Group projects generate more grading complaints than almost any other assignment type, and it's almost always for the same reason: one shared grade gets applied to unequal work. Here's a structure that avoids the usual fights.
Separate the product grade from the process grade
Grade the actual deliverable (the presentation, the report, the product) as one score that the whole group shares — that part genuinely is a group outcome. Then grade individual contribution separately, based on a different source of evidence entirely. Conflating these into a single number is where most of the unfairness actually comes from.
Use a structured peer evaluation, not a vague one
"Rate your group members" invites either uniformly generous scores (nobody wants to be the reason a friend loses points) or occasional score-dumping from an unrelated conflict. A structured version — rating specific, observable contributions like "met deadlines," "contributed ideas in planning," "completed their assigned section" — produces evidence you can actually act on instead of a popularity contest.
Require an individual accountability piece
A short individual reflection, a section of the project explicitly attributed to one student, or a brief individual quiz on the group's topic gives you a piece of evidence that isn't dependent on anyone else's effort. This is often the single most fairness-protective addition you can make to a group assignment.
Set expectations before the project starts, not after a complaint
Explaining upfront exactly how the product grade and the individual grade will be calculated, and what happens if peer evaluations flag someone, removes most disputes before they happen. Explaining it for the first time after a parent emails about an unfair grade is a much harder conversation.
Keep the rubric criteria separate by grade type
If you're using a written rubric, keep "group product" criteria and "individual contribution" criteria in clearly separate sections rather than blended together — it makes both grading and explaining the grade dramatically simpler, and it maps cleanly onto rubric-based tools if you're using one to help score the product portion, like ClassGrade Pilot.
Group work teaches real collaboration skills, but only if the grading behind it holds up to the obvious question every student eventually asks: "why did I get the same grade as someone who did less?"