Build and Race Challenges

Build It, Then Test It: Vehicle Build Challenges in Singapore

In Singapore the build-and-race format is sold almost entirely at the two extremes, and the middle, which is where the useful version lives, is barely offered at all. At one end are the flat-pack kit builds, where teams assemble a near full-size car from pre-cut parts, decorate it, and push it down a hotel ballroom. At the other end sit the cardboard boat builds, where teams get sheets of cardboard and packing tape and one hour, and then paddle the result at Sentosa. Roughly six operators in this market list something in this family. Between the two extremes there is a version that is far better for a working team and that almost nobody packages: a constrained design exercise with a token budget, a parts market, a weight limit and a test that measures more than one thing.

The page below is laid out to let an HR or admin lead specify and run that middle version themselves, or read a quotation closely enough to see which of the two extremes they are actually being sold. The construction is the easy part. The design of the constraints and the design of the test are where these sessions are won or lost, and neither of those things is usually in the brochure.

Vehicle building as a Singapore corporate format, and the shape of the local offer

A build-and-race is a timed making task in which every team assembles a moving object from supplied parts or raw materials inside a fixed window, and the object is then run against a defined test rather than merely displayed. The vehicle can be a wheeled cart, a small car, a cardboard boat, a balloon or elastic driven model, or a full-size push kart. What makes it a distinct format is the second half: something the teams built has to perform, in front of everyone, against a measurement.

The published local and regional listings cluster into three shapes.

The flat-pack kit build. Teams receive a kit of pre-cut parts and assemble a near full-size racing car, then decorate it and take it through a push race and a podium ceremony. Published group bands for this shape start at 26 to 50 people and step up through 51 to 150, then 151 to 250, then 251 to 500, then anything above that. Published durations are offered as 2.5 hours, 3 hours or 3.5 hours, indoors or outdoors. Per pax rates are not published on the product pages.

The cardboard boat build. Teams receive cardboard sheets and packing tape only, get roughly one hour to build, present their boat, and then race it on water. Published group size for the local version runs from 10 to 500 participants, with a published duration of up to 3 hours, outdoors, with the materials described as deliberately limited. Per pax rates are not published.

The small-scale model build. Rubber band cars, mousetrap cars, balloon-powered buggies and similar desktop-scale builds. These turn up inside mixed activity packages and station rotations rather than as a named product, so published headline figures for them are thin.

Per person pricing for facilitated build formats is generally not published at the product page level in this market. The published general bands for this market run as follows. $25 to $40 a head buys one facilitated activity inside your own office with no food. $45 to $90 a head buys a standard half day at an outside venue, facilitator and light refreshments included. $100 to $180 a head buys a premium full day with catering and a managed room. $200 to $250 and upwards buys a retreat or residential format. A supplied-kit build-and-race with facilitators and a scored finale sits in the standard half day band for most groups. Published loadings are 15 to 25 per cent at weekends and around 30 per cent on public holidays, with bundled catering carrying a published markup of 30 to 40 per cent, and a managed venue adding a published $10 to $40 per pax.

The heat in this category is real but shallow. Several suppliers carry a build-and-race product, they carry very similar products, and the differentiation is almost entirely in the theming rather than in the design of the constraints or the test. That is the gap this page is written into.

Supplied kit or raw materials, and what each version quietly buys you

This is the first decision and it changes the entire day. It is not a matter of budget. The two versions test different things and fail in different ways.

The supplied kit version. Every team receives an identical box of defined parts: a chassis panel, a set of wheels, axles, a drive element, fixings. The design space is bounded. Teams decide how to assemble, where to place mass, how to align the axles, what to leave out. Because every team has the same parts, the comparison at the end is clean, the test result is defensible, and nobody can claim the other side had an unfair advantage.

What a kit gives you is repeatability and fairness. What it costs you is the interesting part of the problem. When the parts are pre-cut and the shape is implied, most teams converge on more or less the same vehicle, and the session becomes an assembly race with a decoration phase. That is not worthless, because assembly under a deadline still exposes coordination, but it is a much smaller exercise than it looks in the photographs.

The raw materials version. Teams receive stock: sheet material, rod, tape, wheels of assorted sizes, string, containers, ballast. There is no implied design. A team has to decide what a vehicle even is inside the rules you wrote, and two teams in the same room will arrive at genuinely different objects.

What raw materials give you is a real design decision, which is the whole reason to run this format rather than something cheaper. What it costs you is judgeability. Two vehicles built to two different interpretations of the same brief are hard to compare, and if your rules leave gaps, the argument at the end will be about the rules rather than about the vehicles. That problem has a solution and it has a section of its own further down, but you have to design it in beforehand rather than referee it afterwards.

A third option, which is the one to reach for. Run a defined parts market rather than either extreme. There is a published price list of parts, teams have a token budget, and every team can buy any part. The parts are standard, so measurement stays fair, but the selection is open, so the design decision survives. This gets you most of the raw materials exercise with most of the kit version's defensibility, and it is the shape the rest of this page assumes.

How to choose:

• Choose the kit if the group is large, the room is mixed in seniority, the facilitation is thin, or the session has to be visually spectacular for a photographer.

• Choose raw materials if the group is small, technically confident, and the point of the day is to watch a team make a difficult trade-off in public.

• Choose the parts market if you want the design decision but need the result to hold up when the losing team asks how it was scored.

The constraint set that turns assembly into design

Handing a team a box and a deadline produces assembly. What produces design is scarcity, and scarcity has to be engineered. Four constraints do almost all of the work, and they compound: each one on its own is mild, and together they force a genuine argument inside every team.

