Before you price your San Diego County home, let AI interview it like a buyer
Updated September 2026
For San Diego, Orange and Riverside County homes that aren’t listed yet, Zillow’s median Zestimate error is approx 6.20%, per Zillow figures cited in the August 18, 2026 Reports on Housing. Half of those estimates missed the eventual sale price by more than that. Ray Stendall of Stendall Realty Group wrote the free AI prompt on this page so sellers can test their value in 2026 the way a buyer would.
A 6.20% miss on a $1 million estimate is $62,000. Once a home hits the market, Zillow recalculates using the asking price and listing description, and the same report puts that on-market median error at 1.64%. So the estimate most owners look at while they’re still deciding is the least reliable one they’ll see.
Why might buyers know your home’s value before you do?
Buyers walk into showings with their homework done. They pull recent sales on Zillow and Redfin, compare price per square foot, check how long the house down the street sat, and more of them paste it all into an AI assistant before they ever call an agent.
You see the house through everything you’ve put into it. Say your kitchen remodel cost $85,000. A buyer sees that kitchen next to every other home in their budget, and they pay for it only as far as those alternatives let them. The gap between what an owner put in and what the market pays back is where a lot of pricing mistakes start.
What does a buyer-side AI analysis do differently?
It works in the order a careful appraiser or buyer would, and it saves the automated estimates for last. The prompt has the AI value your home from closed sales first, then pending sales, then the listings competing for the same buyer. It reads expired and withdrawn listings as signs of prices buyers turned down, and it gives them less weight than closed sales. Only after it forms its own opinion does it look at the Zillow and Redfin estimates, and it has to explain any big gap instead of averaging them together. It also checks whether the market has shifted since each comparable went under contract, and it won’t treat a sale as credit-free when nobody knows whether the seller paid concessions.
Then it switches sides. The AI takes its own number and asks what else a buyer could purchase for roughly the same money, within about 10% up or down. Your home has to win two contests at once. The beauty contest is about appeal against those alternatives. The price war, which the prompt calls the value contest, is about whether your price feels like the better deal.
Everything points at one figure, the highest defensible price an informed buyer is reasonably likely to pay. That’s a different number from the highest price someone will agree to list at. Sellers who mix up the two tend to learn the difference one price cut at a time.
Why target a contract within 30 days when Expected Market Time is 107 days?
Because the price that pulls an offer quickly is the price buyers read as a clear value. Expected Market Time is the number of days it would take to sell every San Diego County listing at the current buying pace, which makes it a gauge of supply against demand, and individual homes can sell much faster or much slower. It reached 107 days in the August 18, 2026 Reports on Housing, up from 99 days a year earlier. Detached homes came in at 101 days. For homes priced above $2 million, the figure was 144.
Demand is thin. The same report counted 1,682 new pending sales over the prior month, the lowest mid-August reading since tracking began in 2012. When buyers are this selective, a home priced like an average home competes with every average home, and the one priced as the obvious choice gets the showings first.
The prompt holds the window at 30 days on purpose, because that’s the question most sellers would rather skip. Any price that needs more time lands in the stretch zone, along with what has to go right to get it.
How to run the home value prompt
Use any AI assistant that can search and open live web pages, such as ChatGPT, Claude, Gemini, or Perplexity with web search turned on. The prompt’s first instruction makes the AI confirm live access and stop if it has none, which keeps you from getting comparable sales invented out of stale training data.
Copy everything between the two lines below. Replace [ENTER FULL PROPERTY ADDRESS] with your address, then paste it in. The AI will ask up to 10 questions about condition, updates, and permits, so answer them honestly, including the garage conversion nobody pulled a permit for. Zillow and Redfin sometimes block AI tools, and when that happens the prompt tells the AI to say so and keep working with other sources.
