Below is a practical, reusable guide — with ready-to-use prompt templates and examples — to help you write “perfect” AI prompts for generating PrestaShop product content, category descriptions, and CMS pages. Use these templates with ChatGPT (or any LLM) and adapt the fields/constraints to your store and data import format (CSV/JSON/PrestaShop import).
Short checklist before you prompt
- Provide context (who you are / store / target audience).
- Give the LLM the exact output structure you need (JSON/CSV/HTML/text).
- Give required fields, optional fields, and formats (word counts, HTML allowed, units).
- Provide examples and one perfect sample output if possible.
- Specify tone, SEO keywords, locale/language, and any brand voice rules.
- Tell the model to not invent product specs (or to mark guessed specs clearly).
General prompt principles
- Role + Objective: Start with a system-like role: “You are an e-commerce copywriter and SEO specialist.”
- Data input: Paste the product’s factual data (title, specs, price, attributes) to ground the model.
- Output format: Be explicit — JSON keys, CSV columns, HTML snippets, or free text.
- Constraints: word counts, SEO/meta lengths, banned words, HTML allowed/disallowed.
- Examples: Provide 1–2 examples of desired output.
- Error handling: Ask the model to mark missing facts as “TBD” or leave empty rather than inventing.
Product content: fields you’ll likely need
- name (product name)
- short_description (HTML or plain text, 50–120 words)
- description (full HTML, 150–400+ words)
- features / bullets (5–10 bullet points)
- meta_title (<= 60 chars)
- meta_description (120–160 chars)
- seo_keywords / tags
- image_alt_texts (one per image)
- category_suggestion
- price / currency
- reference / SKU, EAN13/UPC
- condition (new/used)
- combinations/variants (color, size, SKU per variant)
- attributes/specs (weight, dimensions, material, care)
- stock_quantity / available_now_text
- warranty / manufacturer
- shipping info (size/weight notes)
- visibility (catalog/search/both)
- customization/attachments if relevant
Template: Product prompt (general)
"You are an experienced e-commerce copywriter and SEO specialist for PrestaShop stores. Using the factual data below, create the following outputs in JSON format exactly as specified. Do not invent any technical specs; if a spec is not provided, set its value to null or 'TBD'. Keep tone: [friendly/professional/technical/brand voice]. Language: [en-US]. Target audience: [example]. SEO target keywords (priority order): [keyword1, keyword2]. Word counts: short_description = 50–90 words (HTML allowed,
OK), description = 160–300 words (HTML allowed). meta_title <= 60 chars, meta_description 120–155 chars. Provide image_alt_text for up to 5 images. Provide up to 8 bullet features. Output fields exactly: name, short_description, description, features (array), meta_title, meta_description, seo_keywords (comma-separated), image_alt_texts (array), category_suggestion, sku, price, currency, weight, dimensions, stock, variants (array of {name, sku, price_diff, stock}), tags (array), warranty, shipping_notes. Here is the product data: [paste product factual data]. Example output: { ... } Now produce the JSON."
Example: Product prompt with sample data (you can copy/paste and adapt)
"You are an experienced e-commerce copywriter and SEO specialist for PrestaShop stores. Using the factual data below, create the outputs below in JSON and HTML where specified. Do not invent any specs; if a spec is missing, set it to null.
Factual data:
- name: EverTrail Women’s Waterproof Hiking Jacket
- material: 100% recycled nylon with DWR finish
- waterproof rating: 10,000 mm
- breathability: 10,000 g/m²/24h
- weight: 350 g (size M)
- colors: Slate Blue, Forest Green, Black
- sizes: XS,S,M,L,XL
- sku_prefix: ET-W-JKT-
- price: 129.99
- currency: USD
- stock: 120
- manufacturer: EverTrail
- images: front.jpg, back.jpg, hood.jpg
Requirements:
- Tone: friendly, practical
- short_description: 55–75 words, allowed HTML , , - , no images
- description: 180–260 words, include 2 usage scenarios and a quick-care instruction
- meta_title: <= 60 chars, include top keyword 'women’s waterproof hiking jacket'
- meta_description: 140–155 chars
- provide features: 6 bullets
- image_alt_texts for each image (3)
- variants: include color/size combinations example for slate blue size M (sku ET-W-JKT-SB-M)
- Provide category suggestion: 'Women > Clothing > Jackets'"
Then provide an example output (short, valid JSON). (Omit long real JSON here — but when using, include one.)
Product prompt example output (illustrative)
{name: "EverTrail Women’s Waterproof Hiking Jacket", short_description: "Lightweight, fully waterproof...", description: "
Designed for ...