A token budget. Every team starts with the same number of tokens, and every part has a price. A hundred tokens is a good scale because it makes percentages readable. The budget must be tight enough that no team can buy everything they want, which in practice means pricing a full specification vehicle at roughly 140 per cent of the budget. The budget converts a shopping trip into a set of trade-offs, and it lets you award points for tokens left unspent, which is the clause that stops teams simply buying the maximum of everything.

A parts market where the shelf is finite. Do not simply publish a price list and hand out unlimited stock. Put a limited number of each part on the table: nine sets of the good wheels for five teams, four drive units, two of the light chassis panels. Teams now have to decide fast, and they have to trade with each other, because the team holding the second light chassis panel does not need both. Trading between teams is legal and should be stated as legal in the brief. What you get from a finite shelf is a decision made under time pressure with incomplete information, which is the behaviour that is hardest to produce in an office activity and the easiest to observe here.

A weight limit, or a mass floor, or both. A ceiling on the empty weight of the vehicle stops the team that would otherwise solve everything by building a tank. A floor on the weight stops the team that would otherwise solve everything by building nothing. Publishing both, for example an empty vehicle between 400 grams and 1.2 kilograms, closes off both degenerate strategies and forces the design into the interesting middle. Weigh at scrutineering, on one set of scales, in front of everyone.

A payload requirement. The vehicle must carry something: a 2 kilogram load, a full open cup of water, an egg, a stack of files. The payload is the single most valuable constraint available, because it directly attacks the speed-only strategy. A vehicle that is fast because it weighs nothing cannot carry two kilograms without folding, and the team that spends the first twenty minutes making the lightest possible chassis discovers this at the test lane with no time left to change it.

Two optional constraints worth knowing about:

• A banned part. Name one obvious solution and forbid it: no pre-made wheels, or no adhesive, or no part longer than 30 centimetres. It forces improvisation and it makes the design phase noticeably longer, so add time if you add this.

• A design lock. Teams must submit a drawing and a parts order before any material is issued, and after that they may not buy anything else. This is the most aggressive constraint on the list and the one that most reliably produces a real argument, because a team has to commit to a design before they have touched a single part.

Step by step, from the parts market to the final classification

What follows is the running order for a do-it-yourself session built on the parts market version. It assumes a two hour block, five teams and one facilitator with a helper.

1. Set up the test lane before anyone arrives. Tape a start line, a 10 metre run, a finish line and a 30 centimetre stopping box, and place the ramp if you are using one. Teams must be able to see the test before they design for it. A team that has watched the lane for two hours designs better than a team shown it at minute ninety.

2. Hand each team the brief and the scoresheet in the first ten minutes, in writing, one copy per team. The brief covers what counts as a vehicle, the weight window, the payload, the parts price list and the four scored tests with their weights. Do not describe the scoring verbally and hope. Scoring only changes behaviour if teams can read it while they design.

3. Run a design lock block of fifteen minutes with no parts on the table. Teams sketch on paper and write a parts order against their token budget. Nothing may be bought until the order is submitted. This block is the reason the exercise is a design exercise, and it is the first thing every organiser cuts when running late. Do not cut it.

4. Open the parts market for ten minutes. Teams present their order at the shelf and take what is available. When something runs out, it is out, and the affected team has to re-plan at the table. Announce clearly that teams may trade parts with each other at any point for the rest of the session.

5. Open the build window of thirty-five to forty-five minutes. Call the time remaining at the halfway point, at fifteen minutes, at five and at one. Build teams lose the clock completely and there is a section below on why.

6. Open a practice window of ten minutes with a booked slot per team. One team on the lane at a time, two runs each. This is where teams discover that the vehicle pulls left. Book the slots and post the order, because an unbooked practice window turns into a queue and the last team gets nothing.

7. Close the build and run scrutineering, five minutes. Every vehicle is weighed and measured against the published gates by one person with one instrument, in view of the room. Pass or fail, no discussion. A vehicle that fails a gate may be corrected during scrutineering only if the published rules say so, and the rules should say so, because disqualifying a team at minute ninety-five ruins their afternoon and teaches nothing.

8. Run the four tests in the published order, twenty minutes. Everyone watches every run. Announce each result out loud as it happens and write it on a visible scoreboard.

9. Read the classification, five minutes. Read the component scores before the totals, so the room can see where the points actually came from. This is the moment the multi-criteria design earns its keep.

10. Debrief for at least ten minutes, in the room, before anyone leaves.

Why a straight race is the worst test you can run

Here is the failure that ruins most of these sessions, and it happens at the very end where it cannot be fixed.

If the only test is a race down a flat lane, the optimal strategy is to make the lightest thing that rolls. That is not an interesting design problem, it has one right answer, and every experienced person in the room knows the answer within about ninety seconds. The teams that work it out build a bare chassis and two axles and win. The teams that build something considered and robust lose. The lesson the room takes home is that thinking about the problem was punished, which is precisely the opposite of what you booked the afternoon for.

A single-criterion test has three further problems worth naming.

It collapses the design phase. When there is one thing to optimise, there is nothing to argue about, and the design conversation that you are actually trying to produce lasts under two minutes. The team goes straight to building.

It makes the result a coin toss. Five vehicles down a 10 metre lane will finish within a couple of seconds of each other, and the winner is frequently decided by which one was pushed straightest. Nobody believes that result, including the team that won.

It gives you nothing to debrief. Ask a team why they won a straight race and the honest answer is that their thing was lighter. Ask them why they came second in speed but first overall and you get a real answer about a decision they made at minute twenty.

The fix is not to make the race harder. It is to test more than one property, weight the properties so that no single one can carry a team, and publish the weights before the design phase starts.

A four part test, and the weights that stop the lightest thing winning

This is the recommended test design. It measures four properties that genuinely conflict, so no vehicle can be optimal at all of them, and every team has to pick a position. Total is 100 points.

