Using Scribe Home Finder

This guide covers the practical experience of using Scribe from both visitor and agent perspectives.

For Visitors

Getting Started

When you first arrive on a HomeStar site, Scribe waits quietly. After you’ve browsed a few properties (typically 3-5), the floating button appears in the bottom-right corner with a badge showing how many properties you’ve viewed.

Tip

The button only appears after you’ve generated enough browsing data for meaningful insights—usually after 10 minutes and 3+ properties viewed.

Viewing Your Preferences

Click the Scribe button to open a sheet showing:

• Detected preferences with confidence indicators
• A freeform text field for additional requirements
• Options to confirm, dismiss temporarily, or opt out permanently

Example preferences you might see:

• Price range: $400,000 - $600,000 (85% confidence)
• Bedrooms: 3+ bedrooms (75% confidence)
• Location: Twin Falls area (90% confidence)
• Features: Pool, granite countertops (60% confidence)

Adding Your Own Preferences

You don’t have to wait for Scribe to detect preferences. You can manually add them:

1. Navigate to the My Data page (/my-data)
2. Click “Add preference” in the preferences section
3. Select preference type (price, bedrooms, location, features)
4. Enter the value
5. The preference is saved with 100% confidence

User-added preferences:

• Are always shown, even before Scribe becomes “ready”
• Cannot be overwritten by inferred preferences
• Can be deleted at any time by you

Note

Manual preferences are perfect for requirements that might not show up in your browsing behavior, like “must be near Lincoln Elementary” or “need home office space.”

Confirming & Rejecting Preferences

Scribe gets things wrong sometimes. You can refine inferred preferences:

Action	Result
Confirm	Sets confidence to 100%, permanently included
Reject	Sets confidence to 0%, excluded from all matching
Ignore	Keeps original confidence, may change over time

Confirmed and rejected preferences persist across sessions and won’t be re-inferred.

Adding Freeform Notes

The “Anything else you’re looking for?” field allows you to express requirements that don’t fit structured preferences:

“Need to be near Lincoln Elementary, home office space required, prefer mountain views from the back deck”

These notes are:

• Stored with your session
• Visible to matched agents
• Transferred to CRM if you convert to a lead

Use this for context that can’t be captured in dropdowns—school districts, accessibility needs, specific lifestyle requirements.

Choosing an Agent

When you confirm your preferences, Scribe presents agents who specialize in what you’re looking for. Each suggested agent card shows:

• Agent photo and name
• Match label (Excellent/Great/Good/Available)
• “Best Match” badge on the top-ranked agent
• Up to 2 personalized reasons why they’re a good fit

Example match reasons:

• “Specializes in Twin Falls family homes”
• “Expert in $400k-$600k price range”
• “Knows the best neighborhoods with pools”

Click on an agent to select them. Once assigned, Scribe becomes personalized to that agent and you can continue refining preferences together.

Privacy Controls

View My Data

See exactly what data has been collected via /my-data

Temporary Dismiss

“Not now” snoozes Scribe for 24 hours

Permanent Opt-Out

“Don’t show again” permanently disables Scribe

Complete Deletion

Delete all session data from the /my-data page

For Agents

Making Your Profile Work

Scribe matches visitors to agents based on semantic similarity between visitor preferences and your agent profile specialties. The better you describe what you specialize in, the better the matches.

Generic profile (poor matching):

“I sell homes in Idaho.”

Specific profile (excellent matching):

“I specialize in helping first-time buyers find affordable homes in Twin Falls and Jerome. I know the school districts, the neighborhoods with the best yards, and how to negotiate in competitive markets.”

Tip

Update your specialties in Admin → Agents → [Your Profile] → Lead Matching Preferences

Understanding Match Scores

When Scribe assigns you a lead, you’ll see a match score indicating how well the visitor’s preferences align with your specialties:

Score	Label	What It Means
≥80%	Excellent Match	This is your wheelhouse—preferences strongly align with your specialties
≥60%	Great Match	Good fit for your expertise, should be able to help well
≥40%	Good Match	Relevant experience, might need to collaborate with specialists
Under 40%	Available	You’re available but this isn’t your primary specialty

High match scores mean higher conversion rates because you already know the market they’re interested in.

Viewing Lead Preferences

When a visitor converts to a lead, their session data syncs to your CRM:

• Budget range — from properties they viewed and filters they applied
• Location interests — cities and neighborhoods they focused on
• Property requirements — bedrooms, bathrooms, property type
• Feature interests — pools, garages, views, etc. from photo engagement
• Freeform notes — anything they typed in the notes field

This means your first conversation starts with context: “I see you’re interested in 3-bedroom homes in Twin Falls around $500k with pools. Let me show you some great options…”

Troubleshooting: Why Am I Not Getting Matches?

If you’re not being suggested to visitors, check:

1. Is your agent profile active? — Must be marked as “active” in the system
2. Have you set lead matching preferences? — Need specialties configured
3. Are your embeddings generated? — System must have processed your profile
4. Is your specialty too narrow? — Might need to broaden your description slightly

Note

Contact your brokerage admin if embeddings aren’t generated. They can trigger a re-indexing.

Common Scenarios

Visitor Browsing Multiple Times

Scribe recognizes return visits and boosts confidence on those properties:

• First visit: Normal preference weighting
• Return visit: Confidence boosted to 90%
• Multiple returns: Weight multiplied by visit count

When a visitor views the same property 3 times, Scribe knows they’re seriously interested in that style, price, and location.

Visitor Preferences Change

If a visitor starts browsing different property types or price ranges, Scribe adjusts. Recent browsing behavior has more weight than older patterns.

Preferences aren’t locked in—they evolve as the visitor explores.

No Good Matches Found

If semantic matching returns fewer than 3 agents:

• Your brokerage’s primary/default agent is added with 50% score
• System falls back to first eligible agents
• Visitor still sees options, but match quality may be lower

This is rare if agents have good profile descriptions.

Visitor Opts Out

If a visitor clicks “Don’t show again”:

• Scribe is permanently disabled for that session
• A flag is stored in localStorage (scribe_disabled)
• The visitor can still use all other site features
• Their session data remains available via /my-data but Scribe won’t appear

Respect their choice. Some people prefer to browse independently.

Best Practices

For Brokerages

1. Train agents on profile quality — Better descriptions = better matches
2. Monitor match scores — Track which agents get high-quality leads
3. Keep embeddings updated — Re-index when agents update their profiles
4. Respect privacy requests — Honor opt-outs immediately

For Agents

1. Update your specialties regularly — Market focus changes over time
2. Be specific about neighborhoods — “Twin Falls suburbs near schools” beats “Twin Falls”
3. Mention price ranges you excel in — Helps match with budget-conscious buyers
4. Follow up quickly — Scribe-matched leads are warm, don’t let them cool

For Visitors

1. Browse naturally — Don’t try to game the system, just look at homes you’re interested in
2. Add manual preferences early — If you have firm requirements, add them upfront
3. Use freeform notes — Context helps agents help you better
4. Choose your agent thoughtfully — Match scores are guidance, but trust your gut

Related Features

• Understanding Scribe — Learn how Scribe works
• Scribe Detection Reference — Technical thresholds and algorithms
• Scribe API — API endpoints for developers
• Preference Curation — Manual preference management