Six years of AR campaigns:
what I would tell myself at the start

The first AR campaign I directed went live on the wrong platform, at the wrong moment in the user journey, with an effect that made about three seconds of visual sense and then became noise. It reached its numbers. The impressions were there. Nobody complained. And I knew, watching real people use it, that it had not done what we had hoped it would do.

Nobody in the debrief said this out loud. We talked about the numbers. We discussed what we might do differently with the pacing. We moved on. But I had seen the expressions on the faces of people trying it for the first time, and the expression that kept repeating was not delight. It was polite confusion followed by moving on.

That campaign taught me more than the ones that worked. What follows is the set of things I wish I had known at the beginning: six lessons from six years of directing AR campaigns, 1.5 billion impressions, and a growing certainty that the technical floor does not matter if the conceptual foundation is wrong.

Lesson 1: the brief is the most expensive part

Every hour of vagueness in the brief costs four hours of rework in production. I have seen this pattern so many times now that I treat it as a law. A brief that uses the word "immersive" without defining what that means will generate at least two rounds of conceptual revision after production has started. A brief that uses the word "wow" as a success metric will not have a success metric.

The question I ask every client now, before any concept work begins, is this: what do you want someone to say when they describe this experience to a friend ten minutes after using it?

That question does more brief-clarification work than any creative strategy document. It forces specificity. "I want them to say it was cool" is not an answer. "I want them to say they felt like they were actually inside the world of the show" is an answer you can design toward. "I want them to say they have never done anything like it before" is a brief that tells you the experience needs novelty as its primary driver, which has specific implications for platform choice, interaction model, and how much of the budget goes to the effect versus the content.

The brief is not a form you fill in before the real work starts. It is the most important creative decision the project makes. Get it right and everything downstream becomes easier. Get it wrong and nothing you do in production will fully recover it.

Lesson 2: the platform constraint is a creative brief, not a limitation

Early in my career I spent months fighting Snapchat's 90-second filter limit. I wanted longer experiences. I wanted audiences to stay. The 90-second cap felt like a cage. I wrote feedback about it. I lobbied internally. I worked around it in ways that made the product worse, not better.

Then I stopped fighting it and started designing for it.

Ninety seconds is enough time to do one thing well. It is not enough time to do three things adequately. When the constraint forced us to choose one thing, the one thing we chose was always better than the three things we had been trying to fit in. The platform was telling us something about attention, and we had been ignoring it because we thought we knew better.

Every platform constraint is a design insight. TikTok's vertical format is not a limitation for brands who were making horizontal video: it is information about how the audience holds their body when they engage with content, and designing for that posture produces better work. Snap's filter system is not a delivery mechanism you squeeze concepts into: it is a set of creative rules that, when followed, produce formats the audience already knows how to use. Meta's AR glasses run without a touchscreen, which means every interaction has to work with voice or hands, which means your interface has to be intuitive enough that nobody needs to read instructions while wearing it.

I stopped fighting platform constraints the year I started winning with them. That is not a coincidence.

Lesson 3: impressions taught me almost nothing useful

One campaign I worked on reached 1.5 billion impressions. That number appears in decks, in press releases, in awards entries. It is genuinely large. I am proud of it.

It taught me almost nothing about whether the experience had worked.

Impressions measure exposure. They do not measure what the person who was exposed did, thought, or felt. An impression can be a person who saw a thumbnail for 0.3 seconds while scrolling and did not stop. It can be a person who used the filter for eleven seconds and then immediately sent it to four people. The metric treats both the same way.

The metric that actually tells you whether an AR experience landed is simpler and harder to hit: did anyone come back voluntarily a second time? Not because of a retargeted ad or a notification. Voluntarily. Because they wanted to use it again.

Return rate is not a standard metric on most platforms. You have to ask for it or proxy it from session data. But it is the only number that tells you the experience had enough intrinsic value for someone to seek it out again. Everything else can be gamed by distribution spend. Voluntary return cannot.

I now ask for return rate in every brief. Clients sometimes look at me as if I have said something strange. That is usually a sign the brief has not yet defined what success actually means.