Speed on the flat, 20 points. A timed run over 10 metres, launched by hand from behind the start line with no running push, or released from a fixed ramp if the vehicle is unpowered. Points by finishing order: 20, 16, 12, 8, 4. Two runs, best time counts.

The payload run, 25 points. The same 10 metres carrying a 2 kilogram load that must be attached or contained by the vehicle, not held by a person. Completing the run intact with the load still aboard is worth 15 points flat. On top of that, the completing teams are ranked by time for a further 10, 8, 6, 4 or 2 points. A team that fails to complete this scores zero on the whole criterion. This is the criterion that kills the featherweight strategy, and 25 points is the right weight for it.

The coast test, 15 points. The vehicle is released, unpowered, from a fixed ramp of roughly one metre run and 20 centimetres of drop, and the distance travelled before it stops is measured. This rewards alignment, low friction and straight running, which are the invisible qualities of a well built vehicle. Points by distance order: 15, 12, 9, 6, 3.

Reliability, 25 points. Three consecutive runs of the payload course with no repair, no re-taping and no part replacement between them. A facilitator watches the vehicle between runs and nobody touches it except to reload the payload. Each clean run is worth 8 points, with 1 bonus point for completing all three, so the achievable scores are 0, 8, 16 or 25. This is the criterion that punishes the beautiful thing held together with hope, and teams consistently underestimate it.

Budget remaining, 10 points. One point per unspent token, up to 10. This is a small weight and it must stay small, but it is what stops every team simply spending everything on the strongest possible parts and turns the token budget into a live consideration during the market.

Brief compliance and design, 5 points, judged. Neatness, use of parts, whether the vehicle does what the team said it would. Keep this to 5 per cent. Judged points are where disputes live, so the smaller the judged share, the less arguing there is at the end.

Why these four properties in particular. Speed and payload pull directly against each other, because mass helps one and hurts the other. Coast distance rewards precision, which costs build time that could have gone into strength. Reliability rewards over-engineering, which costs speed and tokens. There is no vehicle that wins all four, so every team has to choose a shape and defend it, which is exactly the conversation you want to be watching.

The order of the tests matters. Run speed first, because it is the one everyone expects and it gets the room engaged. Run payload second, because that is where the featherweight builds fail publicly and the room understands the format. Run coast third. Run reliability last, because it is the slowest and by then everyone is invested. Announce running totals only after all four, not between rounds, or the teams behind will stop trying.

Where the two hours actually go, and the overrun you should plan for

Build time and test time are two different budgets and organisers publish only the first.

Build time. Thirty-five to forty-five minutes is the working range for a desktop-scale vehicle with a parts market in front of it. Under thirty minutes you get a chassis with wheels taped on and nothing else. Past sixty minutes the improvement stops and the fiddling starts, and last-minute fiddling in the shadow of scrutineering makes a vehicle less reliable, not more. For a full-size kit build, published durations of 2.5 to 3.5 hours reflect a longer assembly window and a decoration phase, not a longer design phase.

Design time. Fifteen minutes, before any parts are issued, and it is a separate block. Teams that design while building do not design, they build and then rationalise. If you take one recommendation from this page, take this one.

Market time. Ten minutes. Long enough to force a decision, short enough that a team cannot deliberate its way out of the scarcity.

Practice time. Ten minutes with booked slots. This is not optional. A vehicle that has never been on the lane will do something unexpected during the scored run, and the team will feel cheated by a test they were never allowed to see.

Test time. Twenty minutes for four tests and five teams, and this is the number every organiser gets wrong. Count it properly: the speed test is five teams times two runs, the payload test is five teams times one run plus reloading, the coast test is five teams times two runs, and reliability is five teams times three runs. That is roughly fifty individual runs. At twenty seconds of running and resetting each, plus the transitions between tests, twenty minutes is tight and twenty-five is honest. If you have eight teams rather than five, the test block roughly doubles while the build block does not change at all, which is the single most useful piece of scheduling arithmetic in this format.

Why teams always overrun the build

They overrun for a structural reason, not because they are careless. A build has no natural stopping point. There is always one more brace to add, one more strip of tape, one more adjustment to the axle. The vehicle is never finished, it is only taken away. Compare that with a quiz round or a timed challenge, where the end is defined by the task itself.

Three things reliably reduce the overrun:

• A hard tools-down at a stated minute, called twice. Say the number at the start, call it at five minutes out, and enforce it. Enforcing it once, visibly, on the first team that tries to keep working, buys you compliance for the rest of the day.

• Practice slots that are booked before the build ends. A team with a slot at minute eighty stops building at minute seventy-eight of their own accord, because they want to use it. This converts your deadline from an imposition into an opportunity, and it works far better than shouting.

• A published rule that anything added after tools-down is removed at scrutineering.State it, then apply it. It is a cleaner conversation than an argument about whether the tape was already there.

Do not solve the overrun by extending the build, because the time comes out of the test, and the test is the part with the learning in it. If you are running short, cut five minutes of build, never five minutes of testing.

Team size, team count, and the ceiling the test lane imposes

Team size. Five to six people. Four is workable but thin, because the roles below start to double up. Seven or more and you have guaranteed spectators, because a desktop-scale vehicle physically has room for two pairs of hands and no more.

Team count. Four to six teams for a single lane and a single facilitator. Every additional team adds roughly ten individual runs to the test block, and the test block is what constrains the session, not the build. Above six teams, either add a second identical lane with a second timekeeper, or split into two heats and accept that the room watches half of the runs.

