REALTOR AI PLAYBOOK · 28
Better Marketing. Stronger Conversations. Practical AI.
A market report can help you explain a useful question. It cannot automatically tell a specific seller what their home will sell for next month. The challenge is to turn a broad statistic into clear education without turning it into a forecast.
Start by writing four items above your draft: source, reporting period, coverage and limitation. If any item is missing, complete the source check before asking ChatGPT for a caption. This small step prevents an old or broad number from becoming a misleading claim about today's neighborhood.
Here, you will work through one real historical report, a restrained interpretation and a suggested educational post. The example is deliberately dated. It is not a current market update, an appraisal or advice to price a particular property.
The source for this worked example
We use the Orlando Regional REALTOR® Association's September 2025 State of the Market report, checked October 18, 2026. Its scope is residential listings taken or sold by ORRA brokers, wherever located; commercial property and vacant land are excluded. This is not a count of every transaction inside Orlando city limits.
The report records inventory of 13,306 in August 2025 and 13,007 in September 2025. It describes the monthly decline as 2.2%. These are historical figures, not present-day inventory.
What the number can—and cannot—say
The narrow supported observation is that the reported inventory count was lower in the later month for the report's defined coverage. That does not establish why it changed, whether every neighborhood moved in the same direction or what will happen next.
Before translating a statistic into a client conversation, ask whether the property type, location and price segment are comparable to the client's question. A broad report can provide context while still being too broad to answer a property-specific question.
Do not silently replace the report's coverage with “all of Central Florida,” “the city of Orlando” or a particular ZIP code. A familiar place name in a headline is not a methodology. Read the scope note and retain the distinction that matters.
Check the calculation without changing the story
For a percentage change, subtract the earlier value from the later value, divide by the earlier value, then multiply by 100. Applying that calculation here gives approximately −2.247%, which rounds to −2.2% at one decimal place.
Keep the base period explicit. Month-over-month and year-over-year comparisons answer different questions. Never swap them because one produces a more dramatic headline. Likewise, a percentage change is not the same thing as a change in percentage points.
When a report supplies a rounded percentage, small differences from your own calculation may reflect rounding. Larger discrepancies deserve investigation. Check whether you copied the correct row, unit, comparison period and revised value before assuming the source is wrong.
Before and after: a headline that outruns the evidence
Before: “Inventory is dropping. Prices are about to soar—buy before it's too late.”
The statement moves from an observed quantity to a price forecast and then to pressure. It does not identify a period or coverage, and it provides no evidence for the prediction. A confident tone cannot fill those gaps.
After: “A market statistic becomes useful when you know what it measures. Read the period and coverage before applying a broad inventory change to one property.”
The revised opening gives the reader a practical skill. It creates room to show a dated example without pretending that a historical change dictates a personal decision today.
A suggested post based on the source
Educational draft—not a current market alert
Historical example: ORRA reported inventory moving from 13,306 in August 2025 to 13,007 in September, a 2.2% monthly decline within its report coverage. That does not predict a particular home's price. Before applying a broad statistic, check the relevant property type, location and current comparable information. Source: ORRA, September 2025 State of the Market.
Keep the source link beside the finished post, not buried where the audience cannot find it. If you adapt the example for another channel, retain the historical label and enough scope information to prevent the graphic from becoming a misleading standalone claim.
Use ChatGPT to preserve the evidence trail
Copy the entire block below, including the source note and URL, into a new conversation. No information from the rest of this article is needed to run this example. The historical PDF was checked again on October 24, 2026.
Write a short educational real estate post in American English using
only this verified historical source note.
SOURCE NOTE
Publisher: Orlando Regional REALTOR Association (ORRA)
Report: State of the Market, September 2025
Reporting comparison: August 2025 to September 2025
Metric: inventory, measured as number of properties
August 2025: 13,306
September 2025: 13,007
Published monthly change: -2.2%
Coverage: The report covers residential listings taken or sold by
ORRA brokers, regardless of where those properties are located.
Commercial property and vacant land are excluded. Do not relabel
this as a city, county or ZIP-code dataset.
Source page: 1
Source note verified in the editorial audit dated October 24, 2026.
Source URL:
https://www.orlandorealtors.org/clientuploads/Market%20Statistics/Monthly%20Market%20Reports/2025/September/State_of_the_Market_2025.09_Sep.pdf
RETURN
1. One observation supported by the note.
2. A plain-language explanation of the metric and calculation.
3. What this comparison cannot establish.
4. A post of no more than 90 words, clearly labeled as a historical
example, with its source link kept alongside it.
5. A short factual review list for the author.
Do not forecast prices, rates or sales, explain causes, invent
quotations or apply the aggregate result to a specific home.
If a required fact is absent or contradictory, identify the problem
before drafting that claim.For your own exercise, replace the source note with a verified summary of the report you are allowed to use. Provide the relevant facts and definitions rather than uploading an entire proprietary database. The assistant should work from the evidence you can actually inspect.
Adapt the complete input to another report
Fill every bracketed field before using this template. Keep the source beside your notes. If a field is unavailable, write that explicitly; the model should flag the gap rather than guess.
Create the same five outputs using this completed source note. First check every field. If a required value is missing or sources conflict, identify the issue and ask for clarification; do not draft the affected claim.
Publisher: [name]
Report and edition: [title/date]
Reporting period: [period]
Comparison: [earlier period] to [later period]
Coverage and exclusions: [exact scope]
Metric and unit: [definition/unit]
Earlier value: [value]
Later value: [value]
Published change, if supplied: [value or not supplied]
Source page/table and URL: [page/table/link]
Date I checked the source: [date]
Historical or current: [label with date]
Limitations and unresolved differences: [list or none identified]
Return: (1) supported observation; (2) definition and calculation;
(3) limits; (4) post of at most 90 words with appropriate date label
and source link alongside; (5) factual review checklist.
Use only the note. Do not invent causes, forecasts or property-specific
conclusions. A zero comparison base prevents percentage-change calculation.Build a source note before a content calendar
Your source note is the bridge between the report and every adaptation. Include the exact page or table you used, the access date and whether the information is historical or intended to describe current conditions. Keep your interpretation separate from the publisher's statements.
If an update changes a number, review the affected draft rather than changing a figure in isolation. The headline, comparison, caption and graphic may all depend on the original value. A corrected number inside an unchanged dramatic headline can still mislead.
Keep quotations short and attributed. A public report is not automatically permission to reproduce its entire design or redistribute the whole document as your own lead magnet. A small factual summary and a direct source link are often enough for an educational explanation.
Two refinement prompts
Audit the draft against my source note. Mark every claim that is directly supported, an interpretation, or unsupported. Check period, coverage, units and comparison base. Remove predictions and pressure that the source does not justify. Do not add new facts.
This pass is about accuracy, not making the post sound more sophisticated. If the source does not support a sentence, remove it or find a legitimate additional source before continuing.
Rewrite for a busy homeowner without losing the source, date, coverage or limits. Explain one useful question the reader can ask. Keep the historical example clearly labeled and avoid implying that a broad statistic determines their individual property value.
The second pass makes the explanation accessible. Shorter should not mean less accountable. Preserve the information that tells the reader what the number actually describes.
When the source is incomplete
If you cannot open the source, do not build a precise market update from a search snippet alone. Find an accessible primary publication or choose a different teaching example. If two versions disagree, identify which version you are using and resolve the difference before publishing a comparison.
If the only verified report is older, use it as a historical exercise, as this chapter does. Do not change the date to make it feel fresh. The teaching value comes from a method the reader can repeat with a current, appropriate source.
Without AI, the same workflow works on paper: copy the source note, write one supported sentence, add one limitation and propose one useful question. The manual version also makes a good check on an AI-generated draft.
Your reusable report-to-post record
PUBLISHER, TITLE AND SOURCE LINK: REPORTING PERIOD / ACCESS DATE: COVERAGE AND PROPERTY TYPE: METRIC, UNITS AND COMPARISON BASE: VERIFIED FACT: PLAIN-LANGUAGE INTERPRETATION: WHAT THIS DOES NOT ESTABLISH: PROPOSED POST AND HUMAN REVIEW:
Your next action
Choose one accessible report and one statistic. Write the source note first. Produce a single educational post that a reader can trace back to the evidence without needing to trust your confidence.
Separate the market explanation from the property presentation
You now have a supported market statement and a clear limit on what it means. A specific listing needs its own evidence too: photographs that show the rooms, features and context being discussed. A broad market statistic cannot do that job, and attractive media cannot establish a property’s value.
For an upcoming Central Florida listing, explore Magic Lens services and pricing to plan its visual presentation alongside your verified property information.
