Augmented Reality Preview Cost and What Drives It

Augmented reality pricing in real estate looks irrational from the outside. The same brief produces quotes that differ by a factor of five or ten, and none of the vendors is being dishonest.

The variation has one dominant cause and several minor ones. This guide separates them, because a buyer who understands the dominant cause can read any quote correctly in about two minutes.

What the phrase covers, and why that matters to the price

Part of the spread comes from the term itself. An augmented reality preview can describe four quite different deliverables, and buyers rarely specify which one they mean.

The first is a tabletop model. A scaled version of the building appearing on a desk or a floor, viewable anywhere. It is the cheapest, it travels, and it is the least persuasive because it removes the one property that makes the format valuable.

The second is a full scale on site preview. The building at true size in its real place, seen by somebody standing there. This is the version worth paying for and the version everything else in this guide describes.

The third is an interior overlay, where finishes, fixtures or furniture appear inside a space that already exists. That is a product problem rather than a building problem and it is priced quite differently, usually lower.

The fourth is a field verification tool, where a design model is placed against as built conditions to check them. That is an instrument rather than a marketing asset, and the accuracy requirement changes the cost class entirely.

A brief that does not say which of the four is wanted will collect quotes for several of them at once, which explains a surprising share of the confusion.

The dominant variable: who builds the model

Every augmented reality experience displays a 3D model. No augmented reality application creates one from drawings.

So the first question in any quote is whether model production is inside it or outside it, and that single distinction accounts for most of the spread between competing prices.

A vendor quoting to publish an existing model is quoting for a genuinely small job. A vendor quoting to build the model from a document set is quoting for something closer in effort to a rendering package. Both figures can be correct and they are not comparable.

The way to make quotes comparable is to state what you have. If the answer is drawings and no display model, say so in the brief, and the spread narrows immediately.

Why the model you already have usually does not count

This is the most expensive misunderstanding in the category, and it catches sophisticated teams.

A coordination model built in Revit or similar carries construction level information: structure, fastenings, mechanical systems, equipment, layers of data no viewer will see. It can run to millions of polygons.

A phone rendering augmented reality is also running the camera and tracking its own position, and the geometry budget left over is a small fraction of what a workstation handles without noticing. The documentation model refuses to open, opens and crawls, or heats the device until it throttles.

Converting it means rebuilding rather than exporting, and that work is skilled and takes time. It is the line item that separates the quotes, and any proposal that does not mention it has either assumed you have a display model or has not looked.

The cost variables, ranked by how much they move the number

Detail level: the variable buyers control most easily

The difference between a massing view and a hero facade is not a small increment. It is a different production.

A massing view answers scale and position, which is what a neighbour meeting or a lot walk needs. It requires correct dimensions, correct placement and enough material suggestion to read as a building rather than a block.

A hero facade answers what the building will look like, which is what a buyer standing at the entrance needs. It requires modelled openings, real material behaviour, resolved details at close range, and a great deal more testing outdoors.

A large share of augmented reality budget overruns come from buying the second when the first would have answered the question, and it is worth deciding deliberately rather than defaulting to maximum detail.

Context modelling is a real cost and an easy saving

Surrounding buildings have to be in the experience because the questions viewers ask are comparative. That does not mean they have to be beautiful.

Massing at correct height and position, with simple materials, answers every comparative question. Detailed neighbours answer none of them any better and cost several times as much.

The exception is a context where the surroundings are the subject, such as a historic street where the argument is about how a new facade sits against existing ones. There the context deserves real attention, because it is the point of the exercise.

Positioning accuracy: the step change in cost

Phone based placement is included in essentially every augmented reality production. Surveyed placement is a separate purchase with separate economics.

The surveyed route involves positioning hardware, field time, and somebody competent to operate it. It is not marginally more expensive, it is a different class of engagement.

The decision rule is worth applying literally: if somebody is going to make a decision based on where the model appears to sit, pay for surveyed placement. If the purpose is to communicate scale and character, do not.

Paying for surveyed accuracy on a marketing experience is one of the more common ways to overspend here, and skipping it on anything dimensional is the more consequential mistake in the other direction.

Interaction is software cost, not model cost

Finish switching, unit selection, information panels, saved favourites and analytics are all application development.

They are priced by feature and by platform rather than by building, and they carry maintenance obligations the model does not. A static augmented reality view of a building will still work in three years. A bespoke application with a menu system will need attention as operating systems change.

For a first project, the honest advice is to buy less interaction than feels exciting. The single strongest moment in augmented reality is a full scale building appearing in a real place, and menus do not improve it.

The recurring costs that get left out of first budgets

Hosting and delivery. Something has to serve the model and the experience, and it has to keep doing so.

Device and platform updates. Mobile operating systems change annually and augmented reality frameworks change with them, so an experience left untouched for two years frequently stops working correctly.

