Most agencies quote a project at a 35% margin and keep about 13% after tax. The twenty-two points in between rarely disappear into overhead alone. They leak out of delivery — through scope that grows while the fee stays flat, work that gets done twice, capacity you pay for and never bill, and time lost every time a project changes hands. Here is where each leak hides, and how to measure it before it decides your year.
TL;DR
The average digital agency earned a 13% after-tax net margin in 2025, while the average tracked project margin was 35% (Promethean Research, 2026). That gap is the whole game. Four delivery leaks eat it: Scope drift, Rework, The Bench, and The Handoff Tax. Name which one is biggest for you, measure it in hours, and you have found more margin than another new client would give you.
Where does the margin actually go between the quote and the bank?
It goes into delivery you didn't price. In 2025 the average digital agency ran a 13% after-tax net margin, down from a long-run average near 15% since 2015, on an average of $4.43M in revenue (Promethean Research, How Profitable Are Digital Agencies?, 2026, N=119). Among the 59% of agencies that actually tracked project margins, the average project came in at 35%. A project priced to keep a third can still net you a tenth. The difference is operational, and it is fixable — but only if you can see it.
I call the map of where it goes The Delivery Margin Leak Map. Four leaks, in the order they usually bite.
Leak 1 — Scope drift: the work grows, the fee doesn't
Scope drift is the most common and the least invoiced. PMI found that 52% of projects experienced scope creep — uncontrolled changes to what was agreed — up from 43% five years earlier (PMI, Scope Creep Is on the Rise, Pulse of the Profession). It never arrives as a demand. It arrives as a favour: one more revision, a page that was "basically implied," a stakeholder who "just needs a small thing." Each request is too small to bill without looking petty. Stacked across a quarter, they are the single largest unpriced cost most agencies carry.
The trade-off is real: a strict change-order habit protects margin and irritates clients who liked the old looseness. That is the cost of the fix, and it is worth paying.
Leak 2 — Rework: hours you already paid for, spent twice
Rework is margin you burn producing the same deliverable more than once — unclear specs, a quality miss, a junior build a senior has to redo. PMI's same research put total waste from poor project performance at 9.9 cents of every dollar invested. On a $4.43M agency, that is roughly $438,000 a year moving through the P&L with nothing to show for it. Rework rarely gets logged as rework; it hides inside "the project took longer than we thought." A team already at capacity absorbs a spike by doing the work fast and wrong, then doing it again.
Leak 3 — The Bench: capacity you pay for and don't bill
Delivery salaries run whether the work does or not. Parakeeto's benchmark for a well-run agency is a delivery margin of 50–60% or better, with anything below 40% signalling overstaffing or underpricing. Producers — designers, developers, copywriters — should land around 65–80% utilization on an annual basis (Parakeeto, 8 Vital Agency Metrics). The gap between what you pay for and what you bill is the bench, and it is worst in exactly the agencies with lumpy pipelines: you hire for the peak, then carry the salary through the trough.
This is the leak the hire-vs-rent decision turns on. A permanent hire fixes the peak and creates the bench. Rented capacity has no bench by design — you pay for the peak and nothing in the trough. That is the entire economic argument for it, trade-off included: in a busy month, rented capacity costs more than a salaried head would have.
Leak 4 — The Handoff Tax: time lost every time work changes hands
Every time a project moves between people, someone re-learns it. Onboarding a new hire, briefing a freelancer, context-switching a senior across five accounts in a day — none of it is billable and all of it is real. It is the quietest leak because it never appears as a line item. It shows up as smart, expensive people spending Tuesday getting back up to speed on a thing they built in March.
Here is the map in one liftable view:
Leak What it is Where it hides The number to pull
Scope drift Work grows, fee stays flat "We just did them a favour" Unbilled hours vs. SOW hours
Rework Same deliverable produced twice "It took longer than we thought" Revision hours per project
The Bench Paid capacity, unbilled Slow months, lumpy pipeline Producer utilization %
The Handoff Tax Re-learning time on every transfer Onboarding, context-switching Non-billable delivery hours
Why doesn't raising prices fix this?
Because most agencies aren't. Only 28% raised prices between 2024 and 2025 (Promethean Research, 2026), even as AI compressed production hours and clients started expecting the savings. Price discipline matters, but a higher rate on a leaking delivery process just leaks faster. The leaks are hours, not dollars — and you close them by pricing scope tightly, shipping right the first time, and matching capacity to the work instead of to the peak.
FAQ
What's a healthy delivery margin for an agency?
Around 50–60% or better on your P&L, per Parakeeto's benchmark. Below 40% usually means you are overstaffed, underpriced, or absorbing too much unbilled delivery. Delivery margin sits above net margin — it's what's left after delivery costs but before overhead and tax — so a thin delivery margin guarantees a thin net one.
Why is my project margin healthy but my net margin low?
Because project margin is quoted before the leaks and net margin is measured after them. The average agency quotes 35% and nets 13% (Promethean Research, 2026). Scope drift, rework, bench time, and handoff time all land between those two numbers, and none of them show up on the estimate.
Is scope creep really common enough to matter?
Yes. PMI found 52% of projects experienced it, up from 43% five years earlier. For an agency on fixed fees, it is pure margin erosion — the work grows and the invoice doesn't. The fix is a written out-of-scope list and a change-order habit, not heroics.
Does hiring fix a delivery bottleneck?
Sometimes, and it creates the bench when it doesn't. A permanent hire fixes a permanent peak. If your pipeline is lumpy, the hire is fully billable in busy months and a fixed cost in slow ones. That is the exact trade rented capacity is built to avoid.
How do I find out which leak is costing me most?
Pull your last three delivered projects and split the hours into billed, revised, and non-billable. The category with the biggest surprise is your primary leak. Most owners have never done this and are genuinely unsure which of the four is worst.
When does renting delivery capacity beat hiring?
When your pipeline is uneven enough that a salaried head would sit on the bench part of the year. You pay more per hour in busy months and nothing in slow ones. If your delivery is steady and predictable, a hire is usually cheaper. If it spikes, it usually isn't.
The one-page audit (≤7 checks)
Pull your last three projects: billed hours vs. quoted hours.
Log revision hours separately — that's your rework number.
Calculate producer utilization for last quarter (target 65–80%).
Count non-billable delivery hours: onboarding, handoffs, context-switching.
Compare your quoted project margin to your actual net margin.
Name the single biggest leak of the four.
Fix that one before touching anything else.
I'm Basir, co-founder of Opti Webopz. We're a white-label delivery team — WordPress, WooCommerce, SEO/AEO and AI automation, shipped under your agency's brand by a 25+ person team. We also take direct work in healthcare, real estate, and ecommerce. If you want a 20-minute look at your delivery capacity, the link is in my profile.