Amazing Race

How to Design an Amazing Race for a Singapore Office

Amazing Race is the most widely sold team building format in Singapore, carried by at least eight corporate operators under that exact name, and one specialist operator sells nothing else. Vendors here price it from roughly $25 to $130 per person, run it in 2 to 2.5 hours as standard, and stage it everywhere from Pulau Ubin and Fort Siloso to the interior of 313@somerset. The format takes its name and its shape from the television series: teams receive a clue, the clue sends them to a place, they complete a task there, and completing it earns them the next clue.

What almost nobody publishes is the design. Search for it and you get the vibe, the photographs of people in matching T-shirts, and a price. You do not get the thing an HR person actually needs, which is a number of stations for a number of people, a rotation that stops teams walking into each other, a rule for handing over clues, and a scoring method that does not let one station decide the whole result. This page is that design. Everything below can be run by a company without a vendor, and the sections that cost money are marked.

What Singapore vendors actually sell when they say "Amazing Race"

An Amazing Race is a route-based, clue-gated, multi-station team competition. Teams of four to six move around a sequence of checkpoints, complete a scored task at each, and unlock the next checkpoint only by finishing the current one. The result combines task performance with elapsed time.

Three words in that definition do the work, and they are what separate this format from its neighbours.

Route-based. Teams move. The distance and the choice of where to go next are part of the game, not logistics.

Clue-gated. A team cannot go to station four until station three releases it. This is the mechanism that spreads six teams across six stations without a marshal shouting rotations.

Multi-station. Each checkpoint has its own task, its own scoring unit, and its own capacity limit.

Singapore operators put their own badge on the same skeleton, and the published specifications behind those badges are far more alike than the names suggest. Operators describe the format in their own words as sequential and surprise-driven.

Published per-pax rates run from about $25 to $130, with $30 to $80, $40 to $60 and a $50 to $120 mid-range tier all appearing as bands, and $130 per person quoted at 20 to 99 pax falling to $80 at 100 to 499. Published group ranges run from 2 participants at a single fixed location up to 2,000, with 250 given as the ceiling for a single un-staggered session, 500 a common upper bound, and 30 a common floor. Published durations run from 2 to 2.5 hours as the standard, through a 3 to 4 hour half-day, out to 2 to 5 and 2 to 6 hours for the longest formats. Teams are quoted at 4 to 6. Minimum event fees from $2,000 apply to smaller groups.

Inclusions are where the listings genuinely differ: 8 to 15 challenge stations on a custom route, an MC, real-time scoring and one facilitator per 25 to 30 participants at one end; missions delivered through a WhatsApp game bot with photo and video evidence at another; and elsewhere a set of named editions built around fixed Singapore locations.

Read those together and the picture is consistent. The format is elastic in every direction except one: the number of stations has to bear a specific relationship to the number of teams, and that is the single design decision that most in-house attempts get wrong.

How do you actually run one, step by step?

This is the whole job. Work through it in this order.

Step 1: Fix the team count first, then the station count

The team count comes from the headcount. Divide by five. Round to a whole number of teams, then adjust team size between four and six to absorb the remainder. Thirty people is six teams of five. Twenty-eight people is six teams, four of five and two of four. Forty-four people is eight teams, four of six and four of five.

The station count then follows one rule that governs everything else:

Never run fewer stations than teams.

If you have six teams and five stations, then at every moment of the race one team is standing in a queue. A queue is not a neutral cost. It is the single thing participants complain about afterwards, and it compounds: a team held for four minutes arrives four minutes late at its next station, which pushes the team behind it, and by leg four the schedule is fiction.

With stations equal to teams, every team is at a different station at every moment and the schedule is self-enforcing. With stations greater than teams, you gain slack and route choice at the cost of more setup. The practical band is stations equal to teams, or stations equal to teams plus one or two.

Above roughly 60 people the honest answer is that you stop running one race. Either you duplicate the whole station set so that two independent races run side by side and you compare normalised totals at the end, or you split into heats that race in sequence. One operator's published ceiling of 250 in a single session, with staggered starts required above that, is the same admission expressed as a booking policy.

Step 2: Choose simultaneous or staggered starts

There are three start patterns and only one of them is usually right.

Simultaneous distributed start. Every team starts at the same instant, at a different station. This is the default and it is correct whenever stations are equal to or greater than teams. Nobody waits, ever, because the rotation is a closed cycle.

Staggered single-line start. All teams start from one point at fixed intervals and follow the same station order. This is what people imagine when they picture the television show, and in a company setting it almost always fails. With T teams sharing S stations on one route, you avoid a queue only if the release interval I is at least the station play time P multiplied by T divided by S. With six teams, six stations and a nine-minute station, that is a nine-minute interval, which means the last team starts 45 minutes after the first and the event has become a relay. Use this pattern only when you genuinely cannot open more than one or two stations and you are content for the race to take that long.