Total headcount. Twenty to thirty-six for a single-lane session. Published listings for the large kit versions run to 500 and above, which necessarily means many parallel builds and either a very long staggered race or a judged parade instead of a test. That is a different product. If you are being quoted for 200 people, ask precisely how many vehicles are tested, against what, and how long the testing block is. If the answer is that everything is judged rather than run, you are buying a decoration exercise.

How to split the teams. Split deliberately and announce the teams at the door. Two rules that matter more here than in most formats:

• Separate the engineers. In any office there are two or three people who have built something with their hands before. If they land on one team, that team wins every criterion and the day is over at minute twenty. Spread them one per team, and if you have five teams and only two such people, put them both on teams and tell the other three teams that trading with a neighbour is legal.

• Do not let people self-select. Self-selected teams in a build format sort by confidence, and the confident group takes the good parts first at the market.

One structural point about the test lane. Everything about this format scales with lane availability, not with floor space. You can fit a hundred people building at tables in a ballroom. You cannot test a hundred vehicles in one afternoon on one lane. Decide the number of lanes first, then work backwards to the number of teams, then to headcount.

Two builders and four spectators: designing the idle problem out

Construction formats have a reliable and well known failure: two people build and everyone else watches. It is worse here than in most build activities, because the object is small and there is genuinely nowhere for a sixth pair of hands to go. Telling people to participate does not fix it. Structure does.

Assign named roles before the design block and print them on a card per team. Six roles for a team of six:

• Design lead. Owns the sketch and the final shape. Must get every member to initial the drawing before the parts order is submitted.

• Buyer. Takes the parts order to the market, makes the substitution decisions when something has run out, and negotiates trades with other teams. This person is away from the table for a meaningful part of the session, which is the point.

• Fabricator, two of them. The two pairs of hands on the vehicle.

• Test lead. Owns the practice slot, runs the vehicle, records what happened and brings back a change request. This is a real job and it belongs to someone who is not building.

• Scrutineer. Owns compliance. Weighs the vehicle against the published limit as it is built, checks the dimensions, checks the payload attachment, and is the person who says at minute sixty that the vehicle is 90 grams over. Every team that skips this role finds out at scrutineering.

Four rules that actually shift the behaviour:

1. Only two people may touch the vehicle at any moment, and the pair rotates on a whistle every ten minutes. This looks artificial and it works. It also guarantees that at least four people have handled the thing, which changes the debrief completely.

2. The person who has built the most may not run the test. It separates building from evaluating, and it stops the builder explaining away a bad run.

3. The parts order must be initialled by every team member. It forces the design decision to be a conversation rather than an announcement.

4. Give the non-builders a parallel, real sub-task. The payload rig, the team name and one line design rationale for the scoresheet, the run log for the practice window. Sub-tasks must be genuinely needed, because busywork is transparent and resented.

The honest limit here. Even with all of this, a build format will never distribute effort as evenly as a station rotation does. If your primary goal is that thirty people are all equally occupied for two hours, this is the wrong format and something with parallel stations is the right one. What this format offers instead is one high quality decision that a whole team owns, and that is worth some unevenness.

Sourcing the parts, and the realistic per team spend

Figures below are per team for a desktop-scale vehicle, at published local retail as at September 2026.

The parts shelf:

• Wheels. The item that most affects the result. Options are toy wheel and axle sets, bottle caps, castors, CD discs or cut foam. Toy wheel sets run roughly $4 to $12 a pack. Buy two grades, a good one and a poor one, and price them differently at the market.

• Axles. Bamboo skewers, wooden dowel, steel rod or plastic straws. Skewers are around $2 a pack, dowel around $2 to $5 a length. Axle straightness is the difference between a vehicle that coasts four metres and one that coasts one.

• Chassis material. Corrugated cardboard, foam board, thin ply or plastic sheet. A large corrugated roll at local hardware retail is around $22 and yields chassis stock for many teams. Foam board sheets run roughly $3 to $8 each.

• Drive element, if the vehicle is self-propelled. Elastic bands and a spool, a wound string and a falling weight, a mousetrap, or a balloon and a straw. All cheap, all under $5 a team.

• Fixings. Masking tape at around $2 to $4 a roll, cable ties around $3 a pack, glue sticks around $2. Pick masking over clear tape, since it lifts off a hired table without taking the finish with it.

• Ballast and payload. Steel washers, sealed water bottles, sand in a bag. A 2 kilogram payload can simply be two sealed 1 litre bottles taped together, which costs nothing and is identical for every team.

• Tools. One pair of scissors and one ruler per team, around $2 each. A single shared cutting mat and craft knife under supervision, or no knives at all if the group is large.

A realistic per team parts cost is $15 to $35 for a desktop-scale build bought new at retail, and most of it is reusable. A five team, thirty person session can be equipped for roughly $90 to $180 in materials, plus around $30 for shared tools, scales and lane marking. That figure is worth holding in mind when reading a facilitated quotation, because it tells you what you are actually buying: the facilitation, the scoring, the timing equipment, the venue and the setup labour. The parts are the cheapest thing in the room.

Equipment you must not improvise:

• A set of kitchen scales reading to 1 gram, for the weight gate. One set, used for every team, in public.

• A stopwatch or a phone timer, and one nominated timekeeper who times every run. Not a different person per team.

• A tape measure for the coast distance, and floor tape for the lane.

• A ramp of fixed geometry if you are running a coast test. A plank on two books is fine, as long as it never moves.

The full-size kit versions are a different cost structure entirely. Those kits are supplied by the operator, the parts are pre-cut and branded, and the cost is inside the per pax rate rather than itemised. If you are quoted for one, ask whether the kit is yours afterwards, because a flat-pack car is bulky and somebody has to dispose of it.

Space, surface and the test lane you have to find

A build table per team. One trestle table, roughly 1.8 metres, per team of five or six, with at least a metre of clear floor around it. Teams that share a table interfere with each other and steal parts by accident.

