An Agency That Productized Into Software

What happened

An agency sold custom campaign work and grew, but every new dollar needed new hours, so margins hit a ceiling. The team noticed it was rebuilding the same reporting workflow for client after client and built an internal tool to stop repeating itself. They sold that tool to other agencies, and it became their first recurring revenue. By year seven software earned more than services, and the agency work was kept only because it kept showing them what to build next.

The case, start to finish

The workflow they had already been paid to build over and over was the market telling them, slowly and expensively, what to sell.

A profitable business with a ceiling

An anonymized composite, built on an archetype that has been documented publicly many times over. The most famous public version is 37signals building Basecamp out of the frustrations of its own client-project work. The agency described here is not that company, or any other specific one.

For its first three years it did what agencies do. It sold custom campaign work, got good at it, and grew. Demand was never the problem. The shape of the growth was: every additional dollar of revenue required an additional block of somebody's hours. Headcount tracked revenue almost exactly, and gross margin stalled where services margins stall, somewhere in the 30 to 50% band.

That is not a failing business. It is a business whose value is capped by its founders' willingness to keep showing up, and whose sale price, if it ever sold, would be a modest multiple of that continued effort.

The repeated build

In year four somebody noticed that the team kept rebuilding the same client reporting workflow. Each build had been paid for. Each had been slightly customized and substantially identical. Without meaning to, the agency had been running paid research: client after client had independently funded the discovery that this particular workflow was worth money to somebody.

The tempting move at that moment is to keep selling hours, because hours sell today. The second tempting move is to raise rates and specialize, which genuinely improves a services business but optimizes the ceiling rather than escaping it. The third is to build the thing, and the reason most agencies that attempt this fail is not the building. It is the middle.

The trough

Productizing cost roughly two years of agency profit. The two businesses competed for the same people. The developers who could build the product were the developers billing clients, and every hour moved across was an hour not invoiced. The transition was funded out of retainers rather than outside capital, which kept ownership intact and made the pressure entirely internal, which is to say entirely survivable and entirely miserable.

The first attempt made the classic mistake and was scoped as a platform: broad enough to serve everyone, specific enough to serve nobody. It only sold once it had been narrowed back to the single reporting workflow clients had already paid for repeatedly. The market had said what it wanted more than once, and the temptation had been to improve on the answer.

First recurring revenue landed in year five. By year seven software revenue had passed service revenue, and the agency work was kept deliberately, not for its income but as the channel that kept showing the team what to build next.

Why the economics change, not just the revenue

This is worth stating precisely, because "we launched a product" badly understates it. Services carry a cost that rises with every customer, because every customer consumes hours. Software carries a development cost paid once and a marginal cost of serving the next customer that barely moves. That is why software gross margins sit above 80% and services margins do not. The revenue did not merely grow. It was decoupled from the number of hours anyone worked.

For a reader running a services business, the useful version of this is not "go build software." It is narrower and more honest: notice what you keep rebuilding. The thing clients have paid you for repeatedly is the only product idea you have that already carries evidence, and finding it is not the hard part. The hard part is surviving the stretch where you are funding it out of work you are no longer fully doing.

Timeline

• Years 1–3 Agency sells custom campaign work; revenue grows but every dollar requires new hours, so margins cap out.
• Year 4 Team notices it rebuilds the same reporting workflow for every client; builds an internal tool.
• Year 5 The internal tool becomes a paid product for other agencies; first recurring revenue.
• Year 7 Software revenue passes service revenue; the agency work is kept only as a product-research channel.

You're in the owner's chair

Year 4. Your agency is profitable but every dollar still costs hours, and you notice the team rebuilding the same client reporting workflow for the fourth time. What do you do?

• Raise your rates and specialize harder
• Hire more people — sell more of the hours that already sell
• Productize: build the tool once, sell it to other agencies

Business model

Start as services (sell hours, learn the problem deeply, get paid to discover what everyone needs), then productize the repeated part into software sold many times with near-zero marginal cost.

Revenue model

Phase 1: project fees and retainers, high touch and linear with headcount. Phase 2: monthly subscriptions, recurring, decoupled from hours, and valued far more highly per dollar because it compounds.

Cost structure

The agency's cost was payroll scaling with revenue. The software's cost is development and support that barely rises with each new customer, and that gross-margin shift (services ~30–50% → software ~80%+) is the entire point of the transition.

Strategic challenge

The transition trough: building the product ate agency profits for two years, and the two businesses compete for the same people's attention. Most agencies that attempt this stall here.

Key decision

Productize the workflow the agency had already rebuilt a dozen times (proven demand) rather than inventing a product from scratch. Fund it from retainers instead of outside capital.

What worked

Client work was a paid R&D lab: the product solved a problem the team had watched real customers struggle with for years, so early sales came from the agency's own network.

What failed

The first product attempt was too broad ("a platform"); it only sold once it was narrowed to the single reporting workflow clients had actually paid for repeatedly.

Risk factors

The transition trough draining cash; splitting focus between services and product; underpricing the software out of services habit; clients expecting custom work forever.

Lesson summary

Services earn while you learn; software compounds what you learned. The safest product is the one your service business has already been paid to build many times, and the hard part is surviving the middle.

Key data

• 3 Years selling custom hours
• Year 5 First recurring revenue
• Year 7 Software passes service revenue
• Hours-for-fees → build once, sell repeatedly Economic shift

Sources & basis

The business in this story is a stand-in, not a company you can look up. This case is an illustrative composite: the operator, the people and most of the dollar figures represent a pattern rather than reporting one firm's history. What the list below cites is the other half, the documented industry data and public reporting the composite was assembled from, including any real company whose published figures the case draws on by name. The mechanism and the arithmetic are real even where the business is not.

1. Documented archetype (37signals/Basecamp origin, publicly recounted)
2. Composite pattern: see the Business Models category (agency-to-software)