Heats. Split the group in half. Heat one races the full station set while heat two does something else that needs no stations, typically the meal or a briefing. Then swap. Wall-clock time roughly doubles, station setup does not. This is the correct answer for 60 to 120 people in a venue that cannot hold ten teams at once.

Step 3: Build a collision-free rotation grid

With six teams and six stations, the assignment is a cyclic Latin square. Number the stations 1 to 6 and letter the teams A to F. Team A starts at station 1, team B at station 2, and so on. At each leg every team advances one station number, wrapping from 6 back to 1.

Read any column and a team visits all six stations exactly once. Read any row and all six stations are occupied by exactly one team. Collisions are impossible by construction, and no marshal has to think.

The obvious objection is that a carousel is not a race. If every team simply advances one number, the route is a conveyor belt and the sequencing decision the format is supposed to test has been designed out. That objection is correct and it is exactly what separates a good Amazing Race from a Mini Olympics wearing a costume. There are two ways to put the decision back.

Method A, the hidden Latin square. Keep the grid, but never let the teams see it. Each team receives its own clue pack, and the clues resolve to that team's own order. Teams do not know they are on a cycle; they know only that they solved a clue and it sent them somewhere. Zero collisions, zero queueing, and the felt experience is a race. What you lose is that the sequencing decision is simulated rather than real: the teams are solving, not choosing.

Method B, the open board with a one-team lock. Give every team a board showing all six stations at the start. Any team may go to any station it has not yet completed. A station holds exactly one team at a time. If a team arrives at an occupied station, it may not queue: it must divert to another station immediately, and the marshal records the diversion. Now the route is a genuine decision, made under pressure, with real consequences, which is the behaviour this format exists to produce. Add one live variable to sharpen it: at the start of each leg announce a "hot station" whose index score for that leg only is multiplied by 1.25. Six teams then have to decide whether to chase the multiplier and risk arriving to a locked door, or take the safe station nobody wants.

Method B is the better activity and the more demanding one to run. It needs the board, the lock rule, and a marshal or a phone at each station to show occupancy. Method A is the safer default for a first attempt, for very tight schedules, and for groups above eight teams.

Step 4: Set the leg clock

A leg is one complete cycle of travel, play and handover. Budget it in three parts.

• Travel and clue. 1 to 2 minutes inside a single venue. 3 to 8 minutes across a mall or a district. Longer than 10 minutes and you are running a walking tour with games attached, which is a legitimate format but a different one.

• Station play. 8 to 12 minutes. This is the number that has to be equal across stations, and it is the number people get wrong.

• Handover and scoring capture. 2 minutes. The score is written down, the next clue is issued, the clock for the next leg starts.

The leg clock is the longest of these three summed, and every station must fit inside it. Time each station in a dry run with five people who have not seen it. If one station overruns the others by four minutes, either shorten its task or accept that it sets the leg clock for all six, which costs you eight minutes of dead time per leg across the other five stations.

Step 5: Design the clue handoff

The handoff is the mechanical heart of the format and it must be identical at every station. Run it exactly like this.

1. The team arrives and hands its spent clue card to the station marshal, or scans the station's code with the race app. The arrival time is recorded.

2. The marshal reads the station brief aloud. It must be readable in under 60 seconds.

3. The team plays. The station produces a number.

4. The marshal records that number on the team's scorecard, or the team captain enters it into the app and the marshal confirms it.

5. Only then does the marshal issue the sealed envelope for the next clue.

6. The team opens the envelope away from the station and solves it.

Step five is load-bearing. If clues are issued before the score is recorded, teams sprint off and the score never gets written down, which is how a race ends with three missing numbers and an argument.

Clues need an escape valve, because a stuck team destroys the schedule. Publish a burn rule: any team may request the plain answer after three minutes of trying, at a fixed cost of 5 index points. That converts an unbounded schedule risk into a bounded scoring cost, and teams take the burn far less often than you expect, because 5 points is visible and pride is real.

Step 6: Write a station task that works

A station task is not a game. It is a scoring instrument that has to behave the same way six times in a row. Six requirements, and a task that misses any of them will cause you a problem.

1. It produces one number. Not a ranking, not a judgement, not a facilitator's impression. A number, in a stated unit, that a stranger reading the scorecard could reproduce.

2. A team of five completes it in the station play window including instruction, with no player idle for more than about two minutes.