A test lane of 12 to 15 metres of clear, level, hard floor. That lane binds everything else about the format, and it is almost never mentioned in a function room quotation. You need 10 metres of measured run plus a start area plus a run-out. Carpet is the problem: a small wheeled vehicle behaves entirely differently on carpet than on vinyl or polished concrete, and heavy carpet will stop a light vehicle dead. Check the floor before you commit, and if the room is carpeted, either lay a run of hardboard or plywood strips down the lane, or change the test to something that carpet does not destroy, such as a towing or a payload-drag test.

A level lane. Function room floors slope more often than you would think, particularly near drains and doorways. Roll a marble down the intended lane before you commit to it. A lane with a slope makes the coast test meaningless and the speed test unfair depending on direction, so if it slopes, run every test in both directions and take the average, and say so in the brief.

Circulation. A metre and a half of clear floor either side of the lane, with a taped standing line for spectators. Thirty people crowding a lane will kick a vehicle mid-run, and that is a dispute you cannot resolve.

Ventilation and air conditioning. A vent blowing across the lane will move a light vehicle measurably. Identify vents before you place the lane, and if a vent cannot be closed, run the lane across the draught rather than along it.

Published venue cost. Function room hire in this market is published at roughly $40 to $250 per hour, with three hour minimums usual and space planning advice published at 30 to 40 square feet a head for a seated function. That guidance understates this format, because a test lane plus spectator space plus build tables needs more floor per head than a seated function does. An office corridor or a car park deck with a smooth surface is frequently a better lane than a hotel ballroom, and costs nothing.

Judging a raw materials build when every team read the brief differently

This is the central problem of the raw materials version and it deserves to be solved on paper, days before the event.

The situation is predictable: five teams get the same brief and the same stock, and at scrutineering one team presents a four-wheeled cart, one presents a sledge on two runners, one presents something with a single large wheel, and one presents a vehicle that carries its payload slung underneath rather than on top. Are all four vehicles? Under most casually written briefs, three of them are arguably not, and whichever way the facilitator rules, one team feels robbed.

Define the object, in writing, with measurable gates. The brief must contain a definition, not a description. A workable one: the vehicle must travel the full lane without any part of a person touching it after the start line, must carry the payload without a person supporting it, must weigh between 400 grams and 1.2 kilograms empty, must fit inside a 50 by 30 by 30 centimetre box, and must not damage the floor. Every clause there is measurable by one person with scales, a box and their eyes. Note what the definition does not say: it does not say wheels, it does not say four of anything, and it does not say a shape. That is deliberate. Constrain the performance and the envelope, not the solution.

Announce gates to the teams as pass or fail, and keep them off the scoresheet. A gate is a condition of entry, not a scored criterion. Mixing the two produces arguments about how many points a marginal breach costs. Pass or fail is a much easier conversation.

Keep the judged share small. In the scheme above, 5 points out of 100 are judged and 95 are measured. Every point that is judged is a point that can be argued about. If your version needs a creativity award, hand it out as a separate named prize with no bearing on the classification, which lets you be generous without being challenged.

Show the teams the scoresheet before the design block. Teams that can read the weights while they design will design to them, which is the entire mechanism by which the multi-criteria test changes behaviour. A scoresheet revealed at the end is a trick, not a test.

Measure everything the same way, with the same instrument, by the same person. One set of scales, one stopwatch, one timekeeper, one tape measure. Never let team members time their own runs.

Put an appeals protocol in the opening brief, and keep it small. One appeal per team, raised immediately after the run in question, resolved by re-running rather than by argument wherever a re-run is possible. Re-running settles most disputes in ninety seconds and satisfies everyone. Rulings that cannot be settled by a re-run go to the facilitator and are final, and saying so in advance removes most of the heat.

Handle the genuinely novel solution well. Once per session a team will build something that nobody anticipated and that technically satisfies every clause. The correct response is to allow it and to say out loud why it is allowed, referring to the written definition. A team that finds a legal solution nobody expected has done exactly what the brief asked for, and penalising them teaches the room that the real rule was to guess what the facilitator had in mind.

Write the boring clause. Add one sentence to the brief: any question about the rules must be raised before the build window opens, and after that the written brief stands as written. It moves the interpretation conversation to the front of the session, which is where you have time for it.

Making it harder or gentler by moving the budget, not the brief

Change one variable at a time. The budget and the shelf are the cleanest dials, because moving them changes the difficulty without changing what is being measured.

Gentler. Raise the token budget to 130 per cent of a full specification so scarcity bites less. Stock the shelf deeply so nothing runs out. Drop the design lock and let teams buy throughout. Extend the build to sixty minutes. Reduce to three test criteria by dropping the coast test. Supply pre-cut chassis blanks. Allow unlimited practice runs.

Standard. Budget at roughly 70 per cent of a full specification. Finite shelf, with the good parts scarce. Fifteen minute design lock with a submitted parts order. Forty minute build. Four criteria as scored above. Ten minutes of practice in booked slots.

Harder. Budget at 55 per cent. Ban the most obvious part outright, usually the pre-made wheels. Two practice runs only, per team, for the whole session. Add a terrain criterion, such as a 2 metre section of the lane covered with a folded blanket, which changes the optimal wheel size and forces a real compromise with the flat speed run. State a payload in the brief that is awkward rather than heavy, such as an open cup filled to a marked line.

Hardest, and only for a group that wants it. No practice runs at all, so the scored run is the first time the vehicle has been on the lane. A design lock with no revisions permitted after the parts order. A payload announced only after the design lock closes, so teams must design for an unknown load within a published weight range. This last one is genuinely difficult, it is the closest this format gets to real engineering, and it is a very good exercise for a team whose actual complaint is that they commit to designs too early.

