Balanced Allocate

Most allocation tools solve one domain. There are dedicated apps for classroom placement, dedicated apps for sports grading, dedicated apps for seating charts. Each one hardcodes the rules for that specific problem and charges accordingly. If your problem doesn't fit their template, you're back to a spreadsheet.
Balanced Allocate doesn't know what your data means. It doesn't know the difference between a student and a truck. You upload any spreadsheet, define your own constraints, and the solver finds the optimal assignment. The same tool that places 200 students into 8 classes also forms a 10-vendor procurement panel from 40 tender responses. Same solver, same interface, different CSV.
What makes it different:
The solver is real. It's Google's OR-Tools CP-SAT — the same mathematical optimisation engine used in logistics, airline scheduling, and supply chain planning. It doesn't approximate or shuffle randomly. It evaluates the full solution space and returns the mathematically optimal result, or tells you the problem is infeasible and shows you which constraints conflict.
Every constraint type works simultaneously. Budgets cap numeric totals per group. Quotas enforce counts or percentages. Balance spreads values evenly. Relationships keep items together or apart. Preferences steer items toward groups with adjustable strength. Pins lock items in place. Maximise and Minimise concentrate the best or worst values where you want them. Combined budgets share a limit across multiple groups. Set as many as you need — the solver handles the interactions between them.
Validation is built in. After every run, the tool checks every rule you set and shows pass or fail. You don't have to manually verify the results. Export the Excel report and the constraint audit trail is already there — every rule, every result, documented.
The select/reject pattern is native. Most allocation tools assume every item gets placed. Balanced Allocate handles problems where most items are rejected — grant selection, vendor shortlisting, award panels. Set a fixed-size target group and a remainder group. The solver picks the best candidates that satisfy all constraints and puts the rest in the reject pool.
There is no domain lock-in. The nine worked examples cover education, sports, events, healthcare, logistics, government, real estate, manufacturing, and procurement. Users have also applied it to fantasy league drafts, chore rotations, and conference breakout sessions. If it goes into groups and you have rules, it works.