3. The instruction reads in under 60 seconds. If it needs a demonstration, the demonstration is part of the 60 seconds.

4. Every player must contribute a scoring turn. Design the unit so that a team of five needs five turns. Five throws each, five shooting rounds, five frames, one lane each. This is the only reliable defence against two athletic people playing the whole race while three colleagues hold the bags.

5. Failure has a floor. A team that performs badly still records a number and still moves on. Never build a station a team can fail to complete, because that team then sits there and the grid breaks.

6. The set is competency-mixed. Across six stations use no more than two of any one type. A workable spread is two physical or dexterity stations, one accuracy or aim station, one cognitive or puzzle station, one reaction or speed station, and one station with a genuine chance element so the athletic team cannot sweep.

What headcount does it need, and how are teams formed?

Two teams of four, so eight participants, is the floor. The lowest published vendor minimum is 8 pax for the physical format and 5 for the virtual one. Below eight the format collapses into a scavenger hunt with a scoreboard, because with one team there is no race and with two teams the route decisions are trivial.

The comfortable band is 20 to 60 people. Published minimums sit at 20 and at 30. One operator states that its own best fit is groups below 300, and steers larger bookings onto free-for-all board formats instead. That is a candid piece of vendor advice: past a few hundred people, the clue-gated sequential race stops being the right instrument and a simultaneous point-board race takes over.

Team size should be four to six. Published vendor recommendations converge there: 4 to 6 is the common figure, given by one operator as the ideal mix of strategy, teamwork and fun, and the widest published band is 5 to 10, which is the outer edge. The reason six is the ceiling is capacity, not sociology: most station tasks physically seat four to six, and a seventh person becomes a spectator at every single stop for two hours.

Split teams deliberately, not by letting people self-select. Three rules that are worth the two minutes they cost:

• Mix departments. Teams that map onto real org units produce inter-departmental grudges rather than cross-departmental relationships, which is the opposite of the reason the budget was approved.

• Distribute the obvious athletes. One per team, named in advance by whoever knows the group.

• Name a rotating captain, one per leg, and make the captain the only person who may hand the clue card to the marshal. It forces five people to lead once each.

If people arrive late, add them to existing short teams. Never form a seventh team, because a seventh team in a six-station grid re-creates the queue you designed out.

How long does an Amazing Race take, and what can you cut?

A single leg is 11 to 15 minutes: 1 to 5 minutes travel and clue, 8 to 12 minutes play, 2 minutes handover.

A full event for 30 people over six stations is two hours, structured like this:

• Arrival, team formation, rules brief, safety brief: 15 minutes

• Six legs at 13 minutes: 78 minutes

• Schedule buffer: 7 minutes

• Score consolidation, results, awards, debrief: 20 minutes

That two-hour figure agrees with what the market publishes. Published durations cluster at 2 to 2.5 hours as the standard, with 2.5 to 3.5 hours quoted where briefing and debrief are counted in, 3 to 4 hours for the half-day version, and 2 to 5 or 2 to 6 hours for the longest island-wide formats. One operator notes candidly that only 50 to 60 percent of that window is actual gameplay, the rest being travel, briefing and transitions.

Compressing it. The floor is about 75 minutes: four stations, six-minute play windows, a single-venue route with one-minute travel, a five-minute brief and a ten-minute results segment. Below that you are cutting the debrief, and cutting the debrief converts a team building event into an afternoon out.

Extending it. Add stations before you add minutes per station. Ten to twelve stations over three hours holds attention far better than six stations at twenty minutes each, because the format's energy comes from the transitions. Beyond three hours, insert a meal break between two blocks of legs and treat it as two races with a combined total, and consider ending with a head-to-head final between the top two teams at a single station, which gives the event a visible finish that a spreadsheet cannot.

What equipment and what venue does it need?

Split the list into what you must have, what you can improvise, and what you must pay for.

Must have, and cannot improvise:

• A venue with at least as many distinct, scoreable activity points as you have teams. This is the real constraint and everything else is stationery.

• A scorecard per team, or a shared spreadsheet with the normalisation formula pre-entered.

• One person whose only job is scoring, who does not race.

• Sealed clue envelopes, one per team per leg, pre-sorted into team packs in leg order.

Can improvise in-house:

• Team identifiers. Coloured wristbands or lanyards work. Matching shirts are a photograph, not a requirement.

• The clues themselves. Written by someone in the company who knows both the venue and the internal jokes, they outperform any vendor's generic riddle pack.

• Station props for low-tech tasks. Buckets, rope, cups, marker pens, printed puzzles.

• Timing. Phones with a stopwatch, one per marshal.

Must be rented or bought, if the venue does not already supply it:

• Station equipment for anything physical. This is where in-house races blow their budget: six stations of hired props, plus transport, plus setup labour.

