Skip to main content
Learn efficient strategies for reviewing AI-generated content, from single products to thousands, with clear workflows and quality assurance processes.

​
Overview

The review workflow ensures quality control over AI-generated content before it goes live in your store. 
Job Completes → Open Job Details → Review Products → Accept/Edit/Reject → Apply Changes → Verify in Store

​
Getting Started

​
Opening the Review Interface

1

Navigate to Dashboard

Find your completed job in the activity feed
2

Open Job Details

Click the job card to see all enriched products
3

View Product Table

Review success/failure status for each product
4

Access Review Drawer

Click any product row to open the review panel

​
Understanding the Review Interface

The review drawer displays: Left Panel (Original):
  • Current product data from your store
  • Existing content that will be replaced
Right Panel (Generated):
  • AI-created content
  • AI reasoning explaining decisions
Available Actions:
  • ✅ Accept: Use AI content as-is
  • ✏️ Edit: Modify before applying
  • ❌ Reject: Keep original content

​
Reading AI Reasoning

The reasoning section explains the AI’s decision-making process: Example:
“Color identified as ‘Navy Blue’ based on product title ‘Men’s Navy Cotton Shirt’ and dominant color in product image. Material ‘Cotton’ extracted from title.”
What to evaluate:
  • ✅ Logical reasoning from available data
  • ✅ Correct data sources referenced
  • ❌ Incorrect assumptions or interpretations
  • ❌ Missing critical information
AI reasoning often reveals data quality issues in your catalog. Fix these issues to improve future enrichment quality.

​
Keyboard Shortcuts

Speed up your review process with these shortcuts:
ActionShortcutDescription
Next Product→ or TabMove to next product
Previous Product← or Shift+TabMove to previous product
Accept AllCtrl/Cmd + AAccept all attributes for current product
Reject AllCtrl/Cmd + RReject all attributes for current product
Edit FieldE or EnterEdit selected attribute
Save EditEnterSave current edit
Cancel EditEscCancel current edit
Apply SelectedCtrl/Cmd + SApply selected products to store
Select/DeselectSpaceToggle product selection
Filter MenuFOpen filter options

​
What to Check During Review

​
Content Quality Checklist

  • Product features correctly described
  • Specifications match reality
  • No hallucinated or invented details
  • Consistent with product images
  • All key features mentioned
  • Sufficient detail for customer decisions
  • Important specifications included
  • Use cases or benefits highlighted
  • Matches your established tone
  • Appropriate formality level
  • Consistent across product catalog
  • Target audience considerations
  • Correct spelling and grammar
  • Natural language flow
  • Proper formatting
  • Appropriate reading level

​
SEO Optimization

  • Under 55 characters
  • Includes primary keyword
  • Product name prominent
  • Compelling for search results
  • Under 160 characters
  • Clear value proposition
  • Includes call-to-action
  • Natural keyword integration

​
Structured Attributes

  • Correct value selected
  • Matches product reality
  • Consistent with images/descriptions
  • All applicable values included
  • No incorrect values
  • Complete but not over-tagged

​
Review Strategies by Scale

​
Choosing Your Strategy

Decision Framework for Review Approach:
  • 1-20 Products: Individual Review → Review every product thoroughly
  • 20-100 Products: Sample-Based Review → Review first 10, then sample if quality > 90%
  • 100+ Products: Quality-First Sampling → Review first 10-15, then sample 10-20% if quality > 90%
The key decision point is your initial quality assessment:
  • If quality > 90%: Proceed with sampling strategy
  • If quality 70-90%: Expand review sample
  • If quality < 70%: Stop and refine prompts first

​
Quality Assessment Criteria

Use these thresholds to guide your review strategy:
  • Excellent (>90%): Proceed with sampling strategy
  • Good (70-90%): Review more products before deciding
  • Poor (<70%): Stop and refine prompts before continuing

​
Small Jobs (1-20 Products)

Strategy: Individual review of every product Time Estimate: 2-3 minutes per product
1

Start Review

Open first product in job table
2

Evaluate Content

Check each attribute against quality criteria
3

Make Decisions

Accept, edit, or reject each attribute
4

Navigate

Use keyboard shortcuts to move between products
5

Complete

Review all products before applying

​
Medium Jobs (20-100 Products)

Strategy: Quality sampling with spot checks Time Estimate: 30-45 minutes for 50 products
1

Initial Quality Check

Thoroughly review first 10 products
2

Assess Quality

Use quality criteria to determine sampling rate
3

Sample Review

Check every 5th product if quality is good
4

Priority Review

Always fully review high-value/best-selling products
5

Batch Apply

Apply products that pass quality checks

​
Large Jobs (100+ Products)

Strategy: Efficient sampling with focused review Time Estimate: 1-2 hours for 500 products
1

