Weak, missing, or duplicate product descriptions can get products disapproved or buried in Shopping. In Renable you fix this with a prompt that writes a fresh description from each product's own data — its title, attributes, and (with the image-based prompt type) even its photos. Aimée generates the copy, you review it, and it flows into your feed. Generating content spends credits.
Rewrite Thin or Duplicate Descriptions with AI
Weak, missing, or duplicate product descriptions hurt Shopping. Fix them with a prompt that writes a fresh description from each product's own data — Aimée generates the copy, you review it, and it flows into your feed. Generating content spends credits.
Updated September 21, 2026
Why thin or duplicate descriptions hurt your ads#
Ad networks like Google Shopping and Meta expect every product to carry a real, product-specific description. A description that's missing, one line long, stuffed with keywords, or copied word-for-word across every variant gives them little to work with. The networks' own rules are specific: Google gives a product with no description a warning and says it "may experience limited performance", allows up to 5,000 characters and advises putting the most important details in the first 160–500 (Google's description attribute); Meta allows up to 9,999 characters, asks for sentence case rather than all caps, and recommends a description longer than the title (Meta's product description specifications).
Rewriting them by hand doesn't scale — especially across a catalog of hundreds or thousands of products, or across many client accounts. Renable's AI content generation writes a unique description for each product from that product's own data, so you fix the whole group at once and keep fixing new products as they arrive.
Write fresh descriptions with a prompt#
In Renable, AI descriptions are written by a prompt — a reusable recipe that tells Aimée what to write and which products to write it for. A "Write descriptions" prompt reads each product's own information (its name, brand, category, and existing description) and produces new ad copy, one description per product. Because it works from each product's own attributes, no two products get the same generic text.
This article covers the job — turning weak descriptions into strong ones. The prompt editor itself (instructions, steps, scoping, model choice) is covered in Working with prompts; link down to it for any mechanics below.
- Create a new prompt and choose the Write descriptions type — it pre-fills instructions and the right content category, so you're not starting from a blank page.
- Tune the Instructions for your brand — tone of voice, what to emphasize, what to leave out — and add an instruction to write in each product's own language if you sell across markets.
- Scope it with Use as to a product set of just the items with thin or duplicate copy, so you're not rewriting everything.
Then preview the output on a real product and check the estimated credit cost before you roll it out — Working with prompts covers those create, preview, and credit-estimate mechanics.
When the existing text is too thin to work from#
If a product's feed description is empty or barely there, there's little for a text prompt to build on. For those products, use the Write descriptions by looking at the product image prompt type instead. It feeds the product's images into an image-understanding model, so Aimée describes what it actually sees in the photos alongside the product's specifications — useful exactly when you have images but no usable text. See Advanced prompting techniques for how that prompt is built.
When descriptions are duplicated across variants#
Duplicate descriptions usually come from a feed that gives every size or color of a product the same block of text. A "Write descriptions" prompt fixes this by writing from each product's own attributes, so the outputs differ naturally instead of repeating. If you want every product to match a specific house style while staying unique, prime the prompt with an example of your ideal copy — the priming technique in Advanced prompting techniques makes Aimée mimic your tone and formatting on every product.
Review the new descriptions, then they reach your feed#
AI-written descriptions become content on each product — the same raw material every other AI output becomes — and they start out unapproved. Review them on your project's Content page, where you can read what Aimée wrote per product and approve it. See Reviewing and approving content for the review flow and hotkeys.
By default, unapproved content is used in your templates and channels exactly like approved content, so a new description can start flowing into your feed as soon as it's generated — you don't have to approve everything first. If you want tighter editorial control, turn on Only use content approved by a human on the relevant placeholder or channel field, and that surface will skip anything you haven't reviewed. Either way, the generated descriptions are stored alongside your feed's original data — they don't overwrite your source feed.
What this costs in credits#
Generating content spends credits, charged per product the prompt runs on. Text prompts like "Write descriptions" are among the cheaper operations — far less than image or video generation — but the total still scales with how many products you generate for, which is why scoping the prompt matters. The Generate preview shows an estimate before you apply the prompt at scale. See Understanding credits and What uses my credits.
Or ask Aimée#
You don't have to build the prompt by hand. Tell Aimée what you're after — "write short, benefit-led descriptions in my brand's voice" — and she can create or adjust the prompt for you: its instructions, which model runs it, and its steps. She can create it on the right project, but you set the Use as scope yourself — point it at the product set of items that need fixing. She can't run the generation herself either, since that spends credits, so once it's set up she'll point you at Generate to preview and roll it out.
FAQ#
Will this overwrite the descriptions in my product feed?
No. AI-written descriptions are stored as content on the product, alongside your feed's original data. Your channels use the generated copy where it's configured to, but your source feed stays untouched.
Do I have to approve every description before it can go live?
No. By default, unapproved content is used just like approved content, so generated descriptions can reach your feed without a full review. Turn on Only use content approved by a human on a placeholder or channel field when you want that surface to wait for your approval. See Reviewing and approving content.
How do I rewrite only the products with weak descriptions, not all of them?
Scope the prompt with Use as — point it at a product set that contains just the products you want to fix. See Product sets and product queries.
Do the descriptions update when my feed changes?
Yes. When a product changes, Aimée reprocesses it and the prompt runs again on the new data. New products that match the prompt's scope get a description automatically on import. And when you publish a brand-new or edited prompt, every product in its scope is reprocessed within minutes — including the ones already processed — so the new wording reaches your whole feed without waiting for products to change.