Bump Scheduling: How Venues Rotate Staff Within a Shift

A bump is a relief handoff. One person walks to a position, takes it over, and tells the person they just displaced where to go next. That person walks on and does the same thing to somebody else. Nobody swaps, no crew stands down together, and no position is empty for longer than the sentence it takes to hand it over. Bump scheduling is the practice of planning those handoffs in advance: which positions exist, in what order people move through them, how long each turn lasts, and who enters the chain when.

If you run a venue you already do this, probably on a spreadsheet you rebuild by hand three or four times a day. This guide documents the mechanic properly, because almost nothing written about it exists outside enthusiast forums and internal training decks.

First, the word means two different things

In a represented workforce, bumping normally means seniority displacement: a role is cut, a longer serving employee exercises a contractual right to take a less senior colleague's job, and the collective agreement decides how far that ripples. It is a process measured in weeks, and it changes who is employed where.

A relief bump is a handover at a post. It takes about four seconds, it changes nothing about anyone's employment, and it happens dozens of times a day in the same building. The practical fix is small: write relief bump or position bump in anything a shop steward will read. Nobody on the floor will be confused, but people reading a policy document at a desk absolutely will be. The operational sense is the one covered in the definition of bump scheduling.

The position model

A rotation plan is only as good as the list of positions underneath it, and most venues have never written that list down precisely. A position is a place with work attached, and four things about each are worth recording: what it does to a body, whether it can be left empty, who may hold it, and whether anyone actually wants it.

A typical attraction

A ride team usually runs five or six posts. Small attractions merge the last two.

• Greeter. First guest contact. Height and rider requirement checks happen here on many attractions, which makes it the position where refusals get delivered. High verbal load, standing, often the most exposed spot in the queue.
• Grouper. Assigns parties to rows or vehicles. This is the throughput position: a grouper who leaves seats empty costs more capacity per hour than a slow loader does, and almost no rotation plan reflects that.
• Loader. Restraint checks and dispatch readiness. Physical, repetitive, and usually carrying a signed off competency on the restraint check itself.
• Dispatch console. The gated one. On most attractions the ride cannot run without a certified operator here, so this post closes the attraction when it cannot be filled. Low physical load, high consequence, and attention degrades well before the operator notices.
• Unload. The mirror of load, plus loose articles, guests who need assistance, and anything left in a vehicle.
• Exit and stroller area. Crowd flow, lost property, and the guest recovery conversations everything upstream generates. The heaviest post on a bad day, and the one most often treated as easy.

A typical store

Retail has the same shape with different gates.

• Register. Cash accountability changes the handover. Where drawers are assigned by name, a bump means a session close and a drawer handover, costing thirty to sixty seconds that a gate or a platform does not.
• Stockroom. Out of guest view, physically demanding, no service pressure. In practice the recovery position, where you send somebody who has spent two hours facing a queue.
• Floor. Zoning, recovery, and guest questions. Highest discretion and lowest measurable output, which is why it is the first position cut on a short staffed day and the first place rotation quietly stops happening.
• Mobile cart. The hardest position in the building to schedule. One person, exposed to weather, holding cash, nobody to cover a toilet break. It cannot be left, so its relief can never be late, which makes it the constraint the rest of the chain bends around.

Across both models the same four properties do the work: strain, exposure, gating, and desirability. The first three get written down eventually. Desirability almost never does, which is how one person ends up on the till bank every Saturday for a year. Ordering the posts people move through is what position rotation is for.

The bump itself

The spoken handover is one sentence carrying two facts: I have this position now, and you are going to that one. No confirmation step, no board to consult, no message to acknowledge.

That matters more than it looks, because at most large venues the frontline workforce carries no phone, and being seen with one is often a disciplinary matter. The bump is not a convenience, it is the notification layer. The plan reaches people one hop at a time through the mouths of the people executing it, and the only part that reaches a loader is the half sentence spoken by whoever relieved them. So the chain has to be built such that each person's next position is knowable by the person relieving them. Anything depending on information only the office holds will not survive the walk.

A worked cascade

Seven people on a ride team, six positions, one person off the loop at any moment, rotating on a 45 minute interval. Walking distances measured with a stopwatch, not estimated.

1. 14:00. The relief leaves the break room and walks three minutes to the entrance.
2. 14:03. Relief takes the greeter position. The greeter walks one minute to the grouper spot.
3. 14:04. Grouper is bumped, walks two minutes to the load platform.
4. 14:06. Loader is bumped, walks one minute to the console. This person holds a current certificate, which is why the chain is ordered this way and not the other way round.
5. 14:07. The console operator is relieved and walks two minutes to unload.
6. 14:09. Unload is bumped, walks two minutes to the exit area.
7. 14:11. The exit attendant is bumped and walks four minutes to the break room, arriving 14:15.

One relief moved seven people in fifteen minutes, no position was uncovered, and the seven legs average just over two minutes of walking each.

The part that surprises people is that the stagger itself is not unfair. The greeter was relieved at 14:03 and the exit attendant at 14:11, but on the next chain those two are relieved at 14:48 and 14:56, so both served exactly 45 minutes. A constant stagger produces equal turns. Only variation in it creates unequal ones, which is why chain discipline matters more than symmetry.