Quality Baseline

Thoroughly review first 10-15 products
2

Determine Approach

Based on quality assessment, choose sampling rate
3

Strategic Sampling

Review 10-20% of products across different categories
4

Handle Failures

Review all failed products individually
5

Progressive Application

Apply in batches as you complete reviews

​
Editing and Batch Operations

​
Editing Content

When AI content needs minor adjustments:
1

Initiate Edit

Click edit icon or press ‘E’ on the attribute
2

Modify Content

Make changes directly in the field
3

Save Changes

Press Enter or click checkmark
4

Accept Edited Version

Mark the edited content for application
Common edits:
  • Brand name standardization
  • Product-specific details
  • Tone adjustments
  • Factual corrections
  • Formatting consistency

​
Batch Operations

Efficiently handle multiple products: Selection Methods:
  • Click checkboxes for individual selection
  • Shift-click for range selection
  • Use filters then “Select All Filtered”
Bulk Actions:
  • Accept Selected: Apply AI content to selected products
  • Reject Selected: Keep original content for selected products
  • Export Selected: Download for external review
Filtering Options:
  • By Status: Success, Failed, Reviewed, Pending
  • By Collection: Focus on specific categories
  • By Quality Score: If available
  • By Search: Find specific products or SKUs

​
Applying Changes

​
Standard Application Process

1

Select Products

Choose products with approved changes
2

Initiate Apply

Click ‘Apply Selected’ or use Ctrl/Cmd + S
3

Confirm Changes

Review summary and confirm application
4

Monitor Sync

  • Shopify: 1-2 minute sync time
  • Feed Export: Download and import manually
  • API: Real-time updates
5

Verify Results

Spot-check products in your live store
Apply changes in batches—you don’t need to complete the entire job before applying. This allows for progressive updates and easier rollback if needed.

​
Rollback and Undo Procedures

If you need to revert applied changes:
Always test changes on a small batch before applying to your entire catalog.
1

Immediate Undo

Within 5 minutes of application, use the ‘Undo Last Apply’ button in job details
2

Selective Rollback

  1. Navigate to job history
  2. Find the specific application
  3. Select products to revert
  4. Click ‘Rollback Selected’
3

Full Reversion

For complete rollback:
  1. Access Settings → Backup & Recovery
  2. Select pre-enrichment backup
  3. Choose ‘Restore’ for affected products
4

Manual Recovery

If automated rollback unavailable:
  1. Export original data from job details
  2. Use bulk import to restore

​
Team Collaboration

​
Distributed Review

Divide large jobs among team members:
  1. Assign by Category: Each reviewer handles specific collections
  2. Use Filters: Team members filter to their assigned products
  3. Track Progress: Use job dashboard to monitor completion
  4. Coordinate Application: Designated person applies all changes

​
Quality Assurance Workflow

For critical products requiring multiple reviews:
  1. Primary Review: Initial team member reviews and marks products
  2. QA Review: Senior reviewer spot-checks 10-20%
  3. Final Approval: QA lead approves for application
  4. Documentation: Note any patterns for prompt improvement

​
Troubleshooting

​
Common Issues and Solutions

  • Verify sync completed (check sync status)
  • Clear store cache
  • Check for conflicting rules or apps
  • Ensure proper permissions
  • Refresh the page
  • Clear browser cache
  • Check for browser extensions blocking content
  • Try incognito/private mode
  • Verify field is editable in your plan
  • Check field permissions in settings
  • Ensure no active locks on products
  • Reduce batch size (try 50 products)
  • Check for API rate limits
  • Verify all selected products have approved changes
  • Review error logs for specific issues
  • Ensure job completed successfully
  • Check if reasoning was enabled for the job
  • Verify browser isn’t blocking content

​
Best Practices

Focus on accuracy and brand consistency rather than perfection. The 80/20 rule applies—80% quality in 20% of the time is often the right balance.
If you see repeated issues, stop reviewing and adjust your prompts or source data. Fix the root cause rather than editing hundreds of products.
If quality checks consistently pass, trust the process. Over-reviewing wastes time without proportional quality gains.
Take breaks every 30-60 minutes when reviewing large batches. Fresh eyes catch more issues than tired ones.
Keep notes on common problems to improve future prompts and training data.

​
Common Pitfalls to Avoid

Over-Perfecting: Spending excessive time on minor edits for low-traffic products reduces ROI.Blind Acceptance: Never apply content without any review—even excellent AI needs human oversight.Ignoring Patterns: Repeated issues indicate systemic problems. Address the cause, not symptoms.Inconsistent Standards: Maintain the same quality criteria across all reviewers and sessions.

​
Next Steps

Bulk Operations

Advanced techniques for large-scale content management

Customizing Prompts

Improve AI quality to reduce review time

Managing Taxonomies

Better organization for improved AI context

Jobs & Monitoring

Track progress and analyze job history
Need Help? Contact [email protected] or check our Knowledge Base