• Station marshals, one per station, for any task scored by a human. The published facilitator ratio in this market is one per 25 to 30 participants plus a separate MC and scoring coordinator, so a 30-person race under that model carries roughly three paid people minimum.

• A race app, if you want live leaderboards, GPS or code-based check-ins and photo evidence. Operators variously run missions through a WhatsApp bot, market an app-based edition, or supply a game handbook, a live dashboard and in-race support.

The venue question resolves to a single test: can you put one team at each of N points, with those points far enough apart that teams do not overhear each other and close enough that travel does not eat the leg? An office floor plate, a mall, a park, a heritage district and a self-scoring activity venue all pass. A single function room does not, and a group that ends up in a single function room has bought a Minute To Win It event by accident.

How do you score it so one station does not decide the result?

This is the section that most published material skips entirely, and it is where in-house races are actually lost.

Why raw addition destroys a race

Six stations produce six different units. A laser arena produces thousands of points. A basketball station produces thousands. A bowling station produces about a hundred pins. An archery station produces about a hundred and thirty. If you add the raw numbers, the two stations that happen to score in thousands contribute over ninety percent of the total and the other four are decorative. You will have set up, staffed and run four stations that could not affect the outcome.

The station index

Convert every station to the same band before adding anything. For each station, take the six team totals, find the lowest and the highest, and place each team inside that range:

Station index = 40 + 60 x (R - Rmin) / (Rmax - Rmin)

where R is the team's raw score at that station, Rmin is the lowest raw score at that station and Rmax the highest.

Every station now contributes between 40 and 100. Six stations give a maximum of 600. Three properties matter:

• The 40 is a participation floor. A team that has a disastrous station loses 60 points, not 3,000, so no single station can end the race in leg two and no team gives up.

• The 60 is the spread. Raise it to 70 and lower the floor to 30 if you want performance to matter more; the sum stays 100.

• The unit disappears. Pins, points, hits and seconds all become the same currency.

Two adjustments you will need:

Lower-is-better stations, such as lap times or completion times, invert the numerator: 40 + 60 x (Rmax - R) / (Rmax - Rmin).

All teams tied, where Rmax equals Rmin, would divide by zero. Award every team 70, the midpoint, and move on.

Weighting, if you deliberately want one station to matter more, multiply its index by a weight w and keep the weights summing to the number of stations, so the maximum total does not move. Weights outside 0.75 to 1.5 make the race feel rigged.

The worked example: 30 people, 6 teams, 2 hours, 6 stations

Six teams, A to F, on the rotation grid above. The stations and their units, with illustrative raw team totals:

Take S3 as the walk-through. The lowest is 180 and the highest 305, so the range is 125.

• Team C scored 180, the lowest, so its index is 40.0.

• Team E scored 305, the highest, so its index is 100.0.

• Team A scored 210. That is 30 above the floor, 30 divided by 125 is 0.24, times 60 is 14.4, plus 40 gives 54.4.

• Team B scored 275. That is 95 above the floor, 95 divided by 125 is 0.76, times 60 is 45.6, plus 40 gives 85.6.

• Team D scored 240, giving 68.8. Team F scored 195, giving 47.2.

Apply the same three lines to all six stations and the table becomes:

Final standings: B 423.9, D 419.4, E 416.4, C 405.5, A 398.8, F 394.7.

Now look at what the index did. Team C won two stations outright with perfect 100s and finished fourth, because it also floored three times. Team B never won a single station and won the race, because its worst result was one 40 and everything else sat above 56. That is the format telling the truth about consistency, and it is a far better story at the awards than "the team with the ex-national bowler won".

Compare against raw addition on the same numbers. Raw totals are C 17,438; F 16,835; A 16,694; B 15,745; D 15,707; E 15,564. The raw winner is C and the raw last place is E, which the index places fourth and third respectively. The ranking is almost exactly inverted. The reason is arithmetic: across all six teams the laser arena contributed 50,400 of the 97,983 raw points in the race, or 51.4 percent, while the bowling station contributed 0.58 percent, archery 0.81 percent and throwing 1.43 percent. Three of your six stations were decoration. The index costs one spreadsheet column and fixes it completely.

Penalties

Publish these before the start, in writing, or do not apply them.

The safety penalty is the only one that should never be negotiated or waived on the day, and saying so in the brief is what makes the other five stick.

Tie-breaks

Apply in order, and publish the order in advance:

1. Consistency. The team with more stations scored above 70 wins.

2. Head-to-head. If both teams were at the same station in the same leg at any point, the higher raw score there wins.

3. Elapsed time. Lowest total time from first clue opened to final scorecard signed.

4. Sudden death. One nominated player per team, one throw each at the throwing station, highest ring wins. Repeat until it breaks.