Two dials to leave alone. Do not lengthen the lane, because it makes the tests slower without making them harder. And do not shrink the build window below thirty minutes to add difficulty, because what you get is not a harder design problem, it is an unfinished vehicle, and an unfinished vehicle scores zero on everything and teaches nothing.

The five predictable breakdowns, and the pre-emptive fix for each one

The featherweight strategy, discovered too late. Symptom: a team builds the lightest possible chassis, wins the speed run, and then watches it fold under the payload with no time to change anything. Fix: publish the weights before the design block, and state the payload mass in the brief in the first ten minutes. Then let it happen anyway to any team that ignores it, because that specific failure is the most instructive thing that will occur all afternoon and it belongs in the debrief rather than in a warning.

The build that is not finished at tools-down. Symptom: at the deadline, one team's vehicle is in three pieces. Fix: enforce the first tools-down visibly, use booked practice slots to pull teams off the build voluntarily, and add a published partial credit rule so that an unfinished vehicle can still attempt the criteria it is capable of. A team with nothing to run has forty minutes of nothing left in their afternoon, which is a worse outcome for you than for them.

One person builds and five people watch. Symptom: at minute fifty, one person is soldering the whole thing together and the rest of the team is on their phones. Fix: the named roles, the two-hands rule with a ten minute rotation, and the initialled parts order. Apply all three, because any one of them alone is easy to ignore.

The rules argument at scrutineering. Symptom: a vehicle is 40 grams over the limit, or the payload is slung in a way nobody anticipated, and there is a fifteen minute argument in front of everyone. Fix: measurable gates published in writing, one instrument, pass or fail, an appeals protocol with one appeal per team, and a stated correction window during scrutineering so a marginal breach can be fixed rather than fought over.

The test block that runs out of clock. Symptom: the reliability criterion gets cut because it is last and the room is out of time, which retroactively deletes 25 per cent of the scoring the teams designed against. Fix: count the runs during planning, not on the day. Five teams times four criteria is roughly fifty runs. If the arithmetic does not fit, cut a criterion during planning and publish three weights instead of four. Never cut a criterion after the design block, because teams have already spent their tokens on it.

A sixth breakdown, rarer but worth stating. The team that builds something genuinely unsafe: a heavy vehicle launched hard down a lane at ankle height, a sharpened edge, a projectile release. Ban launching by foot, ban any stored energy that could release sideways, cap the payload mass in the rules, and keep the spectator line taped. A build format is one of the few corporate activities where a team can construct a hazard nobody specified.

Two hours, thirty people, five teams: the full worked brief and scoresheet

Thirty people, five teams of six, one facilitator with a helper, one 12 metre lane on a hard floor, and a two hour block.

The schedule.

That runs five minutes past two hours, which is honest. If the block is hard-stopped at two hours, take the five minutes out of the build window, never out of the debrief.

The brief, as issued to teams.

Build one vehicle that travels a 10 metre lane on its own, carrying a 2 kilogram payload, with no part of any person touching it after the start line. The empty vehicle must weigh between 400 grams and 1.2 kilograms and fit inside a 50 by 30 by 30 centimetre box. You have 100 tokens. Parts are priced on the shelf and stock is limited. You may trade parts with any other team at any time. You may not buy anything after your parts order has been submitted. Your vehicle will be scored on four measured tests and one small judged criterion, weighted as published on the scoresheet you have been given.

The parts price list, with limited stock for five teams.

A full specification vehicle, which is precision wheels plus foam board plus dowel axles plus an elastic drive plus a cradle plus tape, costs 93 tokens and leaves 7. Only four elastic drives exist for five teams, and only six precision wheel sets, so at least one team leaves the market without the drive it planned for. That scarcity is the design of the session, not an accident of stock.

The scoresheet as published. Speed on the flat 20, payload run 25, coast test 15, reliability 25, budget remaining 10, brief compliance and design 5.

What actually happened.

Team Alpha read the format as a race. They bought precision wheels, the lightest corrugated chassis, skewer axles and an elastic drive, spent 91 tokens, and built a bare, beautiful, very fast vehicle weighing 430 grams. They tested it empty, repeatedly, and it was quickest every time.

Team Bravo spent fifteen minutes arguing about the payload and then designed backwards from it. They bought basic wheels rather than precision ones, a foam board chassis, dowel axles, the weight and pulley drive, and the payload cradle, spending 96 tokens with a deliberately reinforced chassis. Their vehicle weighed 1.04 kilograms and was visibly not the fastest thing in the room.

Team Charlie over-built. Two chassis panels laminated together, cable ties everywhere, 1.19 kilograms, right on the limit. Slow, and unbreakable.

Team Delta missed the last elastic drive at the market, improvised a weight and pulley, and spent the build window fixing an axle that was not straight.

Team Echo hoarded tokens, spent 41 of them, and built a light vehicle with no payload cradle at all, planning to tape the bottles on at the lane.

The classification.

Alpha won the speed run outright, won the coast test outright, and finished last. Their chassis buckled under the 2 kilogram payload on the first attempt, which cost the entire 25 point criterion, and a skewer axle sheared on the second reliability run, which cost most of another 25.

Bravo won without winning a single test. Third in speed, second in payload, third in coast, full marks on reliability. Every one of those results came from one decision made at minute eighteen, in the design block, before a single part had been bought: design for the payload first and accept being slow.

Charlie's over-engineering was not punished nearly as hard as most people in the room expected, because reliability and payload together are worth half the scoresheet. Slow and unbreakable is a legitimate strategy under these weights, and saying so during the classification is a good moment.

