AI for Agencies: The 3 Billable-Hour Leaks
Firms that sell time leak money in three specific places: non-billable drift, silent scope creep, and proposals rebuilt from scratch. Where each one hides and the fix.
Arrête de tout configurer. Place à la construction.
Des templates SaaS avec orchestration IA.
Problem: Your firm is busy. Everyone is booked. And yet the profit at the bottom is thinner than it should be, and nobody can point to exactly where it went.
Quick Win: If you sell time instead of a product, you leak money in three specific places a software company never touches. Non-billable drift (paid hours that never reach an invoice), silent scope creep (work you deliver but never charge for), and proposals rebuilt from scratch every time a deal comes in. The first one is measurable today: the average firm now bills only 66.4% of its paid hours, the lowest figure SPI Research has recorded in the history of its benchmark, against a healthy range of 74 to 84% (Certinia analysis of SPI Research 2026, Saibon Group). Fix these three before you hire anyone else.
Want this inside your company?
Tell us the outcome you need, and we'll show you what we can build.
Why billable-hour firms leak differently than product companies
A software company sells the same thing a thousand times. Build it once, sell it while you sleep. Its costs and its revenue barely touch after launch.
A firm that sells time is the opposite. Every dollar of revenue is tied to an hour someone worked. Billable utilization, the share of your team's paid hours that actually get charged to a client, is the number that decides whether you make money at all. When it slips, there is no product margin underneath to catch you.
That number is falling across the whole industry. SPI Research, which has run the Professional Services Maturity Benchmark for eighteen years, found billable utilization dropped to 66.4% in 2025, the lowest point in its entire surveying history and more than 8 points below the 75% level that top-performing firms hold (Certinia). A year earlier it sat at 68.9%, already the worst since 2019 (Saibon Group). The direction is the story: down, then further down.
Here is why that matters in cash. Every single percentage point of utilization you recover is worth roughly 2,500 to 4,000 euros per person per year, depending on your billing rate (Saibon Group). A 40-person firm sitting 5 points under where it should be is leaving well over half a million a year on the table, without losing a single client. The people are already on payroll. The hours are already paid for. They just are not reaching an invoice.
Product companies obsess over churn and conversion. Time-sellers should obsess over these three leaks, because they are where a busy firm quietly turns into an unprofitable one.
Leak 1: Non-billable drift and the utilization math
Non-billable drift is the slow migration of paid hours away from client work and into everything else: internal meetings, status updates, chasing information, rework, and the quiet habit of your most expensive people doing your least expensive work.
That last one is the hidden margin killer, and it does not show up on any dashboard. A senior person billing at a high rate spends the afternoon formatting a deck, cleaning up a junior's draft, or re-entering data. The hour still gets logged as worked. It just gets logged at the wrong price, or as non-billable, or not at all. Multiply that across a quarter and you get exactly the industry pattern: everyone is busy, utilization is under 70%, and operating profit is compressing. SPI tied the utilization slide directly to a drop in firm profitability in the same period.
The math is unforgiving because it compounds. Utilization below roughly 74% pushes revenue per person under the break-even line for most firm cost structures, and margins compress rapidly from there (Saibon Group). You do not feel it as a crisis. You feel it as a good year that somehow did not translate to the bank.
Adding people makes it worse before it makes it better. A new hire arrives at low utilization by definition, and if the underlying drift is unfixed, you have just bought more of the same paperwork at a higher payroll. This is the same trap salespeople fall into when they only sell 40% of the week: the answer is rarely more headcount, it is recovering the hours you already pay for.
Leak 2: Scope creep with no early warning
Scope creep is the work you deliver but never charge for. One extra revision. A "quick" favor. A meeting nobody agreed to in the original contract. Each one is small enough to wave through, which is exactly why they never get logged and never get billed.
The numbers are brutal once you add them up. The Project Management Institute finds that roughly 52% of projects experience scope creep, and of the projects that creep, about 85% run over budget, by an average of 27% (PMI, via Digital Applied). On the agency side it is more direct: in Ignition's 2025 survey of 273 agency managers, 57% said they lose 1,000 to 5,000 dollars every month to unbilled projects and tasks, another 30% lose more than 5,000 dollars a month, and only 1% manage to bill for all of their out-of-scope work (Ignition 2025, via Digital Applied).
The reason scope creep is so hard to stop is timing. By the time you notice it, the work is done and the relationship makes it awkward to send a bill for something you already delivered for free. The write-off is baked in before anyone looks at the numbers.
The fix is not tighter contracts, which everyone already has and nobody reads mid-project. The fix is early warning: a system that watches the gap between what was scoped and what is actually being delivered, and flags the drift while the project is still running. That turns an end-of-project loss into a same-week conversation while the client still remembers asking for the extra work. The decision about whether to bill for it stays human. The detection does not need to be.
