Give the assistant a table of verified facts and one rule: it may describe the property and the place, never the people. Draft each neighbourhood page from that table, strip any wording that implies who lives there, and check every figure against its source. Rankings depend on how useful the page is, and nobody can promise them.
Fair-housing law prohibits advertisements that indicate a preference or limitation based on race, colour, religion, sex, national origin, familial status or disability