Echo lost by hoarding. Nine budget points is a rounding error against the 50 points of payload and reliability that a cradle and a stronger chassis would have bought them. The 10 point cap on the budget criterion is what makes that arithmetic obvious, and it is why the budget weight must stay small.

The line to say when the scores go up. Read the four component rows before the totals, then say it plainly: the fastest vehicle in the room came last, and the team that came third in the race won the afternoon. Then ask Alpha what they would spend their first ten minutes on if they ran it again. That question is the whole session.

What it develops, and what it stops developing the moment the build starts

What it produces, reliably:

Role allocation under a deadline. This is the primary output. A team of six with forty minutes, a finite parts shelf and a physical object that only two people can touch has to decide who does what, and it has to decide in the first five minutes or lose the time. Unlike most workplace role allocation, the consequence is immediate and visible: a team with no scrutineer discovers it at the weight gate, and a team with no test lead wastes its practice slot. The mechanism is scarcity of hands rather than scarcity of skill, which is what makes it happen every time rather than sometimes.

Negotiating a design decision to a conclusion. The design lock block, with no parts on the table and a parts order that every member must initial, forces a group to converge on one answer and commit to it. Most workplace design disagreements are resolved by drift, seniority or exhaustion. Here there is a submission deadline and a signature, and the trade-off is legible: fast or strong, cheap or reliable, and the scoresheet says exactly what each is worth. Teams get to watch a decision they made at minute eighteen determine an outcome at minute one hundred, and that causal chain is short enough to see.

Committing to a trade-off in public, and living with it. A team that chose speed cannot quietly re-frame their choice when the payload fails, because thirty people watched the chassis fold. That is uncomfortable and it is the most useful discomfort the format produces.

What it fails to reach, and should never be marketed as reaching:

It does almost nothing for communication under pressure. This is the honest limit and it is worth being precise about, because build formats are routinely sold as communication exercises. Once the build window opens, teams go quiet and heads-down. Two people work on the object, the room's noise level drops noticeably, and the exchanges become short, functional and technical: pass the tape, hold this, is it straight. There is no coordination problem to talk through, no information asymmetry to resolve, and no reason to negotiate with anyone outside the team. Whatever communication happens in a build-and-race happens in the design block, before any material is issued, and that block is fifteen minutes of the session. If communication under pressure is what you actually need, run a format with split information, forced verbal instruction or a time-critical shared task, and run this one for something else.

It is not an icebreaker. It is technical, fiddly and quiet, and confident people take the first twenty minutes. Strangers should do something loud first.

It builds no trust in the interpersonal sense. It builds a shared reference to a decision, which is narrower and different.

It produces no physical energy release. Nobody moves much and nobody sweats.

It quietly favours some people over others. Manual dexterity, prior hands-on experience and comfort with tools are genuinely rewarded. Balance this by spreading experienced builders across teams, and by making the non-building roles real rather than decorative.

It is weak on decision making under genuine uncertainty. The constraints are published, the scoring is published and the test is visible. What is being exercised is committing to a trade-off under known rules, which is a real and useful capability but is not the same thing as deciding without information.

The quarter hour afterwards: questions aimed at the design decision

Do not open with what everyone thought of the activity. Open at the design block, because that is where every outcome in the room was determined, and this format hands you unusually clean evidence.

About the decision itself:

1. What was decided during your design block, and how many minutes passed between the first suggestion and the submitted parts order?

2. Who made the call in the end? Was it the person with the idea, the loudest person, or the person nobody wanted to argue with?

3. Which idea did you reject, and can the person whose idea it was say why they accepted the rejection?

4. Did anyone initial the parts order without agreeing with it?

About the trade-off:

1. You could not have the fastest and the strongest vehicle. Where did your team put itself on that line, and was that a decision or a default?

2. When did you first read the scoring weights properly? Before the sketch, during the build, or when the scores went up?

3. What did you spend your last twenty tokens on, and what would you spend them on now?

About roles and the build:

1. Who did nothing for a stretch of the build window? Was that their choice, or did the structure leave them nowhere to stand?

2. Who ran your practice slot, and did the team act on what they came back and said?

3. If you had to build the same vehicle again in forty minutes, which two people would you put on it and which four would you put somewhere else?

About the result:

1. The fastest vehicle finished last. When did the team that built it realise that was going to happen?

2. Which criterion did your team ignore, and how many points did ignoring it cost?

3. Was the winner the best team, or the team that read the scoresheet earliest? Does the difference matter?

Transfer questions, which are the point of the whole afternoon:

1. Where in our work do we commit to a design before we have understood how it will be measured?

2. What is our real version of the scoring weights, and does everyone on this team know them?

3. When we make a technical decision, do we get a signature from everyone or an absence of objection? Which one did we do today?

4. Name one decision coming up in the next month where a fifteen minute locked design block, with no work permitted, would have changed the outcome.

Question sixteen is the one to sit with. Most teams recognise immediately that what they actually run on is an absence of objection, and this afternoon gave them a clean and slightly embarrassing example of what that produces. Put the answer to question seventeen on a flipchart, take a picture of it, and circulate that picture before lunch the following day.

What suppliers call this, and the formats it should not be confused with

The format trades under a range of generic names in this market. Build and race, vehicle build, car build challenge, kart build, soapbox challenge, cardboard boat build, raft buildand construction challenge all describe versions of the same underlying mechanic: teams build something, and then it is tested. Suppliers here habitually invent their own product names and lay them over the generic ones, so an identical activity may reach you under a label that exists nowhere else. Judge the mechanic and the test, never the label.

Three genuine sub-formats hide among the names, and the difference between them is large:

• The kit assembly build. Pre-cut parts, an implied final shape, a decoration phase and a push race. Photogenic, scales to hundreds, and is largely an assembly exercise with a social finish.

