reach its maximum profit, and that the right price moves profit faster than growing volume does.^[7] Three decades later, the agency version of that sentence is still the one nobody wants to hear. You probably do not have a revenue problem. You have a price you never revisited and a cost you never counted.

Sources