Keep Half of Revenue: The Software-to-Headcount Mix

The bank balance shows the warning before the P&L does: the service work that looked profitable is eating cash. Bootstrapped TripleDart passed $7 million in annual recurring revenue while its EBIT margin stood at 50%. The mix is moving repeatable work into software, pricing the remaining service at fully loaded cost plus margin, and adding people only when product revenue can cover the salary before the next planning cycle.

Unbilled hours are the entry fee

Most services firms lose the margin they think they have because they price the visible hours and forget the invisible ones. Project management, onboarding, reporting, revisions, and the quiet cost of waiting for approval all sit inside the invoice. If you do not charge for them, they become the reason the bank balance looks worse than the P&L.

Profit is a claim on future cash. Cash is what pays payroll, software, and the landlord. A bootstrapped agency that confuses the two will keep hiring until the burn feels like growth. The safer question is whether each dollar of revenue leaves enough cash to cover the next month without extra borrowing.

Repeatable work becomes the product

The first move is to find the service work that repeats across clients. Do the same intake, the same audit, the same reporting, or the same campaign setup for many accounts, and that work has a product form. Package it, automate parts of it, and sell it as a subscription or a fixed-scope product.

The ARR increase came from software, not from adding employees or raising capital. By the announcement, it had 120 people after four and a half years and was managing over $200 million in ad spend for more than 300 client companies. The revenue per employee is a useful check, but the margin mix is the real test.

The combination matters because it shows the margin came from a repeatable system, not from a single hire. TripleDart built Slate internally after deciding that no existing tool was capable of doing the work. Slate is an AI-agent platform. That is the same test you can apply to your own delivery.

Price the people you keep

Once the repeatable work is productized, the remaining service is the human work. That layer should be priced to cover fully loaded cost plus a margin, not to subsidize the product. Fully loaded cost includes salary, benefits, equipment, management time, and the overhead of keeping the person productive.

An underpriced service line quietly becomes the loss leader for the whole company; clients may love it, but the P&L will not. A services team can stay if it is priced like a cost center with a margin, not like a favor. That keeps the relationship honest and the cash predictable.

TripleDart argued the model can work profitably in India without a funding round. It claimed the approach covered the full inbound marketing function, a distinction it sees as a first for an Indian company. For a founder, the useful lesson is separating the product from the people and pricing each one correctly.

Hire only when the product pays

The third move is the hardest. Headcount is added only when the product revenue, not client revenue, can cover the salary before the next planning cycle. When the new person is needed to deliver more service, the service line must already be profitable enough to absorb the cost. When the new person is needed to build or sell the product, the product line should show the revenue path.

Venture investors placed over $300 million into services-as-software startups during 2026. The capital is a market cue. It tells you the market is paying attention to software margins in labor-heavy work. It does not tell you to chase profitable growth faster than your cash can carry.

A founder can judge the mix by asking whether the product line can carry the next salary, whether the service line covers loaded cost, and whether the remaining margin is cash you can keep. A yes gives you a services business with software-like economics. A no gives you a services business that needs a different price, a smaller team, or a clearer product.