The ad was gorgeous. A dog harness on a golden retriever, soft morning light, a meadow so perfect it looked like it had its own skincare routine. People clicked. People liked. People shared it with their friends.
Then people bought it, opened the box, and found a harness that looked noticeably less impressive than the one in the meadow. The returns came in, the reviews got awkward, and the brand learned an expensive lesson. A version of this story did the rounds among marketers recently, and it stuck because so many sellers recognised themselves in it.
It’s a useful story, because generated product images aren’t a fad. They’re cheaper, faster and more flexible than a traditional shoot, and for a lot of brands they’re genuinely the right call. They just come with one rule that’s very easy to break.
The photo shoot problem is real
If you sell physical products, you know the drill. Ad platforms reward fresh creative. Every new campaign, season, channel and format wants its own images. A square for the feed, a tall one for stories, a seasonal version, a sale version, a “new colour” version. A shoot gives you a few dozen good frames, and your ads chew through them faster than you’d like.
So you book another shoot. Even basic product shots can run a few hundred per session, and anything with models, locations or styling costs a lot more. For a small brand, that means either spending a painful share of the marketing budget on photos, or running the same three images until your audience could draw them from memory.
Where generated creatives shine
This is where generated images earn their keep. Take a good photo of your product and you can put it on a kitchen counter, a beach towel, a desk at golden hour or a snowy windowsill in December. You can produce every format a platform asks for. You can test ten different angles to see which one makes people stop scrolling before you spend a cent of your ad budget on the winner.
For backgrounds, seasonal scenes, lifestyle context and sheer volume, a good generated workflow beats a shoot on cost almost every time. It also changes how you work. You stop rationing creative and start testing it.
The one rule: never touch the product
The harness story happened because the generator didn’t just build a world around the product. It quietly “improved” the product itself. A slightly richer colour, a neater stitch, a better fit. None of it was dramatic. All of it made the photo better than the real thing.
That’s the trap. An image that flatters the product brings in clicks from people expecting something that doesn’t exist. Then you pay twice: once for the ad and again for the return, the refund and the one-star review that says “doesn’t look like the picture.” The ad wrote a cheque the product couldn’t cash.
In practice, that means a few simple habits:
- Start from real photos of the real product. Not a description, not a render, the actual item.
- Generate the world, not the item. Scenes, light, backgrounds and props are fair game. Colour, shape, texture and size are not.
- Check it against the thing in your hand. Put the physical product next to the screen before anything goes live. If someone would notice the difference, so will your customers.
- Judge creatives by sales, not likes. A beautiful image that gets engagement and returns is a worse ad than a plain one that sells.
Test it before you bet on it
The nice thing about generated creatives is that you don’t have to take anyone’s word for them, including ours. Take your current best-performing ad and put a handful of generated versions up against it, with the same product, the same offer and the same audience. Give them a small budget and let them run.
Then look past the likes. Compare the cost per purchase, not the cost per click. Keep an eye on returns and on the comments, because nothing tells you an image is off quite like a customer asking “is this the same product?” If the generated versions sell as well or better, without a jump in returns, you’ve just found a cheaper way to keep your ads fresh. If they don’t, you’ve lost very little and learned a lot.
Most brands that run this test end up somewhere in the middle. Some styles work brilliantly, some fall flat, and the winners are rarely the ones anyone on the team would have picked. Which is exactly why you test.
When you should still hire the photographer
Generated creatives aren’t the answer to everything, and pretending otherwise is how you end up in a meadow. A few cases still deserve a real shoot. Your flagship brand campaign, the images that define what your brand looks like. Products where texture or fit is the buying decision, like clothing, jewellery or food. And anything that needs real people genuinely using the product, where authenticity is the whole point.
Most sellers settle on a sensible mix. They shoot the hero images properly, then use generated creatives for everything around them: variations, formats, seasons and tests. You get the quality where it matters and the volume where it counts.
The golden retriever, by the way, was never the problem. Neither was the meadow. Both can be generated, and both probably made the ad better. The harness just had to be the harness.
This is how we make ad creatives at Strale: the real product, a generated world around it, judged by sales. And when a proper shoot is the better call, we’ll be the first to tell you.