Best Practices & Optimization Guide

Example 1: Luxury Fashion Brand
"You are a luxury fashion copywriter creating aspirational content for
style-conscious professionals aged 30-50. Use elegant, sophisticated
language that emphasizes craftsmanship, exclusivity, and timeless design.
Avoid price mentions or urgency tactics. Focus on lifestyle benefits
and premium materials."

Example 2: Tech Gadgets Brand
"You are a tech journalist writing for early adopters and gadget
enthusiasts. Use precise technical specifications, performance benchmarks,
and real-world use cases. Balance technical accuracy with accessibility.
Include model numbers, battery specs, and connectivity standards.
Enthusiastic but never hyperbolic."

Example 3: Sustainable Home Goods
"You are an eco-conscious lifestyle expert writing for environmentally
aware consumers. Emphasize sustainable materials, ethical production,
and environmental benefits. Use warm, approachable language that
inspires positive change. Include certifications and eco-metrics."

Before (Generic):
"Write a product description for {title}. Include features and benefits."

After (Optimized):
You are an outdoor gear specialist writing for experienced hikers and
backpackers who value performance and durability. Focus on technical
specifications and real-world field performance.

Product: {title}
Category: {category}
Weight: {weight}
Material: {material}
Price: {price}

Requirements:
- Lead with the most critical performance spec (weight, waterproof rating, etc.)
- Include 2-3 technical specifications with units
- Mention specific use case (alpine, trail, ultralight)
- Reference field testing or certifications
- Professional but passionate tone about the outdoors
- Length: 350-450 characters

Format:
- First sentence: Key performance benefit
- Middle: Technical specifications
- End: Use case and target activity

Must Include:
- Weight in grams or ounces
- Material composition
- Intended activity level

Avoid:
- Generic outdoor clichés ("embrace adventure")
- Vague quality claims ("high-quality materials")
- Superlatives without evidence ("best", "ultimate")

Setup Custom Variables (Advanced Bulk Generator → Settings):
1. Map `pa_brand` → {brand}
2. Map `pa_model` → {model}
3. Map `pa_warranty` → {warranty}
4. Map `_custom_certifications` → {certifications}

Use in Prompt:
"Highlight the {brand} {model} with emphasis on {warranty} coverage
and {certifications} compliance."

Output:
"Highlight the Sony WH-1000XM5 with emphasis on 2-year warranty
coverage and CE and FCC certifications compliance."

Highlight the key features and specifications of {title}:

Technical Specifications:
- Primary Spec: {attribute1}
- Performance: {attribute2}
- Capacity/Size: {attribute3}

Write a compelling description that:
1. Leads with the standout technical feature
2. Explains how each specification benefits the user
3. Includes precise measurements and units
4. Maintains professional, technical tone

Length: {max_length} characters
Audience: Technical decision-makers and engineers

Create a persuasive description for {title} that focuses on customer
benefits and lifestyle improvements rather than just features.

Product Context:
- What it is: {category}
- Price point: {price}
- Key features: {attributes}

Structure:
1. Opening: Main benefit or problem solved
2. Supporting benefits: 2-3 ways it improves customer's life
3. Social proof hint: Who else uses/loves it
4. Closing: Call to action or value reinforcement

Tone: Warm, customer-focused, enthusiastic
Avoid: Technical jargon, feature lists, specifications
Length: {max_length} characters

Write a detailed technical description for {title} targeting
professional buyers and procurement teams.

Required Specifications:
- Model/Part Number: {model}
- Dimensions: {dimensions}
- Capacity: {capacity}
- Compliance: {certifications}
- Warranty: {warranty}

Format Requirements:
- Lead with model number and category
- List specifications with units
- Include compliance/certifications
- Mention warranty and support
- Professional, formal tone
- No marketing language or adjectives

Length: {max_length} characters
Audience: Procurement managers, technical buyers

Craft an authentic story-driven description for {title} that
emphasizes craftsmanship, materials, and the maker's passion.

