Client Work Books Revenue This Month. A Catalogue Books It for Years. They Can't Share a Team.

Custom game builds and a licensable catalogue are two businesses with opposite cash cycles, and running both from one team quietly starves the catalogue.

Almost every company in licensed games arrives at the same shape eventually. There is a catalogue you licence to portals, operators and brands, and there is client work — a branded build, a reskin, an integration, a custom event game — that someone pays for up front. On a P&L both lines look like revenue. Operationally they are not the same business, and the one that pays this month always wins the argument about who works on what.

That is not a discipline problem. It is arithmetic. Client work has an invoice attached to a named person who will chase it. Catalogue work has no invoice attached to anything until much later, and the person who would chase it is you.

💸 One Line Is Downstream of Someone Else's Greenlight

A custom build converts your capacity into cash at a known rate on a known date. That is genuinely valuable, and anyone who tells you services revenue is beneath a product company has never had to make payroll from a catalogue that is still six months from its first renewal.

What services revenue does not do is compound. Every quarter starts at zero. Worse, the size of that quarter is set by decisions taken inside your clients — portfolio reviews, budget freezes, a publisher deciding to consolidate three campaigns into one. You do not get a vote, and you usually do not get much notice.

The catalogue behaves inversely. It is slow to start and expensive before it earns anything, but a title that is licensed once can be licensed again to a different buyer in a different territory on a different platform without touching the build. The marginal cost of the eleventh licence on a finished game is a contract and a zip file. The marginal cost of the eleventh custom build is eleven times the first one.

So the honest framing is not product-good, services-bad. It is: one line buys you runway, the other buys you the option to stop selling runway. Both are worth having. They just cannot be resourced out of the same pool of people without one of them losing.

📉 The Biggest Services Business in Games Runs on Mid-Teens Margins

If scale fixed the services model, Keywords Studios would be the proof. It is the largest pure-play services company in the industry — art, QA, localisation, co-development, audio — with clients across most of the console and mobile market.

Its first-half 2024 numbers, as reported by PocketGamer.biz in 2024, are worth reading slowly:

• Revenue of $440.4 million, up 6.6% year on year
• Adjusted operating profit down 9.6% to $57.4 million
• Adjusted EBITDA margin of 17.7%, down from 20.2%
• Organic growth of –2%
• A net loss of $18.7 million

Revenue up, profit down, organic growth negative. The company attributed the shortfall to clients recalibrating their operations and game portfolios — which is the polite version of "our customers changed their minds and our revenue changed with them." Keywords was taken private by EQT in 2024 and no longer reports publicly.

Read that as a warning about Keywords and you have missed it. Read it as the ceiling on the model and it is much more useful. At the largest scale anyone has achieved in games services, with the widest client base and the most diversified service lines available, growth still tracked client budgets and margin still sat in the high teens. A five-person studio doing branded builds is running the same business with none of the diversification.

⏱️ Utilisation Is the Metric That Eats Your Catalogue

Here is the mechanism, and it is almost invisible from the inside, because it hides in a number every services business is told to maximise.

Agency benchmarking from Parakeeto puts target utilisation for delivery staff at 75–90% on a weekly basis and 65–80% annually, with the whole business — including non-billable roles — landing at 50–60% or more across a year. Parakeeto also notes that most agencies achieve around a 10–15% net margin, against a recommended target closer to 25%.

Now ask where catalogue work sits in that model. It is not billable. It does not appear on a client project. It goes into the same bucket as admin, internal tooling and sales support — the bucket you are structurally rewarded for shrinking. A studio hitting 80% delivery utilisation is, by definition, spending a fifth of its delivery capacity on everything that is not a client project, and catalogue development is competing inside that fifth against onboarding, estimating, and the work of winning the next contract.

This is why the roadmap slips without anyone deciding to slip it. Nobody cancels the catalogue. It just never wins a week.

The catalogue tasks that get deferred are the ones with external deadlines

The particular cruelty is which work gets postponed. Client work has deadlines set by clients, so it gets defended. Catalogue work has deadlines set by browser vendors, store policy teams and platform holders — people who will not email you, will not accept a slip, and will simply stop your titles working.

• Runtime and SDK upgrades across a catalogue of hundreds of builds, where the work scales with title count and cannot be batched into a sprint at the end
• Store policy compliance — privacy declarations, age ratings, target API levels — which have hard dates attached and apply to every title you have shipped
• Asset and load-time budgets, which drift upward one title at a time until your builds fail somebody's size cap
• Localisation and metadata, the cheapest work in the building and the first thing dropped when a client build runs hot

Each item is small. Each is easy to defer once. Deferred four times across two years, they turn a licensable catalogue into a maintenance liability that a buyer will discount you for.

🧮 Your Client Roster Sits Downstream of an Industry That Is Still Cutting

The 2026 State of the Game Industry report, published by GDC in January 2026 from more than 2,300 industry professionals, found that 28% of respondents had been laid off in the previous two years, rising to 33% in the United States. Half said their current or most recent employer had conducted layoffs in the past twelve months. At AAA studios that figure was 67%; at indie studios, 33%.

Be careful with what that does and does not measure. It is a survey of individuals who work in games, not a census of studios, and it skews toward people who attend or follow GDC. It does not tell you how much external development budget exists in 2026.

