Some things are hard to explain with photos. A machine that has to fit through a particular door. A commercial kitchen that has to work around three pillars and a drain nobody can move. A flat that does not exist yet, sold from a floor plan to someone who has never read one. A piece of equipment whose most important part sits inside a steel housing where no camera can see it.
For businesses like these, a 3D model or a virtual reality walk-through sounds like the obvious answer, and sometimes it is. I have also seen the opposite: a beautiful interactive model that took weeks to build, was shown at one trade fair, and then lived on a laptop nobody opened again.
The difference between the two is rarely the quality of the graphics. It is one mistake made at the very start, before anything is modelled. This article is about that mistake, why it is so easy to make, and what to do instead, so that if you spend money on 3D or VR, it helps people decide rather than just impressing them for a minute.
The mistake: starting with the headset, not the question
The mistake I see most often is starting from the medium. Someone sees a competitor's headset demo at a fair, or a slick spinning model on a website, and the brief becomes "we need something like that". The conversation that follows is about headsets, rendering quality, how realistic the materials look and how much the whole thing will cost.
What nobody writes down is the one thing that matters: which question is a buyer unable to answer today, and how will this help them answer it?
Without that, the project has no way to succeed or fail. It gets judged on how it looks in the demo. It gets built with every detail at the same level of polish, because nobody knows which details matter. And once the novelty wears off, nobody can say whether it was worth it, because nobody said what it was for.
Why it happens so often
It is an easy mistake because 3D and VR are bought the way marketing is bought, by looking at examples. Examples show the finished surface, never the problem underneath. A walk-through of a showroom looks impressive whether or not a single visitor needed it.
It is also easy because the people selling 3D work are usually artists and developers, and they are, rightly, proud of what the technology can do. Asking "what can it show?" gets you a long and exciting list. Asking "what does your buyer get wrong today?" is a different conversation, and a less glamorous one, so it often does not happen.
And there is a timing problem. The decision to "do something in 3D" is often taken a few weeks before an event, which leaves time to build something shiny and no time to find out what visitors actually ask.
Start with the question the buyer cannot answer
Before anything is modelled, I ask the people who talk to buyers every day (the sales engineer, the person who does site surveys, the one who answers the phone) a few plain questions:
- What do people ask again and again, even after they have seen the brochure?
- Where do misunderstandings show up later, after the contract is signed?
- What do you end up drawing on a napkin, or explaining with your hands?
- When do you have to send someone to visit, just to show them something?
Their answers are the brief. If the honest answer is that buyers understand the thing perfectly well from photos and a spec sheet, that is useful too: it means 3D will not change anything, and the money is better spent elsewhere.
Where 3D genuinely earns its keep
In my experience, 3D genuinely changes a decision in a handful of situations.
**When size and fit are the question.** Will this machine fit the hall, with room to open its doors and service it? Will this layout leave enough space for two people to pass with trays? A photo cannot answer that; a model placed in the real room, or in a plan of it, can.
**When the important part is hidden.** Equipment whose value lies in what happens inside a housing, a building's wiring and pipes behind the walls, a device's internal layout. Being able to make the outer shell transparent, or cut through it, explains in seconds what a paragraph cannot.
**When the thing does not exist yet.** Flats sold before they are built, a clinic fit-out before the builders arrive, a classroom layout a school is still choosing between. People who are not used to reading plans make better decisions when they can stand in the space, even a virtual one, before it is poured in concrete.
**When there are many combinations.** A fit-out where the customer chooses layout, finishes and equipment from a set of options. Showing the combination they picked, rather than a generic one, removes a whole category of "I imagined it differently".
Notice that all four are about a decision the buyer has to make. That is the test.
Where it is a gimmick
The same tools become a gimmick when they show what a photo already shows, only more slowly. A spinning model of something whose shape nobody was confused about adds a loading spinner, not information.
VR in particular becomes a gimmick when it is the only way to see the thing. A headset at a trade stand reaches one visitor at a time, needs someone to fit it and explain the controls, and leaves everyone else watching a stranger wave their arms. If the walk-through is genuinely the best way to understand the space, that can still be worth it. If it is there to draw a crowd, a large screen showing the same model draws the crowd too, and lets the sales engineer talk while they point.
The other sign of a gimmick is a model that is more beautiful than it is accurate. If the walk-through shows a finish, a view or a clearance the real thing will not have, it does not help a buyer decide; it creates the argument you will have after handover.
Phone first, headset second
When I am asked for VR, my usual first suggestion is a 3D model that opens in an ordinary web page on any phone or laptop. Not because headsets are bad, but because that is where your buyers already are, and it does not need anyone to be in the room.
The web has become very good at this. Google's open-source model-viewer component, for example, shows a 3D model in a page and, on phones that support it, lets the visitor place it in their own room through the camera. On iPhone, Apple's AR Quick Look does the same from a file the page links to. A facilities manager can stand in their own plant room, point their phone at the floor and see whether the unit fits, then send the link to the person who signs off the budget.
