Most fall schedules fail in the first two weeks, and the cause is almost always something that was skipped in July. Here is the order that works, with a timeline you can copy.
Pick the date of the first lesson and count backward. A fall schedule needs families to confirm their time at least a week before lessons begin, so that is your hard deadline for a finished schedule. Add a week to send it out, a week to absorb change requests, and you can see why the real start of fall scheduling is the second half of July, not the last week of August.
If your first lesson is the Tuesday after Labor Day, here is a workable shape. Eight weeks out you collect re-enrollment answers. Six weeks out you freeze the roster. Four weeks out you build. Two weeks out you publish and fix. Lesson one is the payoff, not the starting gun.
Send one form, not three emails. For returning students you need three answers: are you coming back, what days and times can you do, and is anything different from last year (new school, new sibling, moved house). For new students add instrument, age, experience level, and a teacher preference if you offer one.
Ask for availability as windows, not as a single preferred time. "Tuesdays and Thursdays after 3:30, never Wednesdays" gives a scheduler room to work. "Tuesday at 4" gives it nothing and guarantees a call back. Where possible, ask families to rank two or three acceptable windows.
Teachers change more than owners remember. Before placing a single student, confirm each teacher's days, start and end times, the longest day they will accept, which instruments they teach, and any hard limits such as a standing second job on Fridays. Write it down per teacher. This sheet is the constraint set the whole schedule has to fit inside.
Decide the caps explicitly. A teacher who is "fine with long days" will still run out of patience after a day that mixes back to back lessons with forty minutes of driving. If driving counts as paid work in your studio, count it toward the cap. If it does not, say so out loud, because the teacher is counting it.
Pick a freeze date and hold it. Every student added after the build changes the schedule for someone who already confirmed. Late joiners are not a problem, because they get placed into open slots afterward. Late changes to existing students are the problem, so make the rule simple: after the freeze, existing students keep their slot unless they ask to move.
Do a quick data sanity pass on the frozen roster. Duplicate students, misspelled teacher names, siblings entered under different last names, and availability typed as free text that nobody can read are the four things that quietly break a build. Fixing them takes an hour now and a week of phone calls later.
Build in layers. Place the students with the least flexibility before anyone else: the sibling group that has to be on one teacher the same afternoon, the student whose only open window is Thursday at 4, the family that needs a teacher of a specific gender. Group classes and anything already fixed in the calendar come next because they remove time from teachers.
Then place the flexible majority around them. This ordering is the single biggest quality lever in hand scheduling. People naturally start with the first name on the list, and by the time they reach the hard cases the good slots are gone.
A schedule can have zero double bookings and still be a bad week. Look at each teacher's day as a whole: the length from first lesson to last, the longest gap with nothing in it, whether there is a meal break before the fifth hour, and how much of the day is travel. Teachers feel the shape of a day far more than they feel any one lesson.
This is the step most manual schedules skip, and it is where a scheduling tool earns its keep. Set the Tune scores every teacher day for double-booked minutes, time over the cap, meal-break problems, and drive deficit, so a bad day shows up before a teacher finds it. You can see what that review looks like on the demo.
Send each family their time with the teacher name, the address or room, the start date, and a plain statement of how to request a change. Give the change window a closing date, usually the end of the first week of lessons. Inside the window, handle changes one at a time and re-check the two days each move touches. Outside it, changes wait until the next break.
Keep a short written list of who has asked to move and why. By week three that list tells you whether the schedule has a design problem (the same Thursday slot keeps failing) or just normal churn.
Counted back from a first lesson in the week after Labor Day. Shift the whole thing by the same number of days for a different start.
The two repeat offenders are skipping the freeze and skipping the day review. Without a freeze the schedule never stops moving. Without a day review you ship a calendar that is conflict-free on paper and exhausting in practice. If you only fix two things this year, fix those. For the cases that are harder than a standard fall start, see the notes on season scheduling and multi-teacher studios.
Plan on starting the re-enrollment form about eight weeks before your first lesson and building about four weeks out. That leaves two weeks to publish and fix before lessons begin.
Say before the deadline that late answers are placed into the slots left over after everyone who replied on time. Hold a few open slots per teacher for exactly this reason.
Under about twenty students with one teacher, a spreadsheet works. Once you have several teachers, siblings, travel, or alternating weeks, the number of constraints grows faster than a person can track, and a tool that checks them all is worth it.
Not freezing the roster. Every change after the build ripples through other families. Pick a date, hold it, and place late joiners into open slots.