The Brand Avatar System

The arithmetic that governs every ad account

A performing creative stops performing in roughly two to three weeks. The audience has seen it, and the numbers decay. That is normal, and it is not a quality problem.

A conventional campaign shoot takes six to eight weeks from brief to delivered assets. Which means the production model cannot, structurally, refill the feed as fast as the feed empties. The arithmetic simply does not close.

Most brands respond by re-cutting the same shoot until it stops working, then booking another one. That is not a creative failure. It is a supply problem wearing a creative costume.

Where the weeks actually go

Note that the shoot itself is one to three days of that. The weeks go into casting, booking windows, contracting, usage negotiation and two rounds of client approval. Which is why reshooting a single collection does not cost you a day — it costs most of the six to eight weeks again.

Four stages, two of which happen once

The alternative is not a cheaper shoot. It is taking the shoot out of the path between a new product and a finished creative — by building the face once, as an asset you own, and producing against it indefinitely.

Each stage has a defined input, a defined output, and a gate it has to pass before the next one starts.

It is worth knowing what the conventional route costs on this axis too. A model's day rate buys the shoot, not the pictures — usage is priced separately across term, media and territory. On published Gulf agency rates, one year of GCC digital and social rights adds roughly 155% to the base fee per model, and perpetual rights run 200–300% on top. Whoever is the face of your brand is therefore a renewable expense on a clock that started the day the work was delivered.

What the data actually supports — and what it does not

You are probably asking whether AI-produced creative performs. The honest answer is more useful than the enthusiastic one.

Read honestly, that is parity — not a conversion advantage. No credible independent data shows AI imagery converting better than a good photograph, and anyone telling you otherwise is selling.

So the argument is not that this creative is better. It is that you can have more of it, sooner, at a unit cost that lets you test properly — and that volume and cycle time are themselves performance levers. Zalando's own executive framed it exactly that way, explicitly declining to claim the content was superior.

Two further things we will not pretend about. Product categories with hard geometry — eyewear, watches, footwear, jewellery — are materially harder than apparel, and case studies drawn from clothing do not transfer to them. And creative can only be optimised against numbers somebody is collecting, so if your pixel coverage is partial, that is step zero rather than an afterthought.

The gate every asset clears

"Not noticeable as AI" is, in practice, a geometry problem rather than a skin problem. In eyewear a frame has to sit on a nose, clear a brow, pass over an ear and reflect one light source. Every one of those is somewhere generated imagery fails visibly, and every product category has its own version of the list.

So each image clears a fixed gate before it ships:

• Your product, not an approximation — built from your own photography
• Fine detail survives — engraving, stitching, hardware, wherever the crop shows it
• The geometry holds — contact points, symmetry, nothing floating or passing through
• One light, one reflection, matched to your grade
• Only your marks — no invented wordmarks, no added hardware
• Hands hold at full resolution, or the shot is cut rather than shipped

A shot that fails one of these does not get retouched into place. It gets cut and re-run, because the cost of re-running is now measured in minutes.

What a decode usually surfaces

The specification is the deliverable, but the useful part is often what shows up on the way to it. Three findings recur so often they are worth naming.

One — the campaign contradicts its own line

A brand writes a tagline about connection, then shoots eighty frames in which nobody looks at the camera. The most ownable image available is usually the one the brand has told you it wants and never taken.

Two — one product line is marketed with the logic of another

In the case above, optical frames sat in the same visual language as sunglasses. But that buyer is making a different decision: nine hours a day, weight, comfort, how they look on a video call — not glamour. Same range, different purchase, and usually the highest-return change available without touching the product.

Three — there is campaign imagery, but no performance creative

Beautiful assets that establish a brand is beautiful, and not one of them making an argument. Nothing about what the thing is made of, what it costs relative to the alternative, or why it exists. In the case above, the strongest sales argument the brand owned was engraved on the product — and no image in the library showed it.

Brand imagery and performance creative are different products with different jobs. Most brands have one of the two, and buy more of the one they already have.

Where the arguments come from

An avatar that matches your house look is table stakes. It is not what makes a creative earn money. What determines that is the angle — the argument the ad makes — and the best ones are almost never invented.

We derive them, in this order:

• Your own product copy and packaging. What is already printed, stamped or engraved on the thing.
• Your price against the category. Not against your own range — a narrow internal price spread is not an angle.
• Your reviews. Where use cases you never marketed show up, in your customer's words.
• Comments under your existing ads. You already paid for that engagement; almost nobody reads it.
• The purchase itself. How the product is genuinely used, which is often not how it is shot.

Each angle is paired with a hook direction, a format and a funnel position, then tested by holding the angle constant and changing the format. That is the only way to learn whether the argument failed or the format did.

How this runs once it is running

Six angles across three formats is eighteen concepts. Through a conventional shoot that is a budget conversation. Through stage three it is a production run.

What we would need from you

Short list, and most brands already have all of it.

• Product photography — clean, on neutral ground, front, three-quarter and profile. This is the one non-negotiable, because the product is composited from it rather than generated.
• Your existing campaign assets, whatever exists. The specification is decoded from these.
• Brand guidelines if they exist. If they do not, you get a written specification out of stage 1 either way.
• Your best and worst performing creative, with the numbers attached. This tells us more than guidelines will.
• Where the assets run — feed, story, site, or all of it — so ratios are right the first time.
• Pixel and analytics access, or a clear account of what is currently instrumented.

What we do not need: a shoot, a casting call, a booking window, or a usage negotiation.

What a first call actually covers

1. Your range and cadence. How many products, how often you drop, and how fast creative currently fatigues on your account.
2. What is instrumented. Which channels you can read today, and which you are spending on blind.
3. Casting. Who your customer actually recognises — which is a strategy question long before it is an aesthetic one.
4. Which products lead the first cycle, and on what argument.
5. Who signs off, and how fast. Approval speed sets the cadence more than production does.

If the arithmetic at the top of this document describes your account, the next step is a conversation.

Book a 30-minute call