Leak 3: Proposals and decks rebuilt from scratch
Every proposal, quote, and pitch deck your firm sends contains maybe 20% that is new. The case studies, the credentials, the methodology, the standard pricing, the team bios, that part already exists, somewhere, in a proposal you sent three months ago. Yet most firms rebuild the whole thing by hand every time.
The cost is enormous and almost invisible. Sales teams spend 20 to 30% of their time writing proposals and responding to formal bid requests, and by one breakdown 60 to 80% of that time is pure search overhead, hunting for the current pricing, the latest case study, the approved standard wording, not writing or thinking (Mojar AI). That is a quarter of your revenue-generating capacity spent reassembling documents that already existed.
This is the cleanest leak to fix because it is the most repetitive. Recurring documents built to a known standard, from your own approved content, are exactly the kind of work that should be produced automatically instead of rebuilt by hand. The salesperson's judgment goes into the 20% that is genuinely bespoke: the specific client, the specific problem, the price. The other 80% assembles itself.
What we install for each leak
The three leaks are different problems, so they get three different fixes. Here is the map:
| Leak | What it looks like | What it costs | What we install |
|---|---|---|---|
| Non-billable drift | Busy team, utilization under 70%, senior people doing junior work | 2,500 to 4,000 euros per person for every point of utilization lost | A ranked view of where paid hours actually go, so the wrong-priced work is visible and re-routable |
| Silent scope creep | Extra revisions and favors that never reach an invoice | 1,000 to 5,000+ dollars a month for most agencies | Early-warning detection of the gap between scoped and delivered, flagged while the work is live |
| Proposals from scratch | Every pitch rebuilt by hand, salespeople hunting for existing content | 20 to 30% of selling time, most of it search overhead | Recurring documents assembled automatically from your approved content, to a consistent standard |
None of this is a product you download. Each one is shaped to your firm, using your own time data, your own definition of "in scope," and your own approved content. That is the point of a bottleneck diagnosis: find which of the three leaks is costing you the most first, then fix that one before touching the others. Spreading effort evenly across all three is how firms stay exactly as unprofitable as they started, which is the same mistake the theory of constraints, the idea that one bottleneck limits the whole business, warns against: only fixing the biggest bottleneck moves the number.
When this does not apply
Honesty about where this breaks matters more than a clean pitch.
Very small shops. If you are three people who all know exactly what everyone is working on, you already have the visibility a system would give you. The overhead of installing one is not worth it until the firm is big enough that no single person can hold the whole picture in their head, usually somewhere north of 10 to 15 billable people.
Firms that already run tight. If your utilization is already in the mid-70s and your scope creep is consistently billed, you are not leaking, you are optimizing. The gains here are for firms losing money they cannot see, not firms already near the top of the benchmark.
Pure-product companies. If you sell software, a course, or anything you build once and sell many times, none of this is your leak. Your money leaks through churn, conversion, and pricing, not utilization and scope. This page is specifically for businesses where every dollar of revenue is tied to an hour someone worked.
Anywhere the fix is a rule, not a tool. If deals stall because one partner has to approve every quote, no automation fixes that. That is a decision about who holds authority, and it costs nothing to change and nothing to buy. Naming that honestly is part of a real diagnosis, and it is often the first thing a revenue-leak audit surfaces.
Frequently Asked Questions
How do I know which of the three leaks is worst for my firm?
Start with the one you can measure fastest. Pull your billable utilization number: if it is under 74%, non-billable drift is almost certainly your biggest leak, and every point back is worth 2,500 to 4,000 euros per person per year (Saibon Group). If utilization is healthy but projects keep going over, scope creep is your leak. If your salespeople are drowning in proposal work, that is leak three. Most firms have all three, but one is usually two or three times larger than the others, and that is the one to fix first.
Isn't low utilization just a sign of a slow quarter?
Sometimes, but the industry-wide slide says otherwise. Utilization has fallen for several years running to a record low of 66.4% in 2025, across hundreds of firms in every market (Certinia). A slow quarter is noise. A multi-year decline across the whole sector is a structural problem: too many paid hours going to internal work, rework, and senior people doing junior tasks. That is a leak you can fix, not weather you have to wait out.
Will automating proposals make our pitches feel generic?
Only if you automate the wrong 80%. The repetitive part, credentials, methodology, standard pricing, past case studies, is already generic; you just rebuild it by hand each time. Automating that frees your team to spend more time on the 20% that actually wins the deal: the specific client, their specific problem, and the price. Done right, your pitches get more tailored, not less, because nobody is burning their afternoon hunting for a case study that already exists.
If your firm is busy and still not making the margin it should, the money is leaking through one of these three, and you probably cannot see which one from the inside. We find the largest leak first, using your own data, then install the fix that produces a result your team uses the next morning. See what we build for companies →
Arrête de tout configurer. Place à la construction.
Des templates SaaS avec orchestration IA.