PROMPT STARTS BELOW THIS LINE
HOME VALUE ANALYSIS: WHAT WOULD AN INFORMED BUYER PAY? Property address: [ENTER FULL PROPERTY ADDRESS] Target timeframe: accepted contract within 30 days ROLE AND GOAL Act as a residential real estate valuation analyst using disciplined comparable-sale analysis combined with buyer-side competitive analysis. I own the property above. Estimate the HIGHEST DEFENSIBLE PRICE a well-informed buyer is reasonably likely to pay if the property is properly exposed to the market and goes under contract within 30 days. Assume normal MLS exposure, professional-quality presentation, good photos, and reasonable showing access unless I tell you otherwise. IMPORTANT: I am NOT asking for the highest price I could list the property for. I am asking for the highest price current market evidence shows a well-informed buyer is reasonably likely to PAY within 30 days. I want market reality, even if the answer is lower than I hope. Do not favor my position because I am the owner. Give me substantially the same valuation you would give if you were advising the buyer. Use specific evidence rather than generic statements. Never invent property facts, sales, prices, concessions, dates, links, or market data. GROUND RULES 1. LIVE RESEARCH CHECK First, tell me in one sentence whether you can search and open current web pages in this session. If you cannot, do not create a valuation from remembered or training-data sales. Tell me live market research is unavailable and offer to analyze current property and comparable data that I provide. 2. TODAY'S DATE State the date you are using at the top of your report, taken from your system or your live search results. If you cannot confirm today's date, ask me before you start. Measure every recency period from that date. 3. SOURCES Use Zillow and Redfin as primary public sources for property, listing, and sale information whenever accessible. Cross-check important information using Realtor.com, county assessor or property records, brokerage listing pages, or other credible current sources. If Zillow, Redfin, or another source blocks access, say so and continue with the sources you can reach. Never imply you reviewed a source you could not access. 4. LINKS Provide direct Zillow and Redfin links for the subject and important comparables only when you opened those pages and confirmed they match the correct address. Never manufacture a Zillow or Redfin URL from an address pattern. 5. EVIDENCE INTEGRITY For material facts that affect the valuation, mark which kind of fact it is: VERIFIED: confirmed from an accessible current source REPORTED: found in a source but not independently confirmed OWNER-PROVIDED: supplied by me ESTIMATE: your analytical estimate UNKNOWN: could not be established Use labels where they separate known information from assumptions or estimates. Do not label every sentence. 6. FAIR HOUSING Base all location analysis on property and place characteristics. Never consider or mention the race, color, religion, national origin, sex, disability, familial status, or any other protected characteristic of the people who live in an area. STEP 1: VERIFY THE SUBJECT PROPERTY Verify as accurately as possible: - Address - Property type - Bedrooms and bathrooms - Living area - Lot size and usability - Year built - Garage and parking - HOA and dues, if applicable - View - Neighborhood, subdivision, or tract - Micro-location - School district, when relevant - Previous listing and sale history - Any other characteristic likely to materially affect buyer demand Pay attention to location factors a simple distance radius misses, including road noise and traffic, gated versus non-gated location, view orientation, street position, lot usability, road access, power lines, proximity to commercial uses, and school boundaries. If credible sources disagree about a material fact, show the conflict and name the sources. Do not choose whichever version supports the highest value. STEP 2: ASK ME ABOUT WHAT THE INTERNET CANNOT SEE Ask me no more than 10 concise questions about what you cannot reliably determine online. Combine topics as needed, and cover, where relevant: - Overall condition - Kitchen and bathroom renovations - Flooring - Roof, HVAC, and windows - Solar, and whether it is owned or leased - Pool, landscaping, and lot improvements - Views - Additions, ADUs, and garage conversions - Permitted versus unpermitted improvements - Deferred maintenance - Any unusual positive or negative feature Then wait for my answers before completing the valuation. If this platform cannot pause for answers, state the assumptions you must make and lower your confidence accordingly. Treat my answers as information to test against market evidence, not automatic proof of added value. Do not assume the cost of an improvement equals the value buyers will assign to it. STEP 3: BUILD VALUE FROM MARKET EVIDENCE Do not use any Zestimate, Redfin Estimate, Realtor.com estimate, or other automated valuation model (AVM) yet. Zillow and Redfin pages display their estimates next to the listing data you will be reading. Set those figures aside until Step 8. First, independently analyze: A. CLOSED SALES B. PENDING / UNDER-CONTRACT PROPERTIES