Model updates. Designs change during a project, and an augmented reality preview showing a superseded facade is worse than none.

Tooling risk. Adobe Aero, a mainstream augmented reality authoring tool from a major vendor, reached end of support on 6 November 2025. Anything built on a single proprietary application inherits that kind of event, which is an argument for owning the model rather than only the published experience.

How to make three quotes comparable

State what geometry exists and in what format, honestly, including whether anybody has confirmed it opens.

State the level of detail required and where the viewer will stand, because those two together determine the modelling effort.

State whether surrounding context is required and whether accuracy of context matters.

State whether any decision will be made from apparent position, which determines whether surveyed placement is in scope.

Ask every vendor to price model production as a separate line, even if they intend to bundle it. The spread will collapse and the outlier will become obvious.

Where the money is best spent first

The model, because it is the only asset that survives everything else. Vendors change, platforms shut down, phones get replaced, and a well built model still produces renderings, animation, a tour and an augmented reality preview.

Staging second, because an experience nobody can use correctly has wasted the model.

Interaction last, and less of it than the demonstration suggested.

When augmented reality is not the efficient purchase

Remote audiences. The format reaches nobody who is not at the site, so for out of state buyers the same money produces far more value in renderings or a tour.

Single asset budgets. If exactly one visual can be afforded, it should be an image, because an image reaches every audience through every channel.

Projects with no site access or no foot traffic. Augmented reality with no visitors is an asset with no audience.

Interiors before a shell exists. Standing in a field looking at a kitchen communicates very little compared with a rendered interior or a walkthrough.

What reasonable value looks like

A useful test is cost per person who actually experiences it, rather than cost per asset.

A community meeting where sixty residents each see the proposal at full scale is a strong return on a single production. A hoarding QR code scanned eleven times in a month is not, even if the production was excellent.

That reframing tends to move budget from model detail toward staging and toward the human handing over the device, which is where the return actually improves.

Volume changes the arithmetic completely

A single building and a phased community are not the same purchase, and the per unit economics move sharply between them.

For one building, everything is a fixed cost. The model, the context, the terrain, the staging and the testing all get paid for once and spread across one asset. That is the least efficient configuration and it is where most first projects sit.

For a community with repeating house types or unit plans, the model work amortises. Each additional plan reuses the same terrain, the same context approach, the same material library and the same staging decisions, so the second costs a fraction of the first and the eighth costs less again.

That is the argument for productised platforms over bespoke production at volume, and it is also the argument for planning the whole programme before commissioning the first piece. Commissioning plan one, then plan two six months later from a different vendor, discards the saving entirely.

Phasing the spend across a project timeline

An augmented reality preview does not have to be bought all at once, and staging the spend usually improves the result.

At entitlement and early marketing, what is needed is scale and mass against context. That is the massing version, it is the cheapest to produce, and it answers the questions being asked at that stage.

At sales launch, the detail requirement rises because buyers are standing closer and asking about materials and openings. That is an upgrade to the same model rather than a new commission, provided the model was built to be extended.

During construction, the experience gets cheaper rather than more expensive, because the real structure supplies context and tracking references that the model previously had to carry alone.

Planning that sequence at the start, and contracting for a model you own, is worth more than any negotiation on the first invoice.

A note on comparing against renderings

Buyers frequently benchmark augmented reality against a rendering price and conclude it is expensive. The comparison is unfair in both directions.

Renderings are cheaper per asset and reach everybody. Augmented reality is more expensive per asset and reaches only visitors, but it answers the scale question that renderings answer worst.

The efficient position for most projects is renderings for reach and one augmented reality experience for the moment somebody is standing on the site, produced from the same model so the second is a derivative rather than a new commission.

Frequently asked questions

Why do augmented reality quotes differ so much for the same brief?

Because vendors make different assumptions about who builds the 3D model. A quote to publish an existing model is a small job. A quote to build the model from drawings is comparable to a rendering package. Ask every vendor to price model production as a separate line.

Can we reduce cost by supplying our own model?

Only if it is a display model. A coordination model built for construction carries far too much geometry for a phone and has to be rebuilt, which is most of the work. Supplying one helps as a reference, but it rarely removes the optimisation cost.

Is surveyed GNSS placement worth paying for?

It is worth it whenever somebody will make a decision based on where the model appears to sit, such as clearance, setback or finished floor elevation. For marketing and communication of scale, phone based placement is sufficient and the extra cost is avoidable.

What ongoing costs should we expect?

Hosting and delivery, periodic updates as mobile operating systems and augmented reality frameworks change, and model updates when the design changes. Budgeting nothing for year two is the usual omission.

Is augmented reality cheaper than a physical scale model?

Generally yes, and it can be updated when the design changes, which a physical model cannot. The trade is that a physical model works in a sales office for everyone who walks in, while augmented reality requires the viewer to be at the site.