- Why most Veo product ad prompts fail
- The five things every good Veo product prompt needs
- Step by step: building your prompt before you touch Veo
- What Veo is bad at, so stop asking
- Prompt templates you can adapt
- Where this fits in your wider content workflow
- A realistic production checklist
- Frequently asked questions
The short version: Veo prompts for product ads work best when you treat them like a shot list, not a wish list, naming camera movement, lighting, surface, timing and what the product is doing on screen. Vague prompts get you a wobbly, generic-looking advert. Specific ones get you something you could plausibly run as a paid social clip.
I have spent a stupid amount of time this year testing AI video tools for clients who want quick product content without booking a studio. Veo is good at this if you stop talking to it like a creative director giving notes in a meeting and start talking to it like you're writing a camera operator's instructions.
Why most Veo product ad prompts fail
The number one mistake I see is people typing something like "a stylish product video for my skincare brand, modern, premium feel." That prompt has zero information Veo can use. There's no camera, no lighting direction, no timing, no surface, no action. You'll get something that looks fine for about two seconds and then falls apart, usually with the label warping or the product floating unnaturally.
Here's the uncomfortable bit nobody wants to say out loud: Veo cannot read your brand guidelines, your mood board or your mental image. It only knows what's in the text box. If your prompt doesn't specify it, Veo will guess, and its guesses tend toward stock-footage beige. Lighting defaults to flat and even. Camera movement defaults to a slow, lifeless push-in. Backgrounds default to a blurred studio void that looks like every other AI video you've already scrolled past.
So the fix isn't a magic prompt phrase. It's structure.
The five things every good Veo product prompt needs
- Camera behaviour: static, slow pan, handheld, orbit, dolly in, macro push. Name it.
- Lighting: golden hour, hard studio light with visible shadow, soft overcast window light, neon rim light.
- Surface and setting: wet concrete, marble counter, linen tablecloth, kitchen counter with morning light.
- The action: what does the product do, or what happens to it, in the shot. A lid opening. A pour. A hand picking it up.
- Timing and pacing: how long the shot runs and where the beat lands, for example "the label becomes readable at the two second mark."
Miss two or more of these and you're gambling. Include all five and you're directing.
A worked example
Say you run a small coffee brand selling bagged beans online, and you want an 8 second Veo clip for an Instagram ad. A weak prompt would be: "coffee bag on a table, cinematic, product ad." Here's what a structured version looks like instead:
"Static camera, eye level, shallow depth of field. A matte black coffee bag with a gold foil label sits on a dark walnut counter. Soft morning window light from the left casts a long soft shadow to the right. A hand enters frame from the right at second 2, picks up the bag, tilts it slightly so the label catches the light, and sets it back down by second 6. Steam rises gently from an out of focus cup in the background. Warm colour grade, slight film grain, no text overlay."
That's six sentences and it gives Veo a camera, a light source, a surface, an action with a clear timestamp, and a mood. You'll still need two or three attempts to get the hand movement looking natural, that's normal, but you're working from something directable rather than rerolling a vague prompt and hoping.
Step by step: building your prompt before you touch Veo
- Write one sentence describing the product and its single most sellable physical detail (the foil label, the pump nozzle, the stitching).
- Pick one camera move. Not three. One. Mixing camera instructions confuses the model and you'll get a shot that can't decide what it's doing.
- Name your light source and direction, not just a mood word. "Warm" isn't a lighting instruction. "Low afternoon sun from camera left" is.
- Describe one physical action tied to a rough timestamp, even if Veo doesn't hit it exactly.
- End with a short style note: colour grade, grain, lens type. Keep it to one clause so it doesn't compete with the main instruction.
That five-step order matters more than people think. If you lead with style words, Veo will prioritise mood over physical accuracy, and you'll end up with a gorgeous looking clip where the product itself looks slightly wrong, like the cap doesn't quite sit right or the bottle proportions shift halfway through.
What Veo is bad at, so stop asking
Readable small text is still unreliable. If your product's selling point is the ingredient list or a specific number on the label, don't rely on Veo to render it, add your text and branding afterwards in editing. Also don't ask for more than one product interaction per shot, "the hand opens the lid, pours the liquid and then wipes the counter" is three actions and Veo will usually botch the handoff between them. One action per clip, cut your ad together from multiple short clips instead of one long continuous take.
This is the part most "10 Veo prompts" listicles skip, because admitting the tool has limits is less fun than pretending there's a perfect prompt out there. There isn't. There's a workflow: generate several short, tightly specified clips, then edit them together, same as any real product shoot uses multiple setups rather than one magic take.
Prompt templates you can adapt
For a cosmetics or skincare product: "Slow 45 degree orbit around a glass dropper bottle on a white marble slab, soft diffused studio light from above, a single drop forms at the dropper tip by second 3, shallow depth of field, clean minimal background, soft pastel colour grade."
Want AI doing the heavy lifting in your marketing?
I build the systems that handle the boring 80 percent, so you get your week back. Done properly, with the human kept in.
For a food or drink product: "Handheld, slightly shaky camera for authenticity, kitchen counter setting with warm tungsten light, a hand twists open a jar lid at second 1, steam or condensation visible, natural unedited colour grade, documentary feel."
For tech or hardware: "Macro camera, static, hard rim lighting against a dark background, a finger presses a button on the device at second 2 triggering a soft blue light glow, high contrast, commercial lens look."
Adapt these by swapping the action and surface, not the whole structure. The structure is doing the work.
Where this fits in your wider content workflow
Veo is one tool in a stack, not the whole stack. For the photography stills to sit alongside your video, I'd pair it with Midjourney prompts for product photos so your static and moving assets match in lighting and mood. If you're building out a fuller video library, it's worth comparing against Sora prompts for product videos since the two tools handle motion differently and some products will suit one over the other. And once you've got the visuals, you still need the words, whether that's product descriptions for an Etsy listing or a fuller set of ecommerce product description prompts to go underneath the ad.
If you're running this across dozens of SKUs and it's eating your week, that's usually the point where it makes sense to get real help setting up repeatable prompt templates rather than rebuilding each one from scratch, something an AI consultant for small business can set up once so your team isn't guessing every time.
A realistic production checklist
- Write the prompt using the five elements above before opening Veo, not while staring at the box.
- Generate 3 to 4 variations of the same prompt, Veo's output varies clip to clip even with identical text.
- Keep each clip to one action and under 10 seconds for reliability.
- Plan to stitch 3 to 5 short clips together rather than expecting one long perfect take.
- Add text overlays, logos and legal copy in editing, not in the prompt.
- Match your colour grade across clips in post so the stitched ad looks like one shoot, not five different ones.
Frequently asked questions
Can Veo generate a full 30 second product ad in one prompt?
Not reliably. You'll get better, more consistent results generating several short clips of 5 to 10 seconds each with one clear action per clip, then editing them together, rather than asking for one long continuous shot.
Why does my Veo product video distort the label or logo?
Veo still struggles with rendering small, readable text accurately. Keep labels generic or blurred in the generated clip and add your real logo and text as an overlay in editing afterwards.
How specific does a Veo prompt need to be for product ads?
Specific enough to cover camera movement, lighting direction, surface, one physical action with rough timing, and a one-clause style note. Vague mood words like "premium" or "cinematic" without physical detail produce generic, flat results.
Is Veo better than Sora for product ads?
It depends on the product and the motion you need. Some products suit Veo's handling of lighting and texture better, others look more natural in Sora, so testing the same prompt structure in both is worth the extra half hour before committing to one tool.
Run a business, a brand or a blog? I feature tools, products and expert guest posts for an audience of 15,000. See how it works