C. ACTIVE COMPETING LISTINGS TIME Start with the most recent 90 days. Expand to 3 to 6 months when necessary, and to 6 to 12 months only when needed to find meaningful comparable evidence. Give greater weight to recent transactions when properties are otherwise similarly comparable. DISTANCE AND LOCATION Start in the same neighborhood, subdivision, tract, or immediate buyer market. Expand geographically only when necessary. Do not assume a closer property is more comparable if it belongs to a materially different buyer market. Tell me how far back in time and how far out geographically you had to search. COMPARABILITY Favor properties similar in: - Neighborhood or subdivision - Micro-location - Property type - Living area - Lot size and usability - Bedrooms and bathrooms - Age - Condition and renovation level - Garage and parking - View - Architecture and amenities, when relevant - Layout and functional utility - Any other characteristic that materially influences buyer behavior Aim for 3 to 6 strong closed comparables. Do not force weak properties into the analysis to reach a minimum number. If fewer than 3 genuinely useful closed comparables exist, say so and lower your confidence. FAILED MARKET EVIDENCE When reliable listing history is available, consider expired, canceled, withdrawn, or substantially reduced listings as secondary evidence of prices the market may have rejected. Never treat an unsuccessful asking price as evidence of value. Give this evidence less weight than closed sales, but use it when it helps explain buyer resistance or a later successful sale. Never select an older, farther, or less comparable property because its price supports a higher valuation. Never exclude a credible recent comparable because it contradicts your conclusion. STEP 4: ANALYZE THE COMPARABLES Present closed sales separately from pending and active properties. For important closed comparables, show whenever available: - Address - Distance - Beds and baths - Living area - Lot size - Original list price - Final list price - Sale price - Sale date - Days on market - Price per square foot - Condition and renovation level - Seller concessions (amount, or UNKNOWN) - Important similarities - Most important differences - Why this comparable matters - Weight in the analysis: HIGH, MEDIUM, or LOW - Zillow link - Redfin link - Evidence status, where useful Use separate tables for pending and active competition, with the same fields minus sale price, sale date, and concessions. ADJUSTMENTS Explain adjustments in plain English. Do not create artificial dollar adjustments or false precision. Put a specific dollar amount on an adjustment only when market evidence reasonably supports it. SELLER CONCESSIONS When verified seller concessions or credits exist, identify them and consider their economic effect on the sale. Do not treat ordinary buyer-agent compensation as a seller concession unless the available data establishes that treatment. When concessions are unknown, write UNKNOWN. Never assume zero. STEP 5: INTERPRET PENDING SALES Pending properties matter because they show what buyers are choosing right now. The contract price is generally unknown until the sale closes, so never assume it equals the list price. Evaluate what can reasonably be learned from: - Original list price - Current or final asking price - Price reductions - Days on market before going pending - Condition - Competition at the time - How quickly the property attracted a buyer Clearly separate known facts from estimates. STEP 6: ACCOUNT FOR MARKET DIRECTION Determine whether market conditions today differ materially from when the key closed comparables went under contract. Consider available evidence such as: - Current inventory - Recent pending activity - Days on market - Price reductions - Sale-to-list price relationships - Buyer competition - Changes in financing conditions, when relevant Do not apply a mathematical market adjustment unless evidence supports one. Explain whether older sales deserve less weight because today's buyer environment looks stronger, weaker, or similar. STEP 7: RUN THE BUYER'S ALTERNATIVE TEST Now switch sides. Assume I am a buyer with a budget centered on your preliminary valuation. Find relevant ACTIVE and PENDING homes about 10 percent above and below that price that a buyer considering my property might also tour. Ask: "If I were a rational buyer with this budget, why would I choose this house instead of the alternatives?" Compare: - Condition - Size - Lot - Location and micro-location - Views - Renovations - Architecture - Garage and parking - Amenities - Age - HOA - Schools, when relevant - Layout - Overall desirability - Price Then answer: BEAUTY CONTEST: How compelling is my property compared with what buyers can purchase instead? VALUE CONTEST: At what price does my property become a clearly compelling value relative to its alternatives? Determine the highest price at which my property remains among the most compelling choices a reasonable buyer would seriously consider, ideally one of the top two when enough comparable alternatives exist. If competing properties are materially more compelling, reconsider the valuation downward. If my property is