Product Details:
- What it is: {title}
- Materials: {materials}
- Made by/in: {maker_location}
- Price: {price}

Narrative Elements:
1. Origin story: How/why it's made
2. Materials: Emphasize quality and sourcing
3. Craftsmanship: Detail the making process
4. Uniqueness: What makes each piece special
5. Connection: Why customers will love it

Tone: Warm, authentic, passionate
Voice: First-person ("we craft") or artisan perspective
Avoid: Mass-production language, generic quality claims
Length: {max_length} characters

Weekly Generation Schedule Example:

Monday:
- Review weekend sales performance
- Identify products needing description updates
- Plan week's generation queue

Wednesday:
- Small batch test (10 products) with new prompts
- Review and refine based on results

Saturday Morning (8 AM):
- Large batch generation (100-200 products)
- Monitor progress, check for errors
- Spot-check quality on sample products

Sunday Afternoon:
- Review generated content
- Approve/reject in Advanced mode
- Apply approved descriptions
- Document any issues for Monday refinement

Example Hypothesis:
"Adding technical specifications to product descriptions will increase
conversion rate for electronics products by 20%."

Variables to Test:
- Control (A): Current generic descriptions
- Variant (B): Descriptions with specifications

Success Metric: Conversion rate (product page → add to cart)
Minimum Sample: 20 products per variant
Test Duration: 2 weeks

Control (A):
- Conversions: 51 / 895 views = 5.7%

Variant (B):
- Conversions: 70 / 963 views = 7.3%

Improvement: +28%
P-value: 0.03 (statistically significant)

Winner: Variant B (technical specifications)

Quality Scoring System (1-10 scale):

Content Accuracy (0-3 points):
- All facts correct: 3 points
- Minor errors: 2 points
- Major errors: 0-1 points

Brand Voice (0-3 points):
- Perfect brand alignment: 3 points
- Mostly on-brand: 2 points
- Off-brand: 0-1 points

SEO Optimization (0-2 points):
- Keywords natural, appropriate length: 2 points
- Keyword density acceptable: 1 point
- No optimization: 0 points

Customer Value (0-2 points):
- Compelling, action-driving: 2 points
- Informative but bland: 1 point
- Unhelpful: 0 points

Total Score: 0-10
Target: 7+ for all products, 9+ for premium

Efficient Batching Strategy:

Scenario: 200 products to generate

Method A (Individual):
- 200 separate API calls
- Each call: 500 tokens (prompt) + 200 tokens (output)
- Total: 140,000 tokens
- Cost: $2.10 (GPT-4o-mini)

Method B (Batched with shared prompt):
- Use Standard Bulk Generator (shared prompt template)
- Prompt loaded once, applied to all
- Total: 45,000 tokens
- Cost: $0.68 (GPT-4o-mini)

Savings: $1.42 (68% reduction)

Scenario: 500 products need descriptions

Option A (Advanced Mode):
- 500 products × 1 credit = 500 credits
- Free tier: Not possible (only 250 credits)
- Pro tier: Uses 40% of monthly allocation
- Cost: Included in Pro plan ($X/month)

Option B (Standard Mode):
- 500 products × 0 credits = 0 credits
- Free tier: Fully supported
- Pro tier: Saves credits for other uses
- Cost: $0 (unlimited Standard mode)

Recommendation: Use Standard Mode for bulk, save Advanced credits for premium products

Strategic Credit Allocation:

High Priority (Advanced Mode):
- Bestsellers (top 20%): 100 products × 2 = 200 credits
- New releases (monthly): 10 products × 2 = 20 credits
- Premium tier (>$200): 15 products × 2 = 30 credits
Total: 250 credits/month (fits Free tier)

Medium Priority (Standard Mode):
- Regular catalog: 300 products × 0 = 0 credits

Low Priority (Standard Mode):
- Clearance/bulk: 200 products × 0 = 0 credits

Result: Optimal quality where it matters, unlimited volume overall

Check Credit Balance:
1. Navigate to AI Product Tools → Settings
2. View "Credit Balance" in header
3. Track daily usage via Ctrl Credit System dashboard