", features: ["10,000 mm waterproofing", "Fully taped seams", ...], meta_title: "Women’s Waterproof Hiking Jacket – EverTrail", meta_description: "Lightweight, breathable women’s waterproof hiking jacket—10k/10k membrane. Packable, eco-friendly.", seo_keywords: "women's waterproof hiking jacket, breathable hiking jacket, packable rain jacket", image_alt_texts: ["EverTrail jacket front view", ...], category_suggestion: "Women > Clothing > Jackets", sku: "ET-W-JKT-000", price: 129.99, currency: "USD", weight: "350 g", dimensions: null, stock: 120, variants: [{name:"Slate Blue / M", sku:"ET-W-JKT-SB-M", price_diff:0, stock:12}], tags:["waterproof","breathable","packable"], warranty: "2 years", shipping_notes: "Ships in 1–2 business days."}
Category description prompt
"You are an SEO-savvy e-commerce copywriter. Create a category description for the PrestaShop category 'Women > Footwear > Hiking Boots'. Provide two lengths: short (40–80 words) for category landing thumbnails and long (150–240 words) to appear at the bottom of the category page. Tone: authoritative but approachable. Include the target keywords: 'women's hiking boots', 'waterproof hiking boots for women'. Also give meta_title (<=60 chars) and meta_description (120–155 chars). Output as JSON: {short_description, long_description, meta_title, meta_description, seo_keywords}."
CMS page prompt (policies, about, shipping)
Example: Shipping policy
"You are an e-commerce policy writer. Write a PrestaShop CMS page for 'Shipping & Delivery' that covers shipping zones, delivery times, handling times, carriers, tracking, free-shipping thresholds, returns link, and a short FAQ. Tone: clear and reassuring. Keep it scannable: use headings and short paragraphs or . Include an optional meta_title and meta_description. Output in HTML suitable for PrestaShop CMS content."
Bulk generation: CSV/JSON for import
If you need to generate many products at once, ask for a CSV-ready output. Provide the exact CSV column headers your PrestaShop version expects. Example prompt fragment:
"Output a CSV with these headers exactly: 'name;reference;price;tax;quantity;short_description;description;meta_title;meta_description;categories;tags;image_urls;features'. Use semicolons as separators. Each product row must escape semicolons. Provide up to 10 sample rows for the products listed below. Do not invent specs."
Multilingual output and translations
- Provide the source language and target languages.
- Ask for content plus a separate localized meta_title/meta_description per language.
- Request that translations preserve tone and keyword intent, and adapt keywords to local search habits.
Example: "Also provide a Spanish (es-ES) translation of short_description and meta fields."
Tone, SEO & formatting tips
- Specify primary and secondary keywords, but don’t over-stuff. Ask the model to use primary keyword near the start of meta_title and early in description.
- Specify meta_title length (<60 chars) and meta_description length (120–155 chars).
- For HTML descriptions: allow only safe tags (,
,- ,,,
,) and state whether to include schema/FAQ markup.
- Ask for bullets/features first as plain text and then as an HTML .
- Give examples of brand words/phrases to include or avoid.
Prevent hallucinations and bad outputs
- Ground the model with factual data (specs and images).
- For specs the model doesn’t know, instruct to output "TBD" or null rather than inventing.
- Ask for a final short "data validation" list: e.g., "Check that price is numeric, sizes match provided sizes, image filenames match supplied list."
Iterative refinement prompts
- If first output is too long/short: "Shorten descriptions by 30% while keeping core features."
- If tone is off: "Rewrite descriptions with more technical tone and remove colloquialisms."
- To create A/B variants: "Generate three alternative meta_descriptions focusing on (1) features, (2) price/value, (3) eco-friendly materials."
Quick sample prompts you can copy/paste
1) Single product, JSON
"You are an e-commerce copywriter and SEO expert. Using this product data [paste], create a JSON object with these fields exactly: name, short_description (HTML, 50–80 words), description (HTML, 160–260 words), features (array of 6), meta_title (<=60 chars), meta_description (120–155 chars), seo_keywords (comma-separated), image_alt_texts (array). Do not invent specs; if missing, use 'TBD'. Tone: friendly, professional. Target keyword: [keyword]."
2) Category content
"Write short (50–80 words) and long (150–220 words) descriptions for the PrestaShop category '[category name]'. Include primary keyword '[keyword]'. Provide meta_title and meta_description and 5 SEO tag suggestions. Output in JSON."
3) CMS page: About
"Write an About Us CMS page for our PrestaShop store. We are [brand description]. Include brand mission, sustainability commitments, manufacturing overview, trust badges, and a short FAQ. Tone: authentic and concise. Output in HTML and include a meta title/meta description."
4) Bulk CSV output
"Produce a CSV using semicolons with these headers [list headers]. For these 5 raw product rows [paste data], output cleaned and SEO-optimized columns (short_description in HTML, description plain text, meta fields). Do not invent facts. If a field is missing, leave blank."
Common mistakes to avoid
- Vague prompts: don’t just say “write a product description.” Provide specs and constraints.
- Not specifying output format: AI will return prose not import-ready data.
- Allowing hallucinations: explicitly forbid inventing specs or ask the model to mark guessed values.
- Too broad SEO instructions: give specific target keywords and max lengths.
- Forgetting localization: specify language/locale and units (cm/inches, kg/lbs).
Testing and verification
- Run samples through your PrestaShop import on a staging store.
- Check character limits for meta fields in PrestaShop theme (different themes may trim metadata).
- Validate that HTML tags are allowed and safely rendered.
- Spot-check generated specs vs. original manufacturer sheets.
If you want, I can:
- Convert one of your product rows into an exact prompt ready to use.
- Generate 3 product description variants for a sample product you give me.
- Produce a CSV-ready sample using your specific PrestaShop import headers.
Which would you like me to do next? If you want, paste one product’s factual data and I’ll produce a perfect prompt and a sample output for PrestaShop.
