Bonsai

Project management platforms are where an agency's capacity, dependencies, and billable assumptions become visible or stay hidden. When model costs fall (Claude Sonnet 5.5 runs 30% faster at up to 30% lower operating cost) and compute constraints push API pricing upward elsewhere, the margin difference between agencies comes from knowing which retainer hours are already committed. A tool choice is a margin decision disguised as an admin decision.

Permission Boundary Cost treats external client access as a design constraint that reshapes internal delivery, not a checkbox added at the end of a rollout. Every shared board, guest seat, or client-facing status view forces a decision about what internal commentary, margin data, and draft work stays hidden, and each of those decisions adds configuration and review time to every project. Agencies that map the boundary before migration avoid rebuilding templates twice; those that grant visibility first usually spend the next quarter walking it back. The cost is real but hard to see: a single shared workspace can mean dozens of per-project permission rules that a delivery lead maintains by hand. The frame matters because client-facing visibility is often the stated reason for switching platforms, yet the internal tax it creates rarely appears in the business case. Productive and Teamwork both bundle client access with financial data, which raises the stakes on where that line sits.

Every project management platform has two surfaces: the internal one your delivery team lives in, and the client-facing one your retainer depends on. The Client Visibility Surface framework says the second surface, not the first, sets your adoption ceiling. A tool with deep internal capacity but clumsy guest permissions forces account managers into screenshot-and-email workarounds, and that manual relay becomes the real coordination cost. Basecamp's deliberately flat client access and Plutio's white-label workspace both trade internal depth for a cleaner external surface, while Wrike and Asana push granular guest roles that need configuration before a client ever logs in. The framework matters because client constraints are a stated evaluation input, not an afterthought. Forrester's Q3 2026 research found workflow integration, not model selection, separates agencies scaling AI profitably, and the same logic applies here: the surface your client touches determines whether shared visibility reduces status meetings or adds a second reporting layer.

Migration drag coefficient is the ratio of switching cost to the workflow gain a new project management platform actually delivers. Agencies routinely price the subscription and ignore the drag: historical task data that will not import cleanly, client-specific permission maps, automations rebuilt from scratch, and the two to six weeks of reduced delivery throughput while teams relearn. A platform that wins on features but carries high drag can cost more in the first quarter than it saves in the first year. The framework forces the comparison onto total cost of transition, not list price. It also explains why standardization stalls: a single client with unusual access rules or an embedded approval chain can hold an entire agency on a legacy stack. The practical move is to score drag before scoring features, and to pilot on one retainer rather than migrating the whole book at once.

Compare total migration cost, including rework, retraining, and client re-onboarding, against the annual seat delta before committing to a switch.

Choose the platform whose workflow, permission model, and reporting match how the agency actually delivers, then negotiate seats; never let seat math drive the platform decision.

IF your agency runs more than roughly 15 concurrent client projects across three or more delivery leads, and no single client contractually dictates the tool your team must use, THEN standardize on one project management platform and absorb the migration cost once. IF client procurement, security review, or an existing enterprise agreement forces a specific system on more than a third of your retainer book, THEN run a federated model where each account keeps its mandated tool and you maintain a thin internal layer for capacity and margin reporting.

A productized engagement that moves an agency's delivery operations and its client-facing visibility into one governed workspace, with permissions, reporting, and adoption handled as explicit workstreams rather than afterthoughts.