materially superior, determine whether actual market evidence supports a premium rather than assuming one. Buyers determine value through alternatives. STEP 8: ONLY NOW CHECK AVMs After completing the independent valuation, report when available: - Zillow Zestimate - Redfin Estimate - Realtor.com estimate Show the difference between each AVM and your independent valuation. Do not average AVMs together. Do not let an AVM override stronger evidence from closed sales, pending activity, and current competition. If an AVM differs by about 5 percent or more, or otherwise materially disagrees with your conclusion, explain the likely reasons, such as stale public records, condition the algorithm cannot observe, an unusual lot, views, additions, remodeling, micro-location, or fast-changing market conditions. STEP 9: GIVE ME FOUR PRICING LEVELS 1. MARKET REALITY The range most strongly supported by closed sales, pending activity, current competition, market direction, and buyer behavior for a contract within 30 days. This is also the most likely sale range. Give your best single-number estimate within it. 2. OBVIOUS CHOICE The price or narrow range where a serious buyer should see my property as among the most compelling choices for the money, ideally one of the top two when enough comparable alternatives exist. This is a positioning price, and it may sit in the lower part of the market reality range or below it. Explain where it falls and why. 3. STRETCH A price above the market reality range that remains plausible but requires stronger-than-average buyer enthusiasm, exceptional presentation, favorable timing, limited competing inventory, or more than 30 days of exposure. Explain what must go right. 4. ASPIRATIONAL A price above the stretch level that current evidence does not adequately support, carrying materially greater risk of extended market time, price reductions, or failure to sell. Explain the evidence that makes it aspirational. If the evidence makes a clean separation between levels misleading, explain the overlap instead of forcing an artificial boundary. STEP 10: ANSWER THE MAIN QUESTION Give me: HIGHEST DEFENSIBLE SALE PRICE (CONTRACT WITHIN 30 DAYS): $__________ This sits at or near the top of the market reality range. It may exceed that range only when you cite specific closed or pending evidence that supports it, and it must stay below the stretch level. MOST LIKELY SALE RANGE (MARKET REALITY): $__________ to $__________ BEST SINGLE-NUMBER ESTIMATE OF CURRENT MARKET VALUE: $__________ PRICE WHERE MY PROPERTY BECOMES AN OBVIOUS CHOICE: $__________ CONFIDENCE: HIGH, MEDIUM, or LOW, with one sentence explaining why. Then explain the evidence behind these numbers in plain English. STEP 11: QUALITY CONTROL Before presenting the final answer, audit the analysis. COMPARABLE AUDIT - Did I choose the truly best comparables? - Did I overweight price per square foot? - Did I use an older sale when a better recent sale existed? - Did I ignore a credible inconvenient comparable? - Did I account for neighborhood and micro-location? - Did I keep closed, pending, and active properties separate? - Did I treat a failed asking price as evidence of value? BUYER AUDIT - With the same money, what else can the buyer purchase? - Are several alternatives more compelling? - Is the subject materially superior? - Does actual evidence support any premium or discount? BIAS AUDIT - Did the owner's description of improvements raise the valuation without transactional support? - Did an AVM, including one displayed on a listing page, anchor the analysis? - Did I confuse replacement cost with market value? - Did I confuse list price with sale price? - Am I telling the owner what they want to hear? - Would I give substantially the same valuation if advising the buyer? FACT AUDIT - Do property links actually open to the stated addresses? - Are important prices, dates, property facts, and statuses supported by current evidence? - Did I present an estimate as a verified fact? - Did I assume unknown concessions were zero? - Do the four pricing levels and the main answer agree with each other? - Does every location comment rest on property and place characteristics only? Correct any problems before answering. Do not reveal hidden reasoning or internal deliberations. At the end, briefly list 2 to 4 material corrections that quality control produced. If none were necessary, say that no material corrections were required. FINAL REPORT Organize the answer as: 1. Executive Summary (5 sentences maximum) 2. Subject Property and Data Conflicts 3. Best Closed Comparables 4. Relevant Pending Sales 5. Current Buyer Competition 6. Market Direction 7. Beauty Contest and Value Contest 8. Zillow / Redfin / Realtor.com AVM Check 9. Four Pricing Levels 10. Highest Defensible Price and Most Likely Range 11. What Could Change the Valuation 12. Data Gaps and Facts to Verify 13. Quality-Control Changes Use specific evidence rather than generic statements. If the available evidence cannot support a reliable valuation, say so and explain exactly what additional information is needed instead of inventing an answer. Label the result: "Market analysis for discussion purposes. This is not an appraisal."
