Search Optimization
Help customers find answers quickly. Relay uses semantic search to match customer queries with relevant articles.
How Search Works
Semantic Search
Unlike keyword matching, semantic search understands meaning:
| Customer Types | Finds Articles About |
|---|---|
| "get my money back" | Refunds, returns |
| "can't log in" | Authentication, passwords |
| "how to start" | Getting started, setup |
Vector Embeddings
Articles are converted to mathematical vectors. Customer queries are compared against these vectors to find the best matches.
Optimizing for Search
Article Titles
Write titles that match how customers ask:
Good:
- "How to Request a Refund"
- "Troubleshooting Battery Issues"
- "Setting Up Your Account"
Less effective:
- "Refund Policy Section 3.2"
- "Battery Information"
- "Account"
Article Content
Include variations of how customers might ask:
- Primary question
- Alternative phrasings
- Related questions
Keywords Naturally
Don't keyword stuff. Write naturally and comprehensively.
Search Analytics
What to Track
- Most searched terms
- Searches with no results
- Click-through rates
- Search to resolution
Finding Gaps
Look for:
- Frequent searches without good matches
- Low-ranking important articles
- High search, low click
Improving Results
When Articles Aren't Found
- Check if topic is covered
- Add article if missing
- Update existing article with search terms
When Wrong Articles Rank
- Review article content
- Make sure title is clear
- Add relevant content to correct article