Context
The client arrived with a contained request: refresh the brand, modernize the presentation materials, keep the existing typography and core colors. A few weeks of work, on paper.
Eighteen months later, they were one of the agency’s longest-running clients, and the engagement had grown to nearly three times its initial value across four distinct phases: brand refresh, ongoing design support, a full presentation system, and a website build. I was the sole project manager across all of it.
The stakes
Long-running engagements fail differently than projects do. Three patterns threatened this one:
Scope creep as a lifestyle. The client routinely expanded requirements after scoping, especially around presentations. Decks scoped at 10 to 15 slides grew to 40 to 60+. Priced naively, that either destroys margin or destroys the relationship.
Requests from everywhere, approval from somewhere else. Requirements came from the founder, the product lead and the marketing manager, while approval authority sat elsewhere. That structure produces ambiguity, rework and delay unless someone actively manages it.
Commercial friction. The engagement started as task-by-task requests, each with its own approval and payment. As volume grew, the administrative overhead grew with it: approval delays, recurring payment discussions, planning that reset every two weeks.
What I did
Redesigned the commercial model. I moved the engagement from individual task approvals to larger fixed-scope packages. Fewer approvals, predictable planning, less payment friction, protected profitability. The client got simplicity; the agency got stability.
Re-engineered how presentations were priced. When deck sizes exploded, I stopped pricing by total slide count and started separating unique slides from repeated, template-based ones. Combined with a template-driven delivery approach and asset reuse standards, this kept large decks profitable without cutting quality. Estimation logic is an unglamorous place to do creative work, and one of the highest-leverage ones.
Built the intake system. When the client hired a marketing manager, I worked with her to set up a structured, Notion-based request system: centralized intake, clear briefs, less ambiguity, faster design cycles. Request quality is a system property, not a client personality trait.
Managed the stakeholder map. I served as the single point of contact across founder, product lead, marketing manager, internal team and freelancers, with recurring alignment calls so that decisions were made once, by the right person, instead of repeatedly by different people.
Outcome
- Client retained for approximately 18 months, across four phases and three service lines
- Engagement grew to nearly 3× the initial contract value
- Margin protected throughout, despite ballooning deck sizes and continuous scope pressure
- Commercial friction reduced through fixed-scope packaging
- Intake and briefing quality permanently improved through the request system
- The relationship outlasted my entire tenure at the agency
What I’d do differently
I would restructure the estimation logic in phase one, not phase three. The slide-count problem was visible early; I solved it after it had already cost us margin on two decks. The lesson generalizes: when an estimate fails, redesign the estimation model, not just the next estimate.