• The raw materials build. Sheet stock and tape, an open design space, and a physical test. Much more interesting, much harder to judge, and does not scale past about six teams per lane.

• The scale model build. Desktop-sized vehicles, a short lane, and a multi-criteria test. The cheapest to run in-house and the version this page is written around.

Formats it is not:

• It is not a segment-chain build, where each team constructs one part of a single machine and the parts are joined at the end. There the whole point is the boundary between two teams' work and the dependency it creates. Here every team's vehicle is entirely its own, and no team's result depends on another team's build.

• It is not a tower or bridge building challenge. Those score a static structure on height or load. Nothing moves, there is no run, and the whole second half of this format, the test, does not exist.

• It is not a station rotation format. Nobody rotates, nothing runs in parallel, and there is one long build rather than a series of short scored challenges.

• It is not a route based race. Nobody travels anywhere, no clues are issued and nothing has to be navigated. The group stays at its tables.

• It is not a go-kart or driving experience. Those are about operating a vehicle somebody else built. This one is about the building, and the running is a measurement rather than a sport.

Nine questions for a supplier, which work whatever the product is called:

1. Is it a supplied kit with pre-cut parts, or open materials? If pre-cut, how much of the final shape is already decided?

2. What is the test at the end, exactly? If the answer is a race, ask what else is measured.

3. How many scored criteria are there, and what are the weights? A single criterion means a single strategy.

4. Is there a budget or a parts market, or does every team simply receive an identical box?

5. How many minutes are set aside for the test block, separately from the build?

6. Do teams get practice runs on the lane before scoring?

7. What surface is the lane, and have you tested it on carpet?

8. How many vehicles are actually tested if we bring 150 people, and how long does that take?

9. What are the written rules for what counts as a legal vehicle, and can I see them before we book?

Questions three, five and nine separate a supplier who has designed this format from one who has bought some kits.

PLAYON's honest answer on vehicle builds

PLAYON does not run build-and-race activities, and no cut-down version of one belongs on an indoor attraction floor: this format wants long build tables, a measured lane and a quiet room, which is simply a different kind of space. Should your shortlist also hold a scored, station based competitive afternoon with nothing to purchase, assemble or throw away afterwards, PLAYON suits that well enough. For a vehicle build, look for a hall, a hard floor, and a supplier willing to state the scoring weights before you sign.

Questions organisers raise before approving a vehicle build

What is a build-and-race team building activity? It is a construction challenge where each team builds a vehicle from a kit or raw materials inside a fixed window, then runs it against a defined test. The test, not the build, is what makes it a competition worth scoring.

Should we use a supplied kit or raw materials? Use a kit for large groups, mixed seniority or thin facilitation, because it is fair and repeatable. Use raw materials for small technical groups who want a real design decision. A priced parts market with a token budget gives you most of both.

How many people does a build-and-race need? Twenty to thirty-six people works best, split into four to six teams of five or six. The limit is the test lane rather than the floor space, because every extra team adds around ten individual runs to the testing block.

How long does a build-and-race take? Two hours is the realistic minimum for a desktop-scale build: ten minutes brief, fifteen design, ten market, forty build, ten practice, five scrutineering, twenty testing and ten debrief. Published full-size kit versions run 2.5 to 3.5 hours.

Why is a straight race a bad test? Because the winning strategy is simply to build the lightest thing that rolls, which anyone experienced works out in ninety seconds. It collapses the design phase, produces a result nobody believes, and leaves you with nothing worth debriefing.

How should a build-and-race be scored? Score four measured properties that conflict: speed 20 points, a payload run 25, a coast test 15, reliability across three unrepaired runs 25, budget remaining 10 and judged design 5. Give the teams those weights before the design block.

What constraints make it a design exercise rather than assembly? A token budget tight enough to prevent buying everything, a parts market with limited stock that forces trading, a published weight window, and a payload the vehicle must carry. Together they force a trade-off no team can avoid making.

How do you stop two people building while four watch? Assign named roles, allow only two pairs of hands on the vehicle at once, rotate that pair every ten minutes on a whistle, require every member to initial the parts order, and bar the main builder from running the test.

Why do build teams always overrun? Because a build has no natural finish point, only a removal point. Call a hard tools-down twice, book practice slots that pull teams off the build voluntarily, and publish that anything added after the deadline is removed at scrutineering.

What does a build-and-race cost to run in-house? Parts run roughly $15 to $35 per team for a desktop-scale build, so a thirty person session costs about $90 to $180 in materials plus around $30 for scales, timing and lane tape. Most of it is reusable.

How do you judge fairly when teams interpret the brief differently? Define the vehicle in writing with measurable gates covering weight, dimensions, payload attachment and no human contact after the start. Make gates pass or fail, keep judged points under 10 per cent, and publish an appeals protocol.

What surface does the test lane need? Twelve to fifteen metres of clear, level, hard floor. Carpet stops small wheeled vehicles dead and changes the result entirely. Roll a marble down the intended lane before committing, because function room floors slope more often than people expect.

What does a build-and-race actually develop? Role allocation under a deadline and negotiating a design decision to a firm conclusion. The scarcity of hands forces roles, and the locked design block with an initialled parts order forces a group to commit rather than drift.

Does it build communication under pressure? No. Once the build starts, teams go quiet and heads-down, exchanges become short and functional, and there is no coordination problem to talk through. Whatever communication happens occurs in the design block before any material is issued.

What does a build-and-race cost through a supplier in Singapore? Published bands run $25 to $40 a head for an in-office activity, $45 to $90 for a half day at an outside venue, and $100 to $180 for a premium full day. Weekends carry 15 to 25 per cent loading.