AI Chatbots Are Reading Your Zillow Score — Not Your Closed Volume

What Just Got Confirmed

A column published this week in HousingWire by real estate coaches Tim and Julie Harris made something explicit that most agents have only vaguely sensed: when a consumer asks an AI assistant — ChatGPT, Perplexity, Google's AI Overview, take your pick — to recommend a top listing agent in their neighborhood, the AI is not pulling MLS production records. It is not auditing closings. It is reading Zillow, Homes.com, and Realtor.com, and then presenting whatever those portals surface as if it were an objective, research-backed answer.

That's the news. The portals rank agents by platform engagement. AI inherits those rankings. The consumer gets an authoritative-sounding recommendation shaped by who paid into — and optimized for — the platform. The agent at the top of the AI's answer may genuinely be excellent. Or they may just be very good at Zillow.

This matters to your business in two concrete ways: one you can't control (how the AI references you today), and one you absolutely can (how easy it is for the AI to find and verify your actual production record).

How Portal Rankings Actually Work — and Why It's Not a Scam, Just Misunderstood

The source column is careful here, and it's worth being equally careful. The major portals are not fabricating data. Zillow's Find an Agent directory, per its published methodology, surfaces agents based on star ratings, review volume, and self-reported sales tied to MLS IDs. Non-paying agents are included.

The problem is that every signal in that formula rewards platform behavior, not market performance. High review count? That rewards agents who aggressively solicit Zillow-specific reviews. Profile completeness? Rewards agents who treat Zillow as a marketing priority. Self-reported sales? Rewards agents who diligently log transactions back into Zillow's system — something a referral-heavy producer with a full pipeline has little incentive to do.

The result is a directory that correlates closely with portal investment and only loosely with independently verifiable production. A top-producing agent who built a referral book and never touched their Zillow profile can rank below a lower-volume agent who works the platform obsessively.

That same ranking becomes the raw material for an AI recommendation — now stripped of the visual cues (ads, sponsored badges, platform branding) that might prompt a savvy consumer to ask questions.

Where Each Major Platform Actually Stands

| Platform | Ranking Basis | Cost to Participate | How AI Sees It | |---|---|---|---| | Zillow Find an Agent | Reviews + profile completeness + platform engagement | Free (organic); Premier Agent pricing is ZIP-based and not publicly listed — typically $300–$1,000+/mo in most markets, rising to $5,000+/mo in competitive metros | Very high — heavily indexed, frequently cited | | Realtor.com Find a Realtor | Reviews + self-reported transactions | Connections Plus typically $200–$1,800+/mo by market (custom quote required) | High — NAR affiliation means broad crawl coverage | | Homes.com | Tiered by spend; paid members sort above free | Varies by market | Growing — increasingly indexed as the portal scales | | HomeLight | MLS production data (invitation-only; closed volume, DOM, list-to-sale ratio) | 33% referral fee per close (30% in select states) | Low direct AI risk, but absent if you opt out | | Redfin Partner Program | Close-rate vetting | 30–35% of buyer-side commission | Lower — Redfin not a primary AI source for agent searches | | UpNest | Commission-discount willingness | Commission reduction | Minimal AI visibility |

The clearest takeaway from this table: the platforms that rank closest to merit (HomeLight, Redfin partners) both gate participation behind a referral fee of roughly a third of your commission. High-performing agents who decline to pay that toll are simply absent from those results — and absent from the AI answers built on them.

There is no neutral, performance-only directory at consumer scale. That gap is what the AI is now filling with portal data.

What Fixing Your AI Footprint Actually Costs

The source column recommends building owned assets. That's correct. Here's what it realistically looks like to execute:

Free moves that matter most:

• Google Business Profile — Free. This is the highest-leverage item on the list. A fully built GBP with consistent NAP (name, address, phone), real production stats in plain text, and a steady review stream is exactly the kind of machine-readable, credible, cross-verified signal that AI tools weight. If you haven't claimed and optimized yours, do it before anything else.
• Diversified reviews — Free. Every review that only lives on Zillow is a leased asset. Google reviews, Facebook reviews, and independent mentions on local news or neighborhood sites build the corroborating web that makes AI more confident attributing production to you specifically.

Low-cost, high-impact:

• Owned domain website — Domain: ~$15–20/year. Hosting: $15–50/month on a basic plan, $50–150/month for a managed WordPress or similar setup. Website builders like Squarespace and Wix offer templates that simplify setup. The site should state your actual production numbers in plain English — neighborhoods served, homes sold, price ranges — not just "trusted advisor" marketing copy. Structured data markup (Schema.org RealEstateAgent type) helps parsers read it cleanly; a developer can add this for a modest one-time fee, or some website builders include it in templates.
• Neighborhood-level content — If you write a 600-word page about median days-on-market in the specific ZIP you dominate, and you actually have the production to back it up, that page can outrank a portal profile for hyper-local AI queries. This is content marketing, but it's also just making your real expertise legible to machines.

What not to waste money on: Buying more portal placement to fix an AI visibility problem is like paying rent on a building to fix the fact that you don't own it. Zillow's algorithm can change; your Google presence cannot be taken from you.

Who Should Act on This Now

You should care if:

• You depend on cold leads — online search, AI referrals, or portal inquiries — for a meaningful share of your pipeline
• You're newer and building a brand from scratch in a competitive market
• You're in a market where two or three agents dominate the AI results and you're not one of them

You can mostly ignore this if:

• Your pipeline is 80%+ referral-based and full. Clients who already know you don't ask ChatGPT for recommendations.
• You operate in a small enough market that your name is already dominant across local web content
• You actively participate in HomeLight or a similar merit-vetted program and are satisfied with that flow

Skip this if...

Chasing portal rankings: Skip investing in Zillow Premier Agent or Connections Plus as a strategy to improve your AI visibility. The spend may buy leads directly, but it doesn't solve the credibility problem — it deepens your dependence on infrastructure you don't control.

The Practical Takeaway

Run this test today: ask ChatGPT and Perplexity who the top listing agent is in your primary ZIP code. Note which agents are named and which sources are cited. If you're not in the results, look at who is and audit what they have that you don't — most likely a more complete GBP, an owned website with specific production claims, or a broader review footprint across multiple platforms.

The AI is not going to start pulling directly from MLS transaction data anytime soon. What it can do is find, read, and corroborate whatever you make publicly available. Building that record is a one-time investment. Renting a portal ranking is a recurring cost on someone else's infrastructure.

The portals shaped the rankings. The AI amplified them. The agents who will win the next phase are the ones who make their own production record the easiest thing for any engine — human or machine — to find and trust.