Understanding your results

The stats panel

After every generation, Clever Groups shows a stats panel below the group display. It contains up to three lines depending on the mode you used:

The bold line is the primary metric for the generation mode you used — the one Clever Groups was optimising for. The other lines are supplementary context. Which line is bold depends on the mode:

• Preference Groups — satisfaction score is bold.
• Demand Groups (all preferences) — any preference placed is bold.
• Demand Groups (first preference only) — first preference placed is bold.
• Random Groups — satisfaction score is bold.

The satisfaction score

This is the score Clever Groups has shown since the beginning, and the one Preference Groups optimises for. It measures how well each group is doing at honouring the preferences of the people in it — then averages that across all groups.

Rank matters — earlier preferences score more

The score doesn’t just count whether a preference was met — it rewards getting an earlier preference over a later one. Getting your first choice contributes more to the score than getting your second, which contributes more than your third.

Consider two arrangements for the same class of 20 students:

• Arrangement A: 15 students get their first choice, 5 get nothing — scores higher.
• Arrangement B: All 20 students get their third choice — scores lower, even though more preferences were technically “met.”

This is why the score reflects the quality of the arrangement, not just a simple count. A score of 84% means the arrangement achieved 84% of the best possible weighted outcome for this team — not that 84% of preferences were placed.

What’s a good score?

This varies significantly by team size, group size, and how many preferences were submitted. Smaller teams with larger groups tend to score higher because there are more opportunities to match preferences. Larger teams with small groups and few preferences submitted will naturally score lower. Use the score to compare runs against each other, not as an absolute benchmark.

Demand Groups scores: first preference placed & any preference placed

When you run Demand Groups, two additional lines appear in the stats panel. These are simpler binary counts — they tell you what happened for each individual, not an averaged score across groups:

• First preference placed — the percentage of students who ended up in a group with their very first preference. Shown as both a percentage and a count: “72% (18 of 25).”
• Any preference placed — the percentage of students who ended up in a group with at least one person from their preference list. “88% (22 of 25)” means 22 out of 25 students got at least one preference placed.

The denominator is students who submitted at least one preference — students who submitted no preferences are excluded from these counts.

Reading the gap between the two lines

If first preference placed is 72% and any preference placed is 88%, it means 16% of students got a preference placed but not their first choice. Whether that’s acceptable depends on the situation — at a school camp you might want to chase a higher first preference rate; for a classroom activity group, 88% any preference placed may be entirely satisfactory.

Preference glyphs

Each student card shows a small glyph indicating how their preferences were met. These appear both in the group display and in display mode:

What to expect by mode

Preference Groups and Demand Groups each optimise for different preference outcomes. Random Groups doesn’t optimise for preferences at all — but the scores are still measured and shown afterwards. The result is that all three modes are directly comparable on the same team data:

Demand Groups may produce a slightly lower satisfaction score than Preference Groups on the same data — that’s expected. It’s solving a different problem: ensuring no one is left without any preference placed, even if the overall average is marginally lower.

For Random Groups, preferences play no part in how groups are formed — the draw is completely blind. The scores appear afterwards as a measurement of what happened to any existing preferences in the random arrangement. They’re not something the algorithm was chasing; they show you what preference outcomes look like when left entirely to chance. If Preference Groups scores 84% satisfaction on your team and a random draw scores 31%, you can see exactly how much the preference data is doing.

Not sure which mode to use? See Group generation modes for a full explanation of when to use each one.