The walking is not free. Ten moves per person across an eight hour shift at just over two minutes each is close to 25 minutes, roughly three hours across a team of seven that no roster ever accounted for. Halve the interval to 20 minutes for a heat exposed post and the same team spends over six hours walking. That subtraction is the argument for treating walk time as a scheduled quantity.

What breaks when a link is late

A chain can run long. It cannot run short. Nobody leaves a post before relief arrives, so every leg has a floor and no ceiling, and delays accumulate rather than cancelling out. If each of seven legs runs 40 seconds over, because a guest had a question, because a drawer needed counting, because the concourse filled up between one post and the next, the chain finishes almost five minutes late. Ten chains without a reset puts you 45 minutes adrift by close.

You get to choose who pays, and the choice is binary. Start each chain on the clock and the drift lands on whoever is off the loop, whose break shortens every cycle. Start each chain when the relief is actually free and the drift lands on the plan, so every stint lengthens and the exposed positions blow past their intervals. Most venues do the second without deciding to.

The repairs are two. Build a buffer into each leg, which costs coverage you have to staff for. Or change where the chain enters, promoting a different person to relief and slotting the late one in further along, which costs nothing but is hard to work out on paper in the ninety seconds you have. That rerouting is the practical case for staggered relief being planned rather than improvised.

Where breaks actually fit

This is where most spreadsheets fall over, because break length is not a policy decision in a bump rotation. It falls out of the geometry.

In the example above, the exit attendant leaves the floor at 14:11, reaches the break room at 14:15, leaves again at 14:45 to start the next chain, and is back on a position at 14:48. That is 37 minutes off the floor, of which 30 are actually sat down. Nobody chose 30 minutes. The interval chose it.

The frequency is worse. With six positions and one person off the loop, a full cycle is seven turns, and at 45 minutes that is five hours and 15 minutes between breaks. An eight hour shift with a meal and two rest breaks does not fit inside that cycle, and no care with the grid will make it fit, because cycle length is fixed by headcount and interval.

There are three honest ways out.

• Shorten the interval. A 20 minute rotation across the same seven slots delivers a break every two hours and 20 minutes, closer to a real policy. It also roughly doubles the moves and the walking, so you pay in coverage.
• Add bodies to the loop. Every extra person is another turn in the cycle, which pushes breaks further apart rather than closer. Adding headcount to fix break frequency makes it worse, and this catches people out constantly.
• Run breaks on their own chain. A dedicated breaker walks a break list on the break schedule while position rotation runs on its own interval. Two chains, two clocks, one set of people. Most large venues converge here, which is why the breaker exists as a named job.

Once breaks run separately, the legal break window becomes a hard constraint. A break due between the third and fifth hour has to land there, the position loop has no reason to cooperate, and the breaker's route gets re solved whenever somebody calls in.

Why the spreadsheet stops working

A grid with one row per person and one column per 45 minute block encodes an assumption you never agreed to: that everybody moves at the same instant, at the column boundary. The chain above took 15 minutes to run. For a third of every interval, the spreadsheet describes a floor that does not exist.

You can live with that for one loop of four positions and a single relief. It collapses once two things are true, and in most venues both already are. Intervals differ by position, so an outdoor platform turning every 20 minutes and an indoor booth turning every 60 no longer share column boundaries. And the plan changes during the day: somebody calls in, somebody is late back, weather closes an outdoor attraction and its entire labor has to land somewhere else.

Rebuilding a grid by hand takes long enough that the plan is stale before it reaches the break room, so people stop rebuilding, print the morning version, and call changes over the radio. Every constraint the plan carried goes with it: certification gating, exposure limits, break windows, and any record of who has already taken the hard positions today. Planning that layer deliberately is what intrashift scheduling means, and it sits below the roster rather than replacing it.

It is a different discipline from intraday management, which tracks volume against forecast. That work is interval based and demand driven. This is position based and constraint driven. Sites with queues and posts need both.

What to write down before you automate anything

Whatever tooling you end up with needs the same inputs, and gathering them is a morning's work with a clipboard.

• Every position, with the minimum headcount that keeps it open and whether it can be closed at all.
• The gate on each one, meaning the certificate or system permission it requires, and that certificate's expiry date.
• An interval per position and its reason, whether exposure, attention, strain, or service quality. One interval for a whole area means the list was never made.
• Measured walk times, taken at opening and again at peak, because the crossing that takes six minutes at open takes eleven once the site fills.
• Break entitlements and their windows, kept separate from the position loop, with the breaker route treated as its own schedule.

Sizing the relief pool against that list is a coverage headcount question rather than a rostering one, and the shift boundaries around it belong to the usual shift patterns reasoning. A printable sheet from the rotation schedule generator will show quickly whether your loop geometry survives your break policy.

The roster and the rotation are two layers and they fail differently. A roster built by auto-scheduling can be legal, fully staffed, and still hand somebody six hours on the hottest post on the property. The bump chain is the layer that decides that, and it deserves as much planning as the shift that contains it.