That last part is easy to underestimate. Most purchases of this kind are decided by more than one person, and most of those people will never visit your stand. A link that works on their phone travels through their organisation; a headset stays where it is.
The catch is speed. A heavy model on a page that takes ages to load on a phone loses people before they see anything, which is why the model needs to be prepared for the web, not exported straight from design software. Fast on a real phone covers how I measure that.
Headsets: plan for a fleet, not a gadget
There are good reasons to use headsets: a full-scale walk-through of a space, a sense of height and distance a screen cannot give, a training scene where the hands matter. If you go that way, plan for the hardware as a small fleet of devices, not a gadget.
That includes thinking about the vendor. In January 2026 Meta announced that it would stop selling its business versions of the Quest headset and its Horizon managed services, the tools companies used to look after fleets of headsets, from 20 February 2026. Existing customers keep support for the current service until 4 January 2030, and the consumer Quest 3 and 3S are still on sale. Meta also ended Horizon Workrooms, its virtual meeting room app, on 16 February 2026. None of that stops anyone from showing a walk-through on a Quest today, but it is a reminder that a VR experience should not depend on one vendor's business programme staying the same for years.
Shared headsets also raise the practical question of what happens between visitors. Apple, for example, added an option in visionOS 26 for a managed Vision Pro to be wiped and set up again for the next user from the Lock Screen. Whatever device you choose, decide before the event how it is cleaned, reset, charged and kept from showing the last visitor's session to the next one.
And the content should not be locked to the headset. If the model is built so that the same scene runs on a laptop and in a web page as well, the work survives a change of hardware.
The model is the expensive part, and you may already have it
Most people expect the cost to be in the headset or the app. It is usually in the model.
The good news is that many businesses already have one. Machines are designed in CAD software; buildings and fit-outs increasingly in BIM tools, where every wall and pipe is an object with a size and a type. Those files are a far better starting point than modelling from photos.
The less good news is that they cannot be used as they are. A design file contains every screw, often hundreds of thousands of parts, because engineers need them. A phone needs a lighter version that keeps what the buyer sees and drops what they do not, with surfaces that look like the real materials. That conversion is skilled work, and it is where a project's budget is best spent: on the parts the buyer's question is about, in full detail, and everything else simplified.
It is also where confidential information can leak. A design file may show internal construction you would rather not hand to a competitor. Deciding what the public model leaves out is part of preparing it, not an afterthought.
Keeping the model true to the real thing
A 3D model shown to a buyer is a promise. If the real machine has a different panel, if the real flat has a column in the living room that the walk-through left out, the model has made the conversation after handover harder, not easier.
So treat the model like any other document that describes what you deliver. Decide where it comes from (ideally generated from the same design files the engineers use), who updates it when the design changes, and which version a buyer saw. For anything sold off-plan or built to measure, I would keep a note of which model version was shown with which proposal, the same way you keep the version of a drawing.
This sounds like paperwork. It is the difference between a model that keeps helping for years and one that becomes quietly wrong after the first design change, and then gets switched off because nobody trusts it.
How to tell whether it worked
Because the project started from a question, it is possible to tell whether it worked. Not by counting views, which tell you people opened it, but by watching what changes in the conversations it was meant to help:
- Do the questions that came up again and again come up less?
- Are fewer site visits needed just to show something?
- Do fewer changes and disputes come up after signing because "I imagined it differently"?
- Do the people who sign off the budget, the ones who never visited, ask better questions?
The people who talk to buyers will usually know within a few weeks. If nothing has changed, it is better to learn that early and adjust, or stop, than to keep paying for a model that only looks good.
What I would build first
If you came to me wanting "something in 3D or VR", this is the sequence I would suggest:
- Spend an afternoon with whoever talks to buyers and write down the questions they cannot answer well today.
- Pick the one question that costs the most when it goes wrong: a wasted visit, a lost deal, a dispute after handover.
- Find out what design files exist, and what must stay private.
- Build one model, aimed at that question, that opens on an ordinary phone from a link.
- Use it for a few weeks in real conversations and listen to what changes.
- Only then decide whether a headset walk-through, more models or a configurator of options is worth adding.
It is a less exciting start than a VR launch at a trade fair. It is also much more likely to still be in use a year later. When the goal is teaching staff rather than helping buyers decide, the reasoning is a little different, and A training simulation instead of a manual goes through that case.
- Which question does a buyer get wrong today without it?
- Who will use it: a visitor at a stand, or the colleague who signs off the budget and never visits?
- Does it open from a link on an ordinary phone?
- Which design files does it come from, and who updates it when the design changes?
- What must the public model leave out?
If the first answer is "nobody gets anything wrong, it would just look modern", that is your answer too.
What do your buyers struggle to picture?
Discuss your projectTell me what they ask again and again, and I will tell you whether 3D would answer it, or whether something simpler would.