Sizing a design team usually starts with a ratio someone half-remembers and ends with a pie chart nobody can defend. This planner makes the assumptions visible: how each design role spends its week, who owns which decisions, what that time buys the business, and which of the numbers people quote actually hold up.
01
Pick a role and stage
Six roles, from product designer to head of design, across three company stages.
02
Move the week
Capacity is fixed. Shift time between activities and watch the outcome profile respond.
03
Check the evidence
Every claim is graded, including the popular ones that do not survive a source check.
Role
Stage
Product Designer
IC, mid–senior · Growth (B–D)
Typical experience
3–7 years
Owns end-to-end outcomes for a product area. The default unit of design capacity. Product works, the org does not yet. Balanced profile.
12
10
UX / IA15%
UI craft18%
10
8
10
Failure mode. Squeezed to ~60% UI production when the PM partnership is weak. That is the most common way a design hire under-returns.
Adjust the week
Capacity is fixed at 100%. Raising one activity pulls the others down in proportion, which is the whole argument.
12%5h
5%2h
10%4h
15%6h
18%7h
10%4h
8%3h
5%2h
10%4h
4%2h
3%1h
0%0h
Split of the week
Hover a segment or slider
Expected outcome profile
Change against the benchmark allocation for this role and stage. Relative, not a forecast. Diminishing returns are modelled, so pouring everything into one activity does not max anything.
They are practitioner heuristics, not measured data. There is no credible public dataset on how designers split their week, so the defaults are a starting position built from experience running design organisations. They ship as sliders precisely because they are arguable.
Nielsen Norman Group's 2020 survey found the most common answer was one designer per ten or fewer developers, and one researcher per five or fewer designers. That puts design at roughly a tenth of product headcount, close to the 10–13% of project budget their earlier usability research pointed to. Domain complexity and the number of surfaces you ship move it considerably.
Each owns different decisions. In this model the product designer is accountable for the solution direction of a feature and for checking the build matches it, the design manager for the quality bar and design review sign-off, the design systems designer for component standards, and the head of design for vision, cross-area consistency and hiring. The Decision rights view shows the full matrix.
Once the system has enough consumers to need an owner, roughly six designers or forty engineers. Below that, a shared component library maintained by product designers is enough, and a dedicated role tends to run at a loss for its first year.
No published study with a method behind that figure has been traced, even though it is widely attributed to Forrester. The defensible version of the argument is that fixing a problem costs about $1 in design, $5 in development and $30 after release.