Carving Off a Product From an Agency: The Internal Spin-Out Playbook

Every agency over a certain size eventually notices the same thing: a tool, a template, or an internal framework that clients keep paying for. The temptation is to call it a product and start selling it. The reality is that most of these attempts quietly die inside the services business that birthed them, starved of attention and confused about who owns what.

This is the playbook we use when a client asks us how to spin a product out of their agency — and the one we've used ourselves. It's not a fundraising guide. It's an operational guide for the first 12 months, when the product is still inside the mothership and could go either way.

Decide What You're Actually Spinning Out

Before you touch org charts, name the thing precisely. There are three patterns that get conflated and they each need different treatment.

• Productized service. A fixed-scope, fixed-price offering delivered by humans, with some tooling behind it. Still a services business. Revenue scales with headcount.
• Internal tool turned SaaS. Something your delivery team built to ship faster, now repackaged for external buyers. Real software, but the buyer profile may not match your agency's clients.
• New product, agency-funded. A genuinely new bet that uses agency cash flow as runway. Often the riskiest, because there's no built-in customer evidence.

The failure mode is treating a productized service like a SaaS company (raising, hiring a VP of Sales, burning cash) or treating a real product like a side project (no dedicated team, no roadmap, no pricing discipline).

A quick test

If you stopped doing all client work tomorrow, would the product still have a reason to exist next quarter? If the answer is no, it's a productized service. Run it accordingly — high margin, operational, no venture mindset.

Separate the Money Before You Separate the People

The single biggest reason internal spin-outs fail is shared P&L. The product looks unprofitable next to a 35% margin services line, so it gets defunded the first time a big client deal needs more engineers.

Fix this on day one with three accounting moves:

1. Give the product its own ledger. Separate cost centre, separate revenue line, separate gross margin calculation. Even if it's still inside the same legal entity, the numbers have to be visible.
2. Charge transfer pricing for shared people. If a services engineer spends 20% of their week on the product, the product cost centre pays the services cost centre for that time at an internal rate. This prevents the product from looking free, and it prevents the services team from resenting the "vanity project."
3. Commit a runway, not a budget. Decide upfront: the product gets, say, £400k or 18 months, whichever runs out first. After that, it either stands on its own commercial feet or it gets shut down. Open-ended funding kills urgency.

The product should feel like a portfolio company of the agency, not a department of it. Different rules, different metrics, different review cadence.

Pick a Team Structure That Doesn't Lie

The most common structure — "we'll have engineers split between client work and the product" — does not work for more than about six weeks. Client deadlines always win, because clients shout and products don't.

Here are the three structures we've seen actually function, ranked by how serious you are.

Option A: The protected pod (minimum viable)

Two or three engineers, one PM, and a designer, ringfenced from client work for a fixed period (we usually say 12 weeks minimum). They report to a product lead, not a delivery lead. Their utilisation is measured in shipped product outcomes, not billable hours.

This is the smallest unit that can ship anything real. Anything less is a hobby.

Option B: The skunkworks with a kill date

Same as the pod, but with a hard go/no-go review at the end of the runway. You agree the kill criteria upfront: number of paying customers, ARR threshold, retention numbers. If those numbers don't materialise, you shut it down and reabsorb the team. The kill date is the feature, not a bug — it forces honest decisions.

Option C: The legal carve-out

New entity, separate cap table, the agency holds majority equity, the product team gets meaningful options. This is the right structure once you have early revenue and a credible path to a venture-style outcome. Doing it earlier is usually premature and creates tax and IP headaches you don't need.

Get the IP and Contracts Right Early

This is the boring part that becomes a nightmare if you skip it. Three things, in order of how often they bite:

Code provenance. If your product was extracted from client projects, you need to know which client contracts assigned IP to the client and which left it with you. Run an audit before you ship anything externally. A simple manifest helps:

module: billing-engine
original_project: client-acme-2023
ip_clause: mutual_license # client owns deliverable, agency retains framework rights
rewrite_required: false
notes: Refactored from internal fork; no client-specific logic remains.

Employee assignment. Make sure every engineer who touched the product has a clean IP assignment in their contract. If you bring on contractors, use work-for-hire language. If you're in a jurisdiction where moral rights apply, get explicit waivers.

Customer contracts. Your agency MSA is not a SaaS terms-of-service. Don't try to retrofit one into the other. Write proper product terms — limited liability, data processing addendum, uptime commitments you can actually meet, and a clear termination clause.

Price Like a Product, Not Like a Project

Agency founders consistently underprice their first product because they're used to thinking in hours. A feature took two weeks to build, so it must be worth roughly two weeks of billable time. That's not how product pricing works.

A few principles that have held up for us and the teams we advise:

• Price against the customer's alternative, not your cost. If your product replaces a £4k/month contractor, you have pricing room up to about £1.5k/month before procurement gets interested.
• Charge monthly from day one, even for design partners. Free pilots become permanent. A £1 invoice is infinitely more useful than a £0 one because it proves the procurement path works.
• Avoid usage-based pricing until you have predictable usage. Seat-based or tier-based pricing is easier to forecast and easier for buyers to approve. You can layer usage components on later.
• Build in an annual discount of 10–20%. Annual contracts smooth your cash flow and reduce churn measurement noise.

If you're not sure where to start, look at how we think about pricing engagements on our services page — the same logic about anchoring to outcomes applies, just at a higher leverage point.

Define the Metrics That Trigger the Next Decision

The product should have three or four numbers that everyone — founder, product lead, agency MD — can recite without checking a dashboard. Anything more is theatre.

What we usually pick for the first 12 months:

1. Paying logos. Not signups, not trials. Companies who have paid an invoice.
2. Monthly recurring revenue. Tracked weekly during the early phase.
3. Net revenue retention at month six per cohort. Tells you whether the thing is actually sticky or just novel.
4. Time-to-value. How many days from sign-up to the customer doing the core action that justifies the subscription.

Review these monthly with the same seriousness as the agency P&L. If they're trending the wrong way at month nine of an 18-month runway, you have a decision to make, not a pep talk to give.

When to Pull the Plug — and How

The hardest part of an internal spin-out is killing it cleanly. Sunk-cost fallacy is brutal, especially when the product has your name on it and people you like are working on it.

Write the shutdown plan before you start. Ours is one page:

• Notice period for any paying customers (typically 90 days)
• Data export commitment and timeline
• Where the team gets reabsorbed (specific roles, specific projects)
• What happens to the codebase (archive, open source, sell)
• What gets carried back into the agency as reusable IP

The existence of this document does not make shutdown more likely. It makes the decision faster and less emotional when the data says it's time.

Where We'd Start

If you're running an agency and staring at something that might be a product, do these four things this quarter, in this order. Pick the category — productized service, internalised tool, or new bet. Separate the P&L with a transfer-pricing model. Commit a fixed runway with kill criteria written down. Hire or assign a product lead whose performance is measured on product outcomes, not utilisation.

Everything else — the carve-out, the cap table, the GTM motion — can wait until those four are in place. Most spin-outs die before they reach that point, not because the idea was wrong, but because the operating model never gave the product a fair chance to prove itself.