Below are practical, ready-to-use principles, templates, and example prompts you can paste into ChatGPT (or another LLM) to generate SEO- and e‑commerce-ready content for PrestaShop products, categories and CMS pages. Each template tells the model exactly what data you want, how to format it, and what constraints to respect — which is the key to “perfect” prompts.
Quick rules of thumb
- Be explicit: specify role, goal, output format, tone, keywords, and length limits.
- Provide product data (brand, features, materials, SKU, target audience, use cases). The LLM cannot invent specifics reliably for technical fields.
- Request structured output (JSON/CSV) so you can import or convert it automatically.
- Include SEO constraints: target keywords, meta lengths, and slug (link_rewrite) rules.
- Ask for multiple variations (titles, meta descriptions) to A/B test.
- Use examples and delimiters for clarity, and include “If missing, say so” checks.
Common PrestaShop field length guidance (practical, not rigid)
- meta_title: ~50–60 characters (max ~70)
- meta_description: ~120–160 characters (max ~160–180)
- link_rewrite (slug): short, lowercase, hyphens, no spaces
- short_description: 100–300 characters (displayed on listing)
- description (full): 150–1200+ characters (detailed HTML)
- tags: 2–8 relevant keywords
Adjust lengths for languages where characters/word lengths differ.
1) Product content — Template (single product)
Prompt to paste:
You are an e-commerce copywriter specializing in optimized PrestaShop product data. Produce JSON only, with no extra text. Follow these rules:
- Use the provided product data (between <<< and >>>) and fill fields below.
- Keep meta_title ≤ 70 chars, meta_description 120–160 chars, short_description ≤ 300 chars.
- Generate a search-friendly link_rewrite (slug): lowercase, hyphens, only English letters/numbers/hyphens.
- Provide three title variations (Title_A, Title_B, Title_C) 45–60 chars each.
- Give 3 meta_description variations for SEO.
- Provide short_description (plain text) and description (HTML allowed, with 2–3 short paragraphs and a 3–5 bullet feature list).
- Output tags as an array and features as [{name, value}].
- Include recommended categories and 3 suggested cross-sell phrases.
- Tone: friendly, concise, benefit-driven. Target audience: [TARGET_AUDIENCE].
- Target keywords (priority): [KEYWORDS].
- If any necessary input is missing, note which fields are missing in the JSON field "missing_fields".
Product data:
<<<
Name: [PRODUCT_NAME]
Brand: [BRAND]
SKU: [SKU]
Material: [MATERIAL]
Color/Finish: [COLOR]
Dimensions/Weight: [DIMENSIONS]
Key features: [FEATURES]
Use / Benefits: [USE_CASES]
Price: [PRICE] (optional)
Target audience: [TARGET_AUDIENCE]
Priority keywords: [KEYWORDS comma-separated]
Other keywords: [OTHER_KEYWORDS]
Language: [LANGUAGE_CODE]
>>>
Output JSON schema example:
{
"missing_fields": [],
"name": "",
"title_variations": ["", "", ""],
"meta_descriptions": ["", "", ""],
"meta_title": "",
"meta_description": "",
"link_rewrite": "",
"short_description": "",
"description_html": "",
"tags": ["", ""],
"features": [{"name":"", "value": ""}],
"suggested_categories": ["", ""],
"cross_sell_phrases": ["", "", ""]
}
Example (input filled quickly):
- PRODUCT_NAME: "EcoBreeze Bamboo Towel Set"
- BRAND: "EcoHome"
- KEYWORDS: "bamboo towels, absorbent towels, eco-friendly towels"
- TARGET_AUDIENCE: "eco-conscious homeowners"
(LLM should return JSON with fields populated.)
2) Category page — Template
Goal: SEO-friendly category name + meta + short description + full description + slug + suggestions.
Prompt to paste:
You are a senior SEO copywriter. Produce a JSON object only. Use the category data between <<< and >>>. Requirements:
- Create category name (if not provided), meta_title ≤ 70 chars, 1 meta_description 140–160 chars.
- link_rewrite: short slug, lowercase, hyphens.
- Short description ≤ 200 chars (for listings).
- Full description HTML with 2–4 paragraphs and a 5–7 bullet list of what customers will find in this category.
- Suggest 8–15 tags (keywords), and 6–10 subcategory ideas (names only).
- Tone: trust-building, product-benefit oriented. Target keywords: [KEYWORDS].
- Output JSON with fields: missing_fields, name, meta_title, meta_description, link_rewrite, short_description, description_html, tags, suggested_subcategories.
Category data:
<<<
Category seed name: [SEED_NAME]
Target keywords: [KEYWORDS]
Audience: [AUDIENCE]
Language: [LANGUAGE_CODE]
>>>
3) CMS page — Template (About Us, Shipping, Returns, Guides)
Prompt to paste:
You are a professional content writer for PrestaShop CMS pages. Produce JSON only. Use the CMS details between <<< and >>>. Requirements:
- Create a clear page title (≤ 60 chars), meta_title ≤ 70 chars, meta_description 120–160 chars.