Two format variants

Points race (the default). Everything above. Time enters only as a penalty for overrun and as a tie-break. It is the fairer format, it produces closer results, and it suits mixed-fitness office groups.

Time race. Every station has a clear condition rather than a score: sink four baskets, clear 1,000 points, land three arrows in the mid ring. The clock runs from start to finish and the fastest team to complete all six wins. This is closer to the television format and it is more exciting to watch, but it has one serious flaw for a company: a team that cannot clear a station is stuck there, and stuck means a person is standing in a corner being watched by their colleagues for twenty minutes. If you run the time race, publish a mercy rule (after six minutes at a station the team may take a two-minute penalty and move on) and staff it properly.

Stopping one or two people from carrying the team

Three mechanisms, in increasing strength:

• Mandatory turns. Every player takes at least one scoring turn at every station, enforced by the marshal. This is the baseline and it should always be on.

• Per-player caps. No player may contribute more than 30 percent of the team's raw score at a station. Where the station scores per player this is easy to check; where it scores per team it is not, so use it selectively.

• Rotating specialist lock. Each player is nominated as the lead at exactly one station, decided at the start before the stations are known. Their score at that station counts double and nobody else may lead there. It forces a genuine allocation decision under uncertainty, which is the best single conversation this format produces.

How do you make an Amazing Race harder or easier?

Change one rule at a time. Every tier below shifts two or three rules relative to the one before it.

Base version. Hidden Latin square rotation, so no route decisions. Clues are photographs of the station, solvable in seconds. Free burns. All stations equal weight. Ten-minute play windows. Penalties limited to the safety rule. Use this for a group that has never done anything like it, for a group with mixed mobility, and for any race where the priority is that everyone finishes smiling.

Advanced version. Open board with the one-team station lock. Clues become riddles with a 5-point burn after three minutes. One hot station per leg at 1.25x, announced when the leg starts. Mandatory turns enforced. Nine-minute play windows with the overrun penalty live. This is the version to run for a normal office group that has done team building before, and it is the version this page is designed around.

Hell version. Open board, no repeats, and three additions. First, chained clues: the clue you receive at one station only resolves if you have already visited another, so route order genuinely determines whether you can proceed. Second, a hidden seventh station named in no clue, whose location must be deduced from the six clue texts, carrying a 1.5x weight. Third, a hard leg clock with no buffer and the full overrun penalty. Optionally allow a team to claim a station it does not intend to play, locking rivals out for a leg at the cost of taking the 40 floor there. Be careful with that last one: it is the only rule in this document that reliably produces genuine bad feeling, and it belongs only in a group that already competes happily.

What goes wrong on race day, and how do you stop it?

Queues at stations. Cause: fewer stations than teams, or one station that runs four minutes longer than the rest. Fix: stations at least equal to teams, and dry-run every station with five people who have not seen it, timing to the second. Shorten the slowest task rather than lengthening the leg.

Raw score addition. Cause: nobody thought about units until the last ten minutes. Fix: build the normalisation spreadsheet before the day, with the formula in the cells and the raw columns empty. It takes fifteen minutes to build and it decides who wins.

Two people play the whole race. Cause: no mandatory-turn rule, plus stations that reward one athletic person. Fix: mandatory turns, rotating captains, and at least two stations where physical strength is irrelevant.

Scoring collapses at the finish. Cause: thirty-six numbers arriving at one laptop in the last five minutes, some of them illegible. Fix: capture per leg, not at the end. A scorer who does not race, collecting six numbers every thirteen minutes, is calm. The same person collecting thirty-six numbers at once is not, and the awards run twenty minutes late while everyone stands about.

Clues too hard, or too easy. Too hard and teams stall and the grid breaks. Too easy and you have built a carousel. Fix: the three-minute burn valve caps the downside, and testing every clue on one person who does not know the venue caps the upside.

Teams that mirror real departments. Cause: letting people self-select. Fix: pre-assign, mix functions, and say why you did in the brief.

Injuries and attire. Cause: office clothing, lanyards, hard floors. Fix: state footwear and jewellery requirements in the calendar invitation, not on the day, and screen anyone with knee, back, heart or balance conditions out of the physical stations rather than out of the event.

The debrief gets cut. Cause: the race overruns and the room is booked until five. Fix: protect the last twenty minutes in the schedule before you decide the number of stations, and cut a station rather than the debrief.

What does it actually train, and what does it not?

Two things, and be specific about the mechanism rather than listing virtues.

Decision-making under time pressure. The mechanism is the open board. A team standing between two unvisited stations, one of which carries a multiplier and might be occupied, has to make a real choice with an unrecoverable cost, in under thirty seconds, with five opinions in the room and a clock running. That happens five times in a two-hour race. Very few workplace exercises produce five genuine, costly, irreversible group decisions in two hours with immediate feedback on each, which is why this format survives.