Set Alerts:
- 75% used (188 credits): Start conserving
- 90% used (225 credits): Critical - only emergencies
- 100% used (250 credits): Wait for monthly reset

Track Patterns:
Month 1: Used 230/250 (92%) - too close
Month 2: Used 195/250 (78%) - better buffer
Month 3: Used 180/250 (72%) - optimal pace

Settings → Content Settings → Short Description Prompt:

"Write a compelling 150-160 character summary for {title} that:
- Starts with primary keyword naturally
- Highlights 1-2 standout benefits
- Includes a subtle call to action
- Encourages search result clicks

Format: [Keyword opening] + [Key benefit] + [Value proposition]
Avoid: Generic phrases, excessive punctuation, ALL CAPS

This will appear in search results - make it count!"

Ensure Descriptions Support Product Schema:

Required Elements for Rich Results:
- Product name (from title)
- Price (from WooCommerce data)
- Description (AI-generated)
- Availability (from stock status)
- Image (from product media)

AI Description Considerations:
✅ Plain text or simple HTML (p, br, strong, em)
✅ Natural language, complete sentences
✅ Avoid special characters that break JSON-LD
✅ Include factual claims (supports review schema)

❌ Excessive HTML formatting
❌ JavaScript or dynamic content
❌ Encoded characters or emoji overuse

Validation:
1. Generate descriptions
2. Test with Google Rich Results Test tool
3. Verify product schema validates
4. Check for errors in Search Console

Regular Maintenance Tasks:

Weekly:
- Optimize product tables: wp_posts, wp_postmeta
- Clean up auto-drafts and revisions
- Index custom fields used in filtering

Monthly:
- Optimize AI Product Tools tables:
  - wp_aipt_bulk_generator_history
  - wp_aipt_automation_jobs
  - wp_aipt_job_logs
- Archive old generation history (>90 days)
- Vacuum/optimize database

Commands (via WP-CLI or phpMyAdmin):
OPTIMIZE TABLE wp_posts;
OPTIMIZE TABLE wp_postmeta;
DELETE FROM wp_posts WHERE post_type = 'revision' AND post_modified < DATE_SUB(NOW(), INTERVAL 90 DAY);

Plugin Updates:
- Enable auto-updates for minor versions (security patches)
- Review changelog for major updates before upgrading
- Test updates in staging environment first (if available)
- Backup database before major version updates

AI Provider Model Updates:
- Subscribe to provider newsletters:
  - OpenAI: platform.openai.com/docs → Changelog
  - Google AI: ai.google.dev/changelog
  - OpenRouter: openrouter.ai/docs/changelog
- Test new models in small batches before switching
- Document model performance changes

Community & Support:
- Join AI Product Tools user community (forums, Slack, Discord)
- Follow plugin updates and feature releases
- Share successful prompt templates with community
- Report bugs and feature requests to support

Phase 1: Master Primary Language (Week 1)
1. Perfect English prompts first
2. Achieve 8/10 quality score
3. Document successful templates
4. Validate with A/B testing

Phase 2: Adapt for Target Languages (Week 2-3)
1. Translate prompt template (not literal, adapt culturally)
2. Adjust for formality level
3. Test with 10 sample products per language
4. Refine based on native speaker feedback

Phase 3: Test Cultural Fit (Week 4)
1. Get native speaker to review outputs
2. Check for cultural appropriateness
3. Verify terminology accuracy
4. Ensure legal/regulatory compliance

Phase 4: Monitor and Improve (Ongoing)
1. Track conversion rates by language
2. A/B test local vs. translated prompts
3. Gather customer feedback
4. Iterate based on performance data

Example Timeline:
- English (primary): 2 weeks to optimize
- Spanish: 1 week to adapt
- German: 1 week to adapt
- French: 1 week to adapt
- Total: 5 weeks for 4-language support

Scenario: Product Launch Event

Pre-Launch (2 weeks before):
"Highlight that {title} is coming soon. Build anticipation. Mention
pre-order availability and exclusive launch benefits. Exciting tone."

