Documentation Menu↓
Evaluating AI Tag Recommendations
Checking confidence ratings and reviewing suggested tag fields.
Overview
CrawlSage generates specific code recommendations for every detected SEO issue. Each recommendation includes a confidence score (from 0% to 100%) and a risk profile (Low, Medium, High). This helps you quickly assess whether a recommendation is safe to deploy immediately or requires manual modification first.
AI Recommendations Detail Drawer
The AI recommendation drawer displaying current meta titles side-by-side with the AI-suggested version, alongside the confidence score rating badge (e.g. 96%).
Prerequisites
- A completed Quick Scan or Full Audit.
- At least one technical SEO issue detected (e.g., missing meta description or missing schema).
Step-by-Step Guide
- Open the **Audit Report** from your dashboard.
- Click on an issue category (e.g. **Meta Tags** or **Technical SEO**).
- Click **Review Recommendation** on any issue card. The AI review drawer will open.
- Check the **Confidence Score**:
- High (85%+): The recommendation is highly safe and matches standard template rules.
- Medium/Low (<85%): The recommendation requires user review. Use the edit trigger to tweak fields.
- Examine the code comparison view to confirm the proposed tags are correct.
Expected Result
You will have a clear understanding of the SEO benefit, the confidence score, and the exact difference between the current HTML header code and the proposed replacement tag.
Confidence Score Calculation
Common Mistakes
- Deploying low-confidence fixes without review: Low-confidence fixes may suggest generic keywords if the crawler was unable to parse main text tags due to a script loader. Always review them.
Troubleshooting
- Drawer shows "Generating..." indefinitely:
The backend may be experiencing OpenAI API delays. Close the drawer, wait 15 seconds, and reopen the recommendation to re-trigger the sync fetch.