Lesson 4: creative direction for AR is choreography, not filmmaking

If there is one thing in this list that I wish someone had said to me before I directed my first AR campaign, it is this.

The mental model that slows most creative directors down when they first work in AR is the filmmaking model. Film has a frame. The audience looks at what you put in the frame. Control over what the audience sees is near-total. You write the script, you shoot the scene, you edit the cut, and the audience watches what you made.

AR does not work like this. There is no frame you control. The audience is in a space, and you are designing what happens in that space when they interact with it. You are not directing their eyes to a specific image. You are designing the conditions under which they discover something, reach for something, turn their head, take a step, lift a hand.

That is choreography. You are writing a score for a body, not a script for a camera.

When I made the shift from thinking about AR as "content people watch with their phones" to "an environment people move through," the work changed. I started designing for physical behaviour first and visual aesthetics second. I started thinking about what the audience's hands were doing. I started asking where in the physical space the interesting moment would happen, and working backward from that to what we needed to build to create it.

The best AR creative directors I know all have this instinct, whether or not they describe it as choreography. They think about the audience's body before they think about the asset.

Lesson 5: AI changed the production floor, not the director's job

AI tools transformed what is possible in AR production. Effects that used to take three weeks to build now take three days. Generative backgrounds, real-time style transfers, AI-driven face and body tracking: these are table-stakes capabilities now, not differentiators. Any competent studio can deploy them.

This changed something important about creative direction: the production floor got faster, which means the quality ceiling is no longer set by what you can build. It is set by what you decide to build. The director's job got harder because the constraint of "we cannot make that" was removed. Now every decision about what to make is a pure creative and strategic choice. There is nowhere to hide behind production limitations.

I use AI tools heavily in production now. I also spend more time in brief development and concept validation than I ever did before, because the production speed means you will build the wrong thing faster if the concept is not right. The tools accelerate the consequences of both good and bad conceptual decisions.

The thing AI tools have not changed: whether something resonates. Resonance is an emotional event. It happens when an experience makes someone feel something they did not expect to feel. You cannot generate resonance. You can only design the conditions for it and then do enough user testing to know whether you have succeeded before you ship.

Lesson 6: the gap between impressive and compelling is almost always emotional

I have been in rooms where people put on AR experiences and their eyes went wide. I have also been in rooms where people put on AR experiences and their eyes went wide, and then they put the device down, and nobody talked about it on the way out.

The difference is emotional, not technical. An experience that impresses you has shown you something you did not know was possible. An experience that compels you has made you feel something you did not expect to feel. The first produces a response that lasts about thirty seconds. The second produces a response that produces conversation, sharing, and the kind of word-of-mouth that no media budget can replicate.

The technically impressive thing is often the enemy of the emotionally compelling thing. When the team is in love with what the technology can do, the experience becomes a demonstration. When the team is in love with what the audience will feel, the technology becomes invisible. The best AR experiences I have seen are the ones where, five minutes after using them, you cannot fully explain how they worked. You just remember how you felt.

This is the thing I keep coming back to when I brief creative teams now: start with the feeling, not the feature. What do we want someone to feel? Everything else is in service of that.

Where this leads: Noodle, wearables, and the next chapter

Six years of AR campaigns on phones built an instinct for what resonates with audiences at scale. Winning MIT Reality Hack 2026 with Noodle on Snap Spectacles taught me what changes when you move AR from a screen people hold to a device people wear.

Almost everything changes. The choreography becomes more intimate. The context is no longer "someone looking at their phone in a moment of leisure." It is "someone doing something in the world, with information or content available in their field of view as they do it." The design problems are harder. The resonance ceiling is higher. And the brief question is the same: what do you want someone to say when they describe this to a friend?

What I am most interested in now is the user moment where taking out a phone creates friction. That is the moment wearable AR owns. The technician whose hands are occupied. The designer who is standing in the space they are designing for. The person who needs a creative partner without leaving the creative flow.

Those moments are where the next six years of interesting work is. I am only just getting started on them.

If you are thinking about what AR or spatial computing could do for a specific moment in your users' experience, that is the conversation I want to have. You can find more about how I think about creative direction for immersive campaigns, or reach out directly.

Frequently asked questions