PROMPT ENDS ABOVE THIS LINE
What should you do with the results?
Check the evidence labels before anything else. The facts that drive the value should carry a VERIFIED, REPORTED, OWNER-PROVIDED, ESTIMATE, or UNKNOWN label, and every link should open to the right address. If a comparable doesn’t hold up, have the AI rerun that step.
The costliest mistake is treating the stretch zone as your list price. Stretch prices need extra enthusiasm, good timing, or patience, and the prompt labels them that way on purpose. Another is crediting improvements at cost. If comparable sales don’t show buyers paying for a new pool, the AI shouldn’t count it either, and it’s built to push back when you ask it to.
Letting a Zestimate settle the argument is a third trap, since the pre-listing median error runs 6.20% in the August 18, 2026 Reports on Housing. And read the confidence rating. A low rating in a thin market like Rancho Santa Fe or Del Mar means the comparable set is weak, and no prompt can fix that. The report is a market analysis for discussion. When a lender, a court, or an estate needs a formal value, that job belongs to a licensed appraiser.
What happens when we run it together?
When we meet, I run this same prompt with you, on screen. Public data can’t tell the AI everything. It won’t know if the roof was replaced two years ago, whether the yard backs onto open canyon or a busy road, or that a neighbor is about to list a similar house.
So I add what I know about the property and the street, and then I ask the AI to pressure-test my pricing recommendation. I don’t want it agreeing with me because I’m the broker in the room. If the evidence backs my number, you’ll see exactly why. If it turns up something that should change the plan, we both want to know before the sign goes in the yard.
Ray Stendall, who has tracked North County San Diego for more than 20 years, built the prompt so a seller sees what an informed buyer sees. For current conditions in one city, the Carlsbad real estate market page breaks down pricing across its four ZIP codes.
Frequently asked questions: AI home value prompt
Can AI tell me what my home is worth?
AI can organize the evidence, but it can’t see inside your home. A well-built prompt makes it value the property from closed sales, pending sales, and competing listings, label the facts that matter, and admit its gaps. Treat the result as a starting analysis to test with Ray Stendall when you meet and to compare with what other agents have shared with you.
How accurate is a Zestimate in San Diego County?
For San Diego County homes that aren’t listed yet, Zillow’s median Zestimate error is 6.20%, per the August 18, 2026 Reports on Housing. A miss that size on a $1 million estimate would be $62,000. After a home is listed and Zillow factors in the asking price, the median error drops to 1.64%.
Which AI tool should I use for this home value prompt?
Use any assistant that can search and open live web pages, such as ChatGPT, Claude, Gemini, or Perplexity with web search turned on. The prompt makes the AI confirm live access first. If the tool can’t browse, it should stop instead of building comparable sales from outdated training data. I suggest the paid ChatGPT version with access to Redfin and Zillow skills, but any of the frontier models will work.
Should I list my home at the highest price the AI gives me?
Not necessarily. The highest defensible sale price, the most likely sale range, and the best list price answer different questions. The prompt estimates what current evidence suggests buyers may pay. Ray Stendall builds list-price strategy by also weighing current competition, presentation, your priorities, and how much market time you’re willing to accept.
Why does the prompt check Zillow and Redfin estimates last?
Automated estimates can anchor the analysis before the real evidence gets weighed. The prompt has the AI build value from actual sales and current competition first, then compare its number with the Zillow and Redfin estimates. Any gap larger than about 5% has to be explained, which often exposes condition or data problems.
Is an AI pricing analysis a substitute for an appraisal?
No. The report is a market analysis for discussion and carries no appraisal standing with a lender, a court, or an estate. Stendall Realty Group uses it to test a pricing recommendation against public evidence. When a sale, a probate matter, or a loan requires a formal value, hire a licensed appraiser.
If you want a specific read on where your home stands in the current San Diego County market, book a private seller strategy call and we’ll run this analysis together. It’s an honest look at your options with no pitch attached. Call or text 858-877-0484, or visit stendallrealtygroup.com. Ray Stendall, Stendall Realty Group, eXp Realty, DRE #02038682.