Explicit role allocation. The mechanism is the station capacity cap. When a station physically seats four and the team has five, somebody sits out, and the team has to say out loud who and why. When a station rewards accuracy and another rewards speed, the team has to name who is good at what, in front of that person. Teams that have never allocated roles explicitly discover in leg two that they have been doing it implicitly and badly.

What it does not train. Trust, in any meaningful sense. Trust requires that you depend on someone else's honesty or care when you cannot verify it, and a race with automated scoring and marshals verifies everything. It does not train deep listening: the clock rewards whoever speaks first and loudest, which is often the opposite of what a quiet team needs. It does not train conflict resolution, because there is no time to resolve anything and the round simply ends. And it does not train cross-team collaboration; it is explicitly competitive between teams, so if your actual problem is that two departments do not talk to each other, this format can make it worse rather than better. For that problem you want a collaborative canvas, a build-and-handover format, or something with a shared rather than divided objective.

It also does not fix a team. It exposes one. A team that argues badly at station one will argue badly at station six, louder. That is diagnostically useful for a manager who is paying attention, and it is worthless to a manager who books the race hoping the problem will dissolve.

Which debrief questions are worth asking after a race?

Run fifteen to twenty minutes, immediately after the results, before people scatter. The results screen should still be visible, because the normalisation table is itself a debrief instrument.

Ask teams to answer in their teams first, five minutes, then take two or three answers to the room:

1. Which station did you choose first, and would you choose it first again? What did you actually know at that moment?

2. Who decided the order? Did anyone disagree at the time without saying so, and what stopped them saying it?

3. Name the leg where you lost the most time. Was it the clue, the walk or the station itself? Be precise.

4. At which station did you know you were losing, and what did you change in response? If you changed nothing, why not?

5. Did you play to your strongest person at every station? What did that cost the people who did not play?

6. Who read the station instructions? Did anyone else read them, or did five people take one person's word for the rules?

7. Look at the index table. Point to the station where your raw score was best and your index was worst. What does the gap between those two things tell you about how you were measuring yourselves during the race?

8. Which teammate did something useful that nobody thanked them for at the time?

9. If we ran the same six stations tomorrow with the same teams, what single change would move you up two places? Not three changes. One.

10. What did you do in the last two hours that you would not do in a Monday planning meeting? What is the difference between the two rooms?

Protect question 7 above the others. It is the only question in the list that only a normalised race can ask, and it reliably lands, because a team that scored highest in raw points at a station and still came fourth has to confront the fact that it was optimising the wrong number.

What else is an Amazing Race called, and what is it not?

Singapore operators sell this format under a long list of names, and every one of them should be read as a badge on the same underlying product when you compare quotes. Counting only what local operators publish, the same route-based race is sold under at least a dozen distinct product names, most of them coined by the operator selling them.

The pattern is easy to recognise once you look for it. Most names are the words Amazing Race with a qualifier bolted on: a company prefix, or Singapore, Style, App-Based, Hybrid, Virtual, World Edition, Around The World, Indoor or Outdoor. A few drop the words entirely and become a hunt of some kind. A further layer is location branding, where the identical design is renamed after the site it runs on, from Gardens by the Bay to Pulau Ubin to a bird park. On top of that, some operators give individual routes proprietary edition names, which describe a fixed route around Chinatown, Fort Siloso, HarbourFront, the Singapore River or 313@somerset rather than a different format.

None of that helps you compare quotes. For this family the single most revealing thing to ask is whether the clue order is locked by the organiser or picked by each team on the day, since a free-choice build with play currency layered on top is a different game with a different debrief. The wider set of questions that pins a supplier down on what is actually being sold has its own treatment, in our guide to the names these activities carry.

Now the distinctions that matter when you are choosing, because several adjacent formats are quoted against this one.

Mini Olympics, or telematch. Parallel, not route-based. Every station opens at once, teams are assigned to stations by a marshal and rotate on a whistle, and there are no clues because there is nothing to unlock. The design work there is in choosing a balanced event mix and weighting a final event. The design work here is in the route and the sequencing. If the quote you are reading says "rotate on the whistle", you are buying a Mini Olympics.

Running Man. Elimination-based. The core mechanic is name tags, hidden identities and betrayal, and players leave the game permanently. There is no fixed route, and its hardest design problem is finding something for knocked-out players to do across the rest of the session. Nothing in this page applies to it.

Minute To Win It. Sixty-second units in a single room with household props. No movement, no route, no clues. It is the format you end up with by accident if you book an Amazing Race into a function room.

