Know what to check
Would You Approve This Ad? is a free two-minute challenge from AI Vidia. Compare six ads with their approved product facts and decide: ready for product review, a mismatch to fix, or not enough evidence. It trains the check that stops a beautiful ad from showing the wrong product.
Why a beautiful ad still needs a product check
A buyer orders the product in the ad, not the one on your shelf. Ecommerce usability research from the Baymard Institute found that 42% of users try to judge a product's size from its images, so size, parts and labels are product facts, not styling.
Generated and edited scenes add a pocket, drop a pipette or change a word on a label while everything else looks right. A review against approved facts catches that; a gut check on how premium it looks does not.
- Ready for product reviewThe product matches every approved fact.
- Fix a specific mismatchSomething in the ad breaks a fact.
- Not enough evidenceThe references cannot settle it.
The six checks
Each ad tests one habit. Your result names the first one you missed.
- 01Claims against the pack
- Read every number and promise in the copy against the pack and the fact sheet.
- 02Parts and components
- Count the parts: caps, pipettes, straps, pieces in the box. Generated scenes add and drop them.
- 03Seeing when it is right
- An unusual style is not a defect. Approve what matches every fact, however stylised.
- 04Product details
- Compare pockets, seams, buttons and trims with the reference, not with what looks plausible.
- 05Size and scale
- If nothing shows the size, ask for the measurements and an in-scale photo instead of guessing.
- 06Label consistency
- Check the label in every frame of a video, not only the first one.
Answer key and explanations
Spoilers: every ad, the right call and why. Also the text version of the challenge.
Ad 1: HOLTBY, Dog food, salmon and oat
- Name:
- HOLTBY Salmon & Oat
- Net weight:
- 2 kg
- For:
- Adult dogs
- Pack:
- Kraft pouch, green label
“Our new 3 kg bag. Salmon and oat for grown-up dogs.”
Answer: Fix a specific mismatch (net weight)
The reference and the pack in the photo both say 2 kg; the headline says 3 kg. The words on an ad are part of the product check: a size the pack cannot back up is a mismatch, however good the photo looks.
Ad 2: HAVREL, Face serum
- Name on the label:
- HAVREL Daily Drop
- Bottle:
- Amber glass with a cream label
- Dropper:
- One glass pipette with a black bulb
- Collar:
- Ribbed black collar
Answer: Fix a specific mismatch (dropper)
The bottle is open and its pipette is in the hand, so the pipette still standing inside the bottle is a second one. The one in the hand also narrows to a tapered tip. This frame comes from a real production test on our fictional HAVREL serum, where both blind reviewers rejected it.
Ad 3: HAVREL, Face serum
- Name on the label:
- HAVREL Daily Drop
- Bottle:
- Amber glass with a cream label
- Dropper:
- One glass pipette with a black bulb
- Collar:
- Ribbed black collar
Answer: Ready for product review
Hard light and a stone block make it look unusual, but every fact holds: one bottle, amber glass, cream label, black collar and one pipette. A stylised scene is not a defect. In the HAVREL production test this frame passed both blind reviews.
Ad 4: ARNVIK, Waxed cotton field jacket
- Colour and fabric:
- Olive waxed cotton
- Collar:
- Brown corduroy
- Pockets:
- Two hip pockets with flaps; no chest pockets
- Closure:
- Brass front zip
Answer: Fix a specific mismatch (pockets)
The jacket in the ad has two chest pockets. The reference has none: only the two hip pockets. Everything else matches, which is exactly why an added detail slips through. A buyer who orders this jacket gets a different one.
Ad 5: PELTO, Ceramic table lamp
- Base:
- Matte sand glaze
- Shade:
- Linen drum shade
- Switch:
- Brass, on a cream cord
- Height:
- Not supplied
“Small enough for any nightstand.”
Answer: Not enough evidence
The headline makes a size claim, but nothing in the references or the frame gives the lamp's size. The honest call is to ask for the measured height and a photo in a real room before approving, not to guess.
Ad 6: OKSELA, Night cream
- Brand:
- OKSELA
- Product on the label:
- Night Recovery Cream
- Size:
- 50 ml
- Lid:
- Brushed gold
Answer: Fix a specific mismatch (product on the label)
At 0:02 the label says Night Recovery Cream; at 0:09 it says Day Recovery Cream. Each frame looks fine on its own, so check every frame against the reference, not just the first one.
What the challenge is not
- It is not an AI detector. Real footage can be wrong and a stylised image can be right.
- A product-accuracy call is one step. Brand, legal and platform checks come after it.
- The score covers these six fictional ads only. It is not compared with anyone else.
Questions about checking ads
How do you check an ad before it goes live?
Put the approved product facts next to every frame: name and label text, size, parts, colours and every claim in the copy. Mark each one as matching, wrong or not shown, and fix or ask before approving.
Is this a test for spotting AI images?
No. Whether a frame was filmed, shot or generated does not decide the call. The question is whether the ad shows the product the buyer will receive, and whether the references can prove it.
When is not enough evidence the right answer?
When the ad claims something the references cannot settle, such as a size with no measurements or a result with no documentation. Ask for the missing reference instead of approving or rejecting on a guess.
Where do these ads come from?
AI Vidia made them for fictional brands. Two frames come from our HAVREL production test, a fictional serum, with the verdicts our blind reviewers gave them. None of these products exist.
Can we use it with our team?
Yes. Send your colleagues the challenge link; everyone gets the same six ads, so the scores compare fairly. The result card shows a score and a blind spot, never the answers.