It does tell you something about who is on the other end of your pipeline. Outsourcing and branded-game budgets are approved by marketing leads, publishing producers and portfolio managers — the layer that gets thinned in exactly these rounds. When a client contact disappears, the project does not usually transfer. It gets re-scoped by whoever inherits it, and re-scoping is rarely upward.

A catalogue does not care who got laid off at your biggest client. That is not a small thing when half the industry's employers have cut in the last year.

⚖️ The Valuation Argument Is Real, but Weaker Than It Was in 2021

The standard pitch for building a product line is that recurring revenue is worth more at exit. It is true, and it is less true than the version you will hear from someone selling you a subscription model.

First Page Sage's service company valuation report, last updated January 2025 and covering deals from Q3 2022 to Q1 2025, records marketing agency EBITDA multiples falling from a range of 8–11.5x in 2021 to 4.5–7x in 2024. Consulting businesses in the same dataset land between 4.3x and 8.4x EBITDA depending on earnings level.

On the recurring side, Aventis Advisors reported a median EV/Revenue multiple of 3.4x for public SaaS companies as of March 2026, against a median of 4.5x revenue across private M&A deals in its 2015–2026 dataset with a top quartile at 8.1x. Aventis also found the SaaS premium over non-recurring software models — roughly 21% in 2024 — had narrowed considerably by 2026.

Those are different metrics on different asset classes and you cannot subtract one from the other. What the two datasets agree on is direction: services multiples compressed hard after 2021, and the recurring-revenue premium that was supposed to be the reward compressed too. If your entire case for building a catalogue rests on a resale multiple you read in 2021, rebuild the case.

The better argument was never the multiple. It is that a catalogue is the only asset in the business you control the roadmap for.

🔁 Running Both Without Starving One

The fix is not a decision. It is a constraint, applied before the quarter starts.

1. Ring-fence capacity as a percentage, not as intent. Twenty per cent of delivery capacity, named people, in the same planning tool as client work. "We'll do catalogue work between projects" is how it dies, because there is no between.
2. Give catalogue work a fake client. Ticket it, date it, estimate it, review it in the same weekly meeting. Work without a requester loses to work with one, every time.
3. Price reuse into the client quote. If a branded build reuses your framework, your ad mediation layer and your leaderboard service, that reuse is why you can quote below a studio starting from scratch. Some of that margin belongs to the catalogue that produced it.
4. Negotiate the IP carve-out in the contract, not afterwards. Custom game contracts are usually work-for-hire: the client owns the deliverable. That is fine, and it is normal. What is not fine is signing away the engine, the tooling and the reusable components underneath it. Carve out your framework, your libraries and anything generic in writing, before the build starts.
5. Count catalogue revenue separately from the first month. Blended revenue hides the fact that one line grew and the other did not. Two numbers, one board slide, every month.

There is also a legitimate third path that gets treated as a failure: licence someone else's catalogue instead of building your own. If your business is genuinely a services business — you win on client relationships, creative and delivery — then spending two years of engineering on a catalogue you will under-maintain is not diversification, it is a hobby with a payroll. Buying access to finished titles under a clear licence and pricing arrangement is the faster route to a recurring line, and it does not compete with your client work for engineers.

🚫 Five Ways Companies Get This Wrong

• Funding the catalogue out of slack. There is no slack in a business at 75% utilisation. Slack is the thing you sold.
• Treating a delivered client build as a catalogue title. If the client owns the IP, the brand assets and the exclusivity, it is not in your catalogue. It is a case study. Check what your last five contracts actually granted before counting them.
• Hiring for the catalogue with client revenue that has not been signed. Services revenue is the least forecastable revenue you have. Hiring against a verbal yes is how a good quarter becomes a redundancy round.
• Building titles nobody asked for. The catalogue is not a side project, it is a product line, and it needs the same buyer conversations client work gets. Talk to portal operators and publishers before the third title, not after the thirtieth.
• Letting maintenance debt accumulate silently. Every deferred SDK upgrade is a bill with interest. At licence renewal, or at diligence, it arrives all at once.

🎮 Where a Licensed Catalogue Fits

Forestry Games has run both lines since 2017 — in-house HTML5 development and branded work on one side, a licensable catalogue of 1,049 titles on the other, distributed to portals, operators, brands and agencies. The catalogue exists precisely so that operators who need a recurring games line do not have to fund a studio to get one. If that is the position you are in, the catalogue and the source code options are the two shapes worth comparing: licensed access when you want titles live quickly, source code when you need to own and modify the build.

🧭 What to Do Before Your Next Client Quote

Pull the last four quarters and split revenue into two columns — work billed to a named client, and revenue that arrived from titles you already owned. Do not blend them. Then answer three questions.

What share of engineering hours went into column two? If it is under 10%, your catalogue is not a business line, it is a marketing asset, and you should either resource it properly or stop describing it as a product. What would happen to column one if your two largest clients cut budgets in the same quarter — as half the industry's employers did last year? And what would column two be worth in three years if it received the same 20% of capacity every quarter without exception?

The answer to the third question is usually large enough to change what you do on Monday. It never gets calculated, because calculating it is not billable either.