Squid Game. Themed elimination with a fixed sequence that every player experiences in the same order. The design problem is the order of the rounds and the handling of the eliminated. The sequence is fixed by the organiser, not chosen by the teams, which is the exact opposite of this format.

Free-for-all board races. Operators distinguish these from the route-based race explicitly, and the distinction is worth borrowing: a raceboard shows every station and its point value from the start, teams choose freely, and nothing is clue-gated. It scales past 300 people where the sequential race does not.

Treasure hunt or scavenger hunt. Collection-based. Teams gather items, photographs or answers, and there is often no scored task at a fixed point. Many Singapore vendors use "treasure hunt" and "Amazing Race" interchangeably in their marketing, so read the mechanics paragraph rather than the heading.

Running an Amazing Race at PLAYON

PLAYON is a single indoor venue with 26 separately scored attractions, which makes it a route-based race that already exists. Booking the whole venue does not give you a room in which to build a race; it gives you the finished station set.

The practical argument

A conventional Amazing Race carries three costs that have nothing to do with the participants' experience.

There is no rig to build. A hired station is a table, a set of props, a rules card and forty minutes of setup before the group arrives, multiplied by the number of stations. Here the stations are permanent installations that were built to be played by strangers with no instruction.

There are no props to hire or transport. Six stations of hired equipment, delivered, set up, staffed and collected, is a large fraction of a mid-range per-head quote. The bows, the balls, the pins, the laser guns and the motion rigs are already in the building.

Most stations do not need a marshal, because they score themselves. This is the substantive point. Colorful Grid registers each tile step and posts to a live board with no human judge, applying its own deductions for wrong tiles and timeouts and running its own streak multiplier. Five-Screen Basketball tracks each of its five lanes on its own overhead screen. Bowling has automatic pin reset and a dual score display. Archery uses digital targets that pick up exactly where each arrow lands and score it by hit zone. King of Guns detects infrared hits and locks the score at the buzzer. Rubber Ball Shooting posts the total instantly when the timer expires. Reaction Testing registers each catch. Hyper-Drift Arena posts lap times live across up to six linked rigs. Motion-Sensing Game and Body-Sensing Arena read a real-time skeleton map and score accuracy and timing automatically. The published market ratio is one facilitator per 25 to 30 participants plus an MC and a scoring coordinator; at a venue where the stations judge themselves, most of that headcount is doing something the machines already do.

A six-station set for 30 people

Six teams of five, six stations, the rotation grid above, thirteen-minute legs.

Substitutes when one of the above does not suit the group: Reaction Testing (automated, head-to-head, catch tally, no deductions), Rubber Ball Shooting (timed, automated, single station so it runs turn-based with a colour key that changes each round), Archery (real bow, rubber-tipped arrows, digital hit-zone scoring, but published at up to 4 in rotation, so a team of five needs the fifth player to sit a round or the round to be split), VR Shooting (automated, shared arena map), VR Large Space (1 to 6 free roam, missions scale with group size, but the modes are cooperative rather than competitive and the scoring method is not published, so it works as a between-legs interlude rather than a scored station), and the two camera-tracked stations, Motion-Sensing Game and Body-Sensing Arena.

Two capacity notes worth planning around. Five-Screen Basketball caps at precisely five players in a round, matching a five-person team exactly and is a good reason to build teams of five rather than six. Hyper-Drift Arena publishes a strictly enforced 120cm minimum height for harness fit and is not recommended for anyone who is pregnant or has heart, neck or back conditions or motion sickness, so it needs a named alternative for anyone screened out, not an apology on the day.

Attire is not a formality. Colorful Grid, Reaction Testing, Motion-Sensing Game and Body-Sensing Arena all require grip socks and prohibit bare feet, and several stations require rings, watches, lanyards and bracelets to be removed. Put that in the calendar invitation.

What still costs you work, honestly

The scoring is automated per attraction and nowhere else. Every leaderboard in the building ranks players at that attraction. Nothing aggregates six attractions into one team total, and nothing normalises across them. Somebody has to walk the six numbers per leg into a spreadsheet and run the index formula. For the worked example that is thirty-six numbers over six legs, which is roughly twenty minutes of a nominated scorer's attention spread across two hours. It is small and it is not zero, and a race that does not plan for it ends with a leaderboard nobody can read.

One station in that set needs a person. Throwing Competition has no automated scoring described: staff select the mode and scores are totalled at the end of the round. That is one tally sheet and one attentive human, not six.