Launch Day:
"Announce {title} is now available! Emphasize limited-time launch
pricing and early-bird bonuses. Urgency without pressure."

Post-Launch (2 weeks after):
"Regular product description with focus on customer reviews and
early adopter testimonials. Proven quality emphasis."

Automation Strategy:
- Schedule automation jobs to update descriptions at each phase
- Use WP-Cron to trigger regeneration on specific dates
- Track conversion rates by phase to optimize future launches

Automation Job Examples:

Job 1: Weekly New Product Descriptions
- Schedule: Every Monday at 2 AM
- Filter: Products added in last 7 days
- Tool: Standard Bulk Generator
- Action: Generate + Auto-approve
- Purpose: Ensure all new products have descriptions within 24 hours

Job 2: Monthly Premium Product Refresh
- Schedule: First Sunday of month at 3 AM
- Filter: Category = "Premium", Price > $200
- Tool: Advanced Bulk Generator
- Action: Generate + Draft review
- Purpose: Keep high-value product content fresh

Job 3: Seasonal Campaign Updates
- Schedule: Manually triggered before campaigns
- Filter: Tagged with campaign tag (e.g., "summer-sale-2025")
- Tool: Advanced Bulk Generator with seasonal prompt
- Action: Generate + Draft review
- Purpose: Update descriptions for marketing campaigns

Configuration:
Navigate to AI Product Tools → Automation Jobs → Create New Job

Scenario: Low-Stock Alert Regeneration

Use Case:
When inventory drops to <10 units, create urgency in description

Implementation:
1. Use WooCommerce inventory hook
2. Trigger automation job for low-stock products
3. Use custom prompt emphasizing scarcity

Prompt Addition:
"Note: Limited stock available ({stock_quantity} remaining). Mention
this creates urgency without being pushy."

Result:
"Premium wireless headphones with 30hr battery and ANC. Only 8 units
remaining at this price. Secure yours today before stock sells out."

Strategy: Regenerate Underperforming Products Quarterly

Process:
1. Export Google Analytics data (last 90 days)
2. Identify products with high traffic but low conversion (<2%)
3. Create filter in automation job by product IDs
4. Regenerate with optimized prompt
5. Monitor conversion rate changes

Expected Results:
- 30% of regenerated products see conversion improvement
- Average conversion increase: 0.5-1.5 percentage points
- ROI: Positive within 2-4 weeks

Model Comparison Test:

1. Select 3 identical products
2. Generate with different models:
   - Product A: GPT-4o-mini
   - Product B: Gemini 2.5 Flash
   - Product C: GPT-4o (if budget allows)

3. Compare quality:

   Example Results:

   GPT-4o-mini:
   "Wireless headphones with 30hr battery and ANC. Comfortable design
   for all-day use. Bluetooth connectivity included."
   Quality: 6/10 (generic, safe)

   Gemini 2.5 Flash:
   "Experience premium audio with advanced active noise cancellation
   and exceptional 30-hour battery life. Bluetooth 5.2 ensures stable
   wireless connectivity for seamless listening."
   Quality: 8/10 (specific, engaging)

   GPT-4o:
   "Immerse yourself in studio-quality sound with these premium wireless
   headphones. Advanced ANC technology blocks ambient noise while
   40mm neodymium drivers deliver rich, detailed audio. The impressive
   30-hour battery ensures uninterrupted listening throughout your week."
   Quality: 9/10 (compelling, detailed)

4. Choose best-performing model for your products

Hypothesis: Prompt is too complex for AI to follow consistently

Test:

1. Current Complex Prompt (200 words):
   [Long detailed prompt with 15 requirements]

   Result: AI confused, inconsistent output

2. Simplified Prompt (50 words):
   "You are a professional product writer. Write a compelling
   description for {title} in the {category} category. Include
   key features from {attributes}. Professional tone. Length:
   {max_length} characters."