- link_rewrite slug.
- Main content in HTML with headings (H2/H3), short intro paragraph, 3–6 info sections, and a short FAQ (2–4 Q&As).
- Tone: [TONE]. Audience: [AUDIENCE].
- Add 2 suggested internal links to other pages (by slug/title).
- Output fields: missing_fields, page_title, meta_title, meta_description, link_rewrite, content_html, suggested_internal_links.
CMS data:
<<<
Page type: [ABOUT / SHIPPING / RETURNS / GUIDE]
Key points to include: [POINTS]
Legal/regulatory notes: [NOTES_IF_ANY]
Target keywords: [KEYWORDS]
Language: [LANGUAGE_CODE]
>>>
4) Bulk generation / CSV-ready output
If you need dozens or thousands of records, ask the model to produce CSV or JSON Lines with exact column headers matching your import template.
Prompt example:
You are a data export assistant. Given the product list (between <<< and >>>) generate CSV only (first row = header) with these columns exactly:
"id","name","meta_title","meta_description","link_rewrite","short_description","description_html","tags","features_json","suggested_categories"
- For tags use semicolon (;) separated string.
- For features_json use JSON string of [{"name":"", "value":""},...]
- Ensure meta lengths and slug rules are followed.
If any product lacks required data, output "MISSING:[field]" in the corresponding CSV cell.
5) Rewriting or optimizing existing content
Ask for an improvement rating and concrete edits.
Prompt example:
You are an e-commerce UX & SEO editor. I will provide the current content between <<< >>>. Return JSON only with keys:
- "score" (1–10),
- "issues" (array of short issue strings),
- "optimized_meta_title",
- "optimized_meta_description",
- "rewritten_short_description",
- "rewritten_description_html",
- "edits_explained" (3–5 short bullets).
If text is fine, explain why.
6) Multilingual content & localization
- Provide the target language code and local SEO keywords.
- Ask the model to produce both original and localized slugs.
- For languages with accents, ask to provide both accented and ASCII slugs.
Prompt fragment:
Language: fr-FR. Provide both "link_rewrite" (ASCII slug) and "slug_native" (accented, display slug). Translate content and adapt units/currency.
7) Add structured data (JSON-LD) for product pages
Prompt:
Generate a JSON-LD Product schema for the product using schema.org/Product. Include name, image (if provided), description (short), sku, brand, offers (priceCurrency, price, availability), aggregateRating if rating provided, and 3 review snippets. Output the JSON-LD only.
8) A/B variations, tone switching, and length control
- Ask explicitly for Title A/B and meta A/B.
- Ask for three tone variants: “Formal”, “Playful”, “Technical”.
9) Example final prompt (fully filled) and expected output
Prompt:
You are an e-commerce SEO writer. Produce JSON only. Use the product details here:
<<<
Name: "UltraGrip Running Shoes"
Brand: "PeakStride"
SKU: PS-UR-42
Material: Mesh upper, EVA midsole
Color: Black/Neon
Dimensions: Men’s sizes 40–46
Key features: enhanced grip sole, breathable mesh, memory-foam insole, lightweight
Use / Benefits: trail running, long-distance comfort, wet-weather traction
Target audience: trail runners and fitness enthusiasts
Priority keywords: "trail running shoes", "grippy running shoes", "waterproof running shoes"
Other keywords: "lightweight running shoes","men's trail shoes"
Language: en-US
>>>
Requirements:
- Provide name, three title_variations, meta_title, three meta_descriptions, link_rewrite, short_description (≤200 chars), description_html (2 short paragraphs + bullet features), tags, features, suggested_categories, cross_sell_phrases. Tone: confident, active.
Expected JSON keys as in template above.
(Then the LLM returns the JSON.)
10) Operational and safety checks
- Always validate slugs for disallowed characters before import.
- Run character length checks on meta fields.
- Use human review on claims (health/safety).
- For legal or regulated products add a compliance check step or a “do not make medical claims” requirement.
11) Prompt engineering tips (practical)
- Use triple quotes/delimiters to prevent data confusion.
- Ask for output in a single language only (or specify bilingual fields).
- Use "Produce JSON only" to avoid extra prose.
- For bulk, include unique id input to match outputs to inputs.
- Use few-shot examples for nuanced tone or examples of perfect output.
12) Quick checklist to paste before any prompt
- [ ] Role and goal stated
- [ ] Input data provided and delimited
- [ ] Exact output format (JSON/CSV) given
- [ ] Field lengths and constraints stated
- [ ] SEO keywords and target audience defined
- [ ] Tone and s specified
- [ ] Any prohibited claims or words listed
- [ ] Ask for missing_fields reporting
If you want, I can:
- Create ready-to-import CSV templates for your PrestaShop import module.
- Generate 5 sample product JSON outputs from a sample product you paste in.
- Help convert an existing product export into a prompt-ready dataset for bulk enrichment.
Which would you like next (CSV template, sample outputs, or convert your product list)?