The route element is compressed. This is a single floor. Travel between stations is sixty to ninety seconds, not the eight minutes an island-wide race gives you, which means the navigational half of the format is largely absent. That has one real consequence for the design: run the open board variant rather than the hidden Latin square, because when travel is cheap the only decision left is which station to claim, and you need that decision to carry weight. Enforce the one-team station lock strictly, use the hot-station multiplier every leg, and put the difficulty into the clue rather than the walk. A venue race is a sequencing race, not a navigation race, and it is better to design for that than to pretend otherwise.

What this venue is not right for

Say it plainly, because HR briefs frequently contain one of these requirements and the venue cannot meet it.

• Heritage, culture or city-discovery races. If the brief says the team should learn something about Singapore or see a part of it they have not seen, this is the wrong building. Operators build exactly that at Chinatown, Fort Siloso, HarbourFront, the Singapore River, Pulau Ubin and Gardens by the Bay.

• CSR-linked races, where checkpoint tasks tie to a donation, a build or a community interaction. Several Singapore operators build those; a self-scoring indoor venue cannot.

• Races above roughly 60 people at once. Twenty-six attractions is a lot, but concurrency at most of them is not published and several are capped low. Above sixty, either run heats or take the group to an operator built for scale: published capacities reach 2,000 and 5,000 participants, and large groups get moved onto free-for-all board formats deliberately.

• Outdoor, endurance or transport-based races. If the point is that the team walks eight kilometres and takes the MRT four times, book the outdoors.

If, after all that, the brief actually says station rotation, scored tasks, a clear winner, two hours, indoors, weather-proof and thirty people, then the argument for a venue like this is simply that the stations are already built, already scored and already tested on thousands of strangers, and the only thing left for the organiser to build is the part that was always the interesting part: the route, the clues and the scoring.

Amazing Race questions HR teams ask

How much does an Amazing Race cost per person in Singapore? Published rates run from $25 to $130 per person. The lowest published rates start near $28, a common mid band is $40 to $60, and $130 per person appears at 20 to 99 pax. Minimum event fees apply below roughly 30 people.

How many stations do I need for 30 people? Six. Run six teams of five and six stations so every team is at a different station at all times, which removes queues entirely. Fewer stations than teams guarantees waiting, and waiting is what participants remember.

How long should an Amazing Race run? Two hours for 30 people across six stations. Singapore vendors publish 2 to 2.5 hours as standard, 3 to 4 hours for half-day formats, and up to 6 hours for island-wide routes. Below 75 minutes the format stops working.

How do I stop one station from deciding the whole result? Normalise every station to the same 0 to 100 band before adding. Convert each team's raw score to 40 plus 60 times its position in that station's range, so a laser arena scoring in thousands weighs exactly as much as bowling pins.

What is the difference between an Amazing Race and a Mini Olympics? Route. An Amazing Race gives each team a clue that sends it to one station at a time, so sequencing is part of the game. A Mini Olympics opens every station at once and assigns teams to them.

How big should each team be? Four to six people. Singapore vendors converge on this: published recommendations are mostly 4 to 6, with 5 to 10 the widest band found. Above six, someone stands and watches at every station, which is the main cause of disengagement.

Do I need staggered starts? Only when you have more teams than stations. With equal numbers, start every team simultaneously at a different station and rotate. Staggering teams down one shared route creates a queue that compounds at every leg and never clears.

What happens if two teams tie? Break it on consistency first: the team with more stations scored above 70 on the index wins. If still tied, compare total elapsed leg time, then run a single sudden-death throw with one nominated player from each team.

Can an Amazing Race run entirely indoors? Yes. Singapore operators run mall editions at 313@somerset and HarbourFront, and venue-based races work in any building with enough separately scored activity points. The trade-off is that travel between stations shrinks to a minute, so route choice matters less.

How many facilitators does an Amazing Race need? Published vendor ratios are one facilitator per 25 to 30 participants plus an MC and a scoring coordinator. At venues where the attractions score themselves, you need only a scorer and one marshal for any station scored by hand.

Does an Amazing Race build trust? No. It builds decision-making under time pressure and forces explicit role allocation, because station caps make teams choose out loud who plays. Trust needs vulnerability and dependence on someone else's honesty, and a scored race supplies neither.

What is the minimum group size? Eight people, two teams of four, is the smallest that works, and the lowest published vendor minimum is also 8 pax. Below 20 the per-head cost rises sharply because vendors apply minimum event fees from $2,000.

Sources

• PLAYON attraction pages: equipment, player steps, scoring, automation, concurrency, age, height and attire rules for Colorful Grid, Reaction Testing, Throwing Competition, Rubber Ball Shooting, Five-Screen Basketball, Bowling, Archery, King of Guns, VR Shooting, VR Large Space, Hyper-Drift Arena, Motion-Sensing Game and Body-Sensing Arena (playon.com.sg)