   Result: Consistent, good quality output

3. Incrementally Add Complexity:
   - Start with simple prompt (test 5 products)
   - Add 1 requirement at a time
   - Test each addition with same 5 products
   - Stop when quality degrades

   Finding: AI handles 5-7 clear requirements well, degrades after 10+

Provider Status Pages:

OpenAI: https://status.openai.com
Google AI: https://status.cloud.google.com
OpenRouter: https://status.openrouter.ai

Actions:
1. Visit status page for your provider
2. Check for recent outages or degraded performance
3. Review incident reports for your region/model
4. If issues present, wait for resolution or switch providers temporarily

Example:
"OpenAI GPT-4o-mini experiencing 15% increased latency in US-East region"
→ Switch to Gemini 2.5 Flash temporarily
→ Test if quality improves
→ Return to GPT-4o-mini after incident resolved

Create Quality Assurance Workflow:

1. Generate with Primary Model:
   - Attempt generation with preferred model (e.g., GPT-4o-mini)
   - Check output quality score

2. If Quality < 7/10:
   - Automatically regenerate with secondary model (e.g., Gemini Flash)
   - Compare outputs

3. Select Best Output:
   - Use higher-quality version
   - Document which model performed better

4. Adjust Model Selection:
   - If secondary model consistently wins, make it primary
   - Update model settings

Implementation:
Currently manual process - future plugin update may automate this

Consistency Audit Process:

1. Export Product Data:
   - Navigate to Products → All Products
   - Export CSV with all fields
   - Focus on category you're generating for

2. Analyze Data Completeness:

   Example Analysis (Headphones Category):

   | Product ID | Title | Attributes | Custom Fields | Completeness |
   |------------|-------|------------|---------------|--------------|
   | 101 | Sony WH-1000XM5 | Brand, Color, Battery, Connectivity | ✅ All 4 | 100% |
   | 102 | Wireless Headphones | None | ❌ 0/4 | 25% |
   | 103 | Bose QC45 | Brand, Color | ✅ 2/4 | 75% |
   | 104 | Premium Headset | Color | ✅ 1/4 | 50% |

   Finding: Product 101 has complete data = excellent output
            Product 102 has minimal data = poor output

   Correlation: Data completeness directly impacts quality

3. Identify Data Gaps:
   - Which attributes are missing most often?
   - Are custom fields populated consistently?
   - Do high-quality outputs correlate with complete data?

4. Fill Data Gaps:
   - Bulk edit products to add missing attributes
   - Populate brand, model, specifications
   - Ensure all products in category have same attribute fields

Data Standardization Checklist:

Before:
Product A: Brand = "sony"
Product B: Brand = "SONY"
Product C: Brand = "Sony Corporation"
Result: AI treats these as different brands, inconsistent output

After:
Product A: Brand = "Sony"
Product B: Brand = "Sony"
Product C: Brand = "Sony"
Result: Consistent brand recognition, uniform quality

Standardization Rules:

1. Brand Names:
   - Title case: "Sony" not "SONY" or "sony"
   - Short form: "Sony" not "Sony Corporation"
   - Consistent spelling: "Canon" not "Cannon"

2. Attributes:
   - Use units consistently: "30 hours" not "30hr" or "30 hrs"
   - Standard format: "Bluetooth 5.2" not "BT 5.2" or "Bluetooth v5.2"
   - Numeric values: "30" not "thirty"

3. Categories:
   - Use full path: "Electronics → Audio → Headphones"
   - Not abbreviated: "Elec > Audio > Headph"

4. Custom Fields:
   - Same fields across all products in category
   - Empty fields: Use "N/A" not blank (AI knows it's intentional)

Implementation:
- Use bulk edit in WooCommerce
- Or CSV export → clean data → reimport
- Verify consistency before regeneration

Problem: Prompt assumes all products have {brand}, but some don't
Result: AI confused when {brand} is empty, produces poor output

Solution: Conditional prompt logic

Before (Assumes Brand Exists):
"Highlight the {brand} {title} with its premium features..."

Problem: If {brand} is empty, output is:
"Highlight the  Wireless Headphones with its premium features..."
(Double space, awkward)

After (Conditional Approach):
"Write a professional description for {title} in the {category}
category. Include key features from {attributes}. If brand information
is available, mention it naturally. If technical specifications are
provided, highlight them. Adapt content based on available product data."

Result: AI handles missing data gracefully
- With brand: "The Sony Wireless Headphones offer..."
- Without brand: "These wireless headphones offer..."

Advanced (Using Variable Checks):
Currently not supported by plugin, but workaround:
- Populate missing fields with "N/A" or generic value
- Train AI to handle "Brand: N/A" by omitting brand mention

Understanding Temperature Settings:

Temperature controls AI creativity/randomness:
- 0.0 = Deterministic (same input → same output)
- 1.0 = Creative (same input → varied outputs)
- Default: Usually 0.7-0.8

Current Issue:
High temperature = inconsistent outputs for similar products

Solution:
1. Check plugin settings for temperature control
   - Currently hardcoded in plugin (not user-adjustable)
   - Future update may expose this setting

2. Workaround: Request consistent outputs in prompt
   "Write a consistent, professional description following this format:
   [Opening sentence with primary feature]
   [2-3 supporting features]
   [Use case or target customer]
   [Closing value proposition]

   Maintain consistent structure and quality across all products."

3. Test Consistency:
   - Generate same product 3 times
   - Compare outputs
   - If wildly different, temperature too high
   - If identical, good consistency

Strategy: Different prompts for different data completeness levels

Tier 1: Complete Data (100% attributes + custom fields)
Use: Advanced prompt with all variables

Tier 2: Partial Data (50-99% complete)
Use: Standard prompt with essential variables only

Tier 3: Minimal Data (< 50% complete)
Use: Simple prompt, focus on title and category

Implementation:

1. Tag Products by Tier:
   - Audit data completeness
   - Tag products: "data-complete", "data-partial", "data-minimal"

2. Create Tier-Specific Prompts:

   Tier 1 Prompt:
   "Write detailed description for {title} by {brand}. Include:
   - Battery life: {battery}
   - Connectivity: {connectivity}
   - Material: {material}
   Highlight {certification} compliance. Target {target_audience}."

   Tier 3 Prompt:
   "Write professional description for {title} in {category}.
   Focus on common benefits for this product type at {price} price point.
   Professional tone, 300 characters."

3. Generate by Tier:
   - Filter by tag
   - Use appropriate prompt
   - Consistent quality within tier

Result:
- Tier 1: 9/10 quality (complete data = detailed outputs)
- Tier 2: 7/10 quality (partial data = good outputs)
- Tier 3: 6/10 quality (minimal data = basic but consistent)
- Overall: Predictable, consistent quality based on data availability

Automated Consistency Check:

After generation, score each output:

Quality Criteria:
1. Length compliance (290-410 chars for 400 target): ✅ or ❌
2. Contains product name: ✅ or ❌
3. Includes price mention: ✅ or ❌
4. Has specific features (not generic): ✅ or ❌
5. Brand voice keywords present: ✅ or ❌

Score Calculation:
5 ✅ = 10/10 quality
4 ✅ = 8/10 quality
3 ✅ = 6/10 quality
2 ✅ = 4/10 quality
1 ✅ = 2/10 quality
0 ✅ = 0/10 quality (regenerate immediately)

Automated Actions:
- If score < 6/10: Flag for manual review
- If score < 4/10: Auto-regenerate with different model
- If score = 10/10: Auto-approve

Implementation:
- Currently manual review
- Use Advanced Bulk Generator's review queue
- Approve high scores, reject/regenerate low scores

Performance Baseline Test:

1. Single Product Generation:
   - Generate 1 product via metabox
   - Time from click to result
   - Record: _____ seconds
   - Expected: 5-8 seconds
   - Status: ✅ Normal | ⚠️ Slow (8-15s) | 🚨 Critical (>15s)

2. Small Batch (10 products):
   - Standard Bulk Generator, 10 products
   - Total time: _____ seconds
   - Expected: 50-80 seconds (5-8s each)
   - Status: ✅ Normal | ⚠️ Slow | 🚨 Critical

3. Medium Batch (50 products):
   - Standard Bulk Generator, 50 products
   - Total time: _____ minutes
   - Expected: 4-7 minutes
   - Status: ✅ Normal | ⚠️ Slow | 🚨 Critical

Document Results:
| Test | Expected | Actual | Status |
|------|----------|--------|--------|
| Single | 5-8s | 18s | 🚨 Critical |
| 10 products | 50-80s | 180s | 🚨 Critical |
| 50 products | 4-7min | 25min | 🚨 Critical |

Analysis: 3x slower than expected across all tests
Next: Identify bottleneck

Diagnostic Process:

1. Test API Provider Speed Directly:

   cURL Test:
   ```bash
   time curl -X POST https://api.openai.com/v1/chat/completions \
     -H "Content-Type: application/json" \
     -H "Authorization: Bearer YOUR_API_KEY" \
     -d '{
       "model": "gpt-4o-mini",
       "messages": [{"role": "user", "content": "Write a product description for wireless headphones"}]
     }'

**Step 3: Optimize Database Performance (30 minutes)**

**Step 4: Optimize Server Configuration (1 hour)**

**Step 5: Reduce Batch Sizes (10 minutes)**

**Step 6: Identify and Disable Plugin Conflicts (30 minutes)**

**Step 7: Optimize Prompt Length (15 minutes)**

Before (Verbose):
"Please write a comprehensive and detailed product description for
the product titled {title} which belongs to the {category} category.
The description should include all relevant features and benefits.
Make sure the description is engaging and compelling for potential
customers who might be interested in purchasing this product. The
tone should be professional but also friendly and approachable.
Please ensure that you incorporate the price of {price} naturally
into the description. The description should be approximately
{max_length} characters in length."

Word count: 95 words
Token count: ~120 tokens
API time: 5 seconds

After (Concise):
"Write a professional product description for {title} ({category})
priced at {price}. Include key features and customer benefits.
Professional yet friendly tone. Length: {max_length} characters."

Word count: 30 words
Token count: ~40 tokens
API time: 3 seconds

Improvement: 40% faster, same quality

Model Speed Comparison:

Speed Test (Same Prompt, Same Product):

| Model | Response Time | Quality | Cost |
|-------|---------------|---------|------|
| **GPT-4o** | 8 seconds | 9/10 | High |
| **GPT-4o-mini** | 4 seconds | 8/10 | Low |
| **Gemini 2.5 Flash** | 2 seconds | 8/10 | Free |
| **Gemini Flash-8B** | 1.5 seconds | 7/10 | Free |
| **OpenRouter qwen3** | 2 seconds | 7/10 | Free |

Recommendation:
- For bulk operations: Gemini 2.5 Flash (fast + free)
- For quality: GPT-4o-mini (balanced)
- For premium: GPT-4o (slow but best)

Implementation:
1. Navigate to Settings → API Settings
2. Change model to Gemini 2.5 Flash
3. Re-test performance
4. Expected: 2-3x faster generation

Speed Improvement Example:
- 100 products with GPT-4o-mini: 400 seconds (6.7 min)
- 100 products with Gemini Flash: 200 seconds (3.3 min)
- **2x faster with free model**

Performance Optimization Summary:

Before:
- Single product: 18 seconds
- 10 products: 180 seconds
- 50 products: 25 minutes
- Status: 🚨 Critical performance issues

After All Optimizations:
- Database indexed: -40% time
- Server config improved: -30% time
- Smaller batches: Stable, no timeouts
- Plugin conflicts resolved: -20% time
- Shorter prompts: -20% time
- Faster model: -50% time

Final Performance:
- Single product: 3 seconds (6x faster ✅)
- 10 products: 35 seconds (5x faster ✅)
- 50 products: 3 minutes (8x faster ✅)
- Status: ✅ Excellent performance

Key Takeaway:
Combined optimizations deliver 5-8x performance improvement
Most impact: Database optimization + faster AI model