The first time I reshuffled seats mid-term, three students told me the draw was rigged before I'd finished writing the new plan on the board. They weren't being difficult — from where they sat, a teacher quietly deciding who moves where looks exactly like a teacher quietly deciding who moves where. That is the real problem with randomising a classroom: the randomness is easy, and convincing thirty people it was random is the hard part.
Why perceived fairness matters more than actual fairness
A genuinely random draw that students suspect was fixed buys you nothing. They'll argue the result, negotiate for swaps, and treat the next draw as a negotiation too. Meanwhile a visibly random draw — even a low-tech one — usually gets accepted with a groan and no argument, because there's nothing to argue with.
Practically, that means the draw has to happen in front of them, in one continuous action, with no pause where you could plausibly have intervened. The moment you say "let me just adjust one thing," the whole thing reverts to being your decision, and you've spent the randomness for nothing.
This is also why a pre-printed seating plan handed out at the door lands badly even when you generated it randomly the night before. Same result, no witnesses.
The no-device version: do it live, badly, on purpose
You don't need software. Numbered lollipop sticks in a cup, drawn one at a time onto a seat map on the whiteboard, does the job and does it very visibly. It's slow — five to eight minutes for a class of thirty — but the slowness is part of why it's believed.
One detail worth getting right: draw the seats in a fixed order and pull names at random, not the other way round. If you pull a name and then choose where to put them, you've reintroduced yourself as the decision-maker. Fixed order, random name, no exceptions.
Announce the rule about swaps before you draw, not after. "No swaps today, we'll redraw in three weeks" is accepted at the start of a draw and resented at the end of one.
Random grouping, and the same three friends
Group work has the opposite failure mode from seating. With seats, students want to sit with friends. With groups, the complaint is usually the reverse — the same cluster ends up together every time and nobody else gets a turn working with the strong students.
A one-click random split fixes this only if you're willing to accept the occasional bad draw. Sometimes randomness produces a group of four who genuinely cannot work together, and pretending otherwise to protect the purity of the method wastes a lesson. My rule: I reroll the whole class, out loud, stating why — "that group won't work, we're doing it again" — rather than quietly moving one student. A visible reroll of everything stays fair; a quiet fix of one group does not.
The other thing worth doing is varying group size between lessons rather than always using fours. Fours reliably split into two pairs. Threes don't, and fives force someone to take a coordinating role. If you always use the same size, students settle into the same roles within it.
The random picker problem: "you always pick me"
Cold-calling by memory is much less random than it feels. Teachers reliably over-pick students sitting in the middle-front, and under-pick the two corners at the back — and students notice the pattern long before we do.
A visible random picker fixes the perception problem, but it introduces a new one: a genuinely random picker will sometimes land on the same student twice in five minutes, and that student will be certain the machine is against them. Two things help. First, say up front that repeats are possible — a draw that can't repeat isn't random. Second, for a fast-recall round, draw without replacement across the round so everyone gets one turn, and only allow repeats in open questioning.
Anxious students deserve a carve-out here. A random picker that can land on a student who genuinely freezes when called on is not a neutral tool for that child. Tell them privately that they're out of the pool, or agree a signal that means "pass" — and don't announce the arrangement to the class.
When not to randomise
- Access needs come first. Hearing, vision, mobility and support-staff seating are not part of the draw. Fix those seats before you start and say so.
- The week before a high-stakes assessment. A new seat neighbour costs a few days of settling. Don't spend them then.
- When two students genuinely should not sit together. Exclude the pairing before the draw rather than fixing it after — and if you can, exclude it silently by fixing one of the seats in advance.
- Exam layouts. These aren't random at all; they're a separate saved plan.
Naming the exceptions out loud at the start actually strengthens the draw rather than undermining it. "These four seats are fixed for reasons I'm not discussing, everything else is random" is a much more credible claim than pretending everything is up for grabs.
Keep more than one layout, not one you keep redrawing
Most classes need three arrangements over and over: normal rows for direct teaching, clusters for group work, and a spaced-out layout for assessments. If you keep a single plan and redraw it each time, you spend a few minutes of every transition rebuilding something you already had.
Keeping them as separate saved layouts turns "rearrange for group work" from a redraw into a switch. It also means the group-work layout can preserve the randomised groups you drew earlier in the week instead of quietly reverting to friendship clusters as students drift back to familiar seats.
How SparkMyClass handles seating and grouping
SparkMyClass's classroom tools cover the three pieces described above: a drag-and-drop seating chart with multiple saved layouts per class, one-click random grouping with instant reroll, and an animated random picker.
The part that matters most for the fairness problem is that all of it runs on the teacher's screen — students don't need devices, don't log in, and don't bring anything. The draw happens on the projector, in front of everyone, in one action. That's the same property the lollipop sticks have; the software just makes it take five seconds instead of eight minutes, and lets you keep the result.
None of it removes the judgement calls. The tool will happily draw a group that you know won't work. Deciding to reroll — and doing it visibly — is still yours.
See the classroom tools, or read about tug-of-war team quiz games for the same no-device-needed approach applied to revision.
Final thoughts
Randomising seats and groups is one of the cheapest things a teacher can do to break up settled social patterns in a room. It fails for a reason that has nothing to do with the randomness itself: students don't believe it. Draw in public, in one action, with the exceptions named in advance, and the arguments mostly stop.