10 Lessons to Scale a Marketing Agency to $35M

There's no moat, so the people are the moat

Run a competitive analysis on a digital agency and you find no patents, no exclusive distribution. "Someone could wake up in any country in the world and decide I am a digital marketing agency." That commoditisation is accelerating, not softening: IBISWorld counts roughly 114,000 US advertising agencies in 2026 (up ~4.4% year on year), inside a $56.9 billion digital-advertising segment growing ~6.6% a year. Demand is not the problem. Differentiation is. Asked why clients stay, the founder circled back to reputation: "we over-incentivize and overpay our employees, and that encourages them to over-deliver to our customers." The moat is never the product. It is the standard of the people producing it.

An agency does not produce anything. The value you're giving your customers is your employees' time.

Force growth, and pay for overachievement

The trap is comfort: a founder reaches profitability at five or ten employees and settles. "Your best performers aren't going to want to stay in a company that's not growing." SmartSites' answer was forced 30–35% annual growth as deliberate policy — because it manufactures the promotions and stretch roles that keep ambitious people from walking. The retention logic is borne out by current numbers, and they are getting worse: Gallup's State of the Global Workplace 2026 puts global engagement at just 20% — the first back-to-back annual decline since Gallup began measuring — and Pew Research finds 63% of quitters cite no opportunities for advancement. Agency turnover runs roughly 18–30% by various benchmarks, well above the 13.5% cross-industry norm.

Compensation follows the same logic. SmartSites pays above-market rates, but on a deliberately low fixed base paired with an aggressive variable upside: PPC analysts and project managers earn a percentage of the cash flow of the clients they manage. "There are literally two reasons people leave, making up 90% of all leaving. Number one, they hate their boss. Number two, they don't feel compensated fairly." The Incentive Research Foundation's 45-study meta-analysis backs the mechanism: incentive programs raise performance by an average of 22%, climbing to 44–48% for programs running longer than six months. To remove the risk for new hires, SmartSites converts the plan into a first-year guarantee — a modest base plus expected variable pay, guaranteed as a single number for year one.

Delegate before you become the bottleneck

Every growth stage demands a different operating model, and clinging to the last one caps you: "What got us to $5 million in revenue — if we kept doing the same thing, we'd have gotten to maybe $6 million ten years later." SmartSites moved through four stages — wearing every hat at 5–10 people, appointing department heads at 30–40, building tight processes at ~100, and, candidly, still hunting for the next model between 400 and 800.

The founder's own confession makes the case for delegation better than any framework: running SmartSites at 200 employees with no HR department, personally doing final interviews and reviewing every contract. "I literally became the bottleneck. We started hiring slower because I was too busy to do that stuff." Inc.'s "80 Percent Delegation Rule" makes it explicit: hand off any task someone else can do at 80% of your quality, because the missing 20% is rarely worth the bottleneck. The end state is replaceability — "if your absence breaks the business, you didn't build one" — which is exactly the discipline behind mandatory-time-off policies some firms use to stress-test whether the business survives without a given person.

Kill the KPIs that reward presence, not output

Organisations accrete silly metrics nobody intentionally created. Going remote forced SmartSites to notice its own: a fingerprint clock-in machine at the office door, quietly rewarding people who "clock in and then sit and stare at the monitor for an hour." The fix was intentional, department-specific KPIs co-designed with each team — because where the product is output quality, measuring input presence measures the wrong thing entirely, especially in a remote, AI-augmented agency where the best people finish in a fraction of the hours.

The same logic pushes decision authority to the edge. Past roughly 15–20 employees, a founder can no longer personally coach every call, so SmartSites runs its own version of the Ritz-Carlton $2,000 rule: every employee, at every level, is empowered to give free work to fix an upset client, no approval required. "Without empowering all your employees to make decisions, you're not going to be agile." Tellingly, SmartSites' most common exit-interview complaint is the opposite problem — giving people responsibility before they're ready — which the founders treat as an acceptable cost of staying fast.

The margin trade-off, and the AI question

None of this is free. "Profit as a percent has been worse and worse. If we were a public company, our stock would be in the garbage." Most of an agency's cost is people plus its own marketing spend, with little price elasticity — you can't raise client fees 10% a year, but employees expect exactly that raise. SmartSites deliberately trades margin for growth, a luxury enabled by having no board to answer to.

AI makes a junior analyst who knows nothing "pretty good," never excellent. The moat is the value of experience and the data set an established agency owns.

The founders are betting AI reverses the compression, and 2026 research agrees on direction while flagging a catch: AI is sorting agency work into fully automated, AI-assisted and fully human tiers — hollowing out precisely the junior rung where strategy used to be learned by doing the manual work. Four moves to run this quarter:

• Set an explicit growth target and tie it to a published promotion ladder — retention insurance, not ego.
• Redesign comp around variable upside, then de-risk new hires with a first-year guaranteed number.
• Audit your accidental KPIs this month — delete anything that rewards hours logged instead of outcomes delivered.
• Give every client-facing employee a defined "Ritz-Carlton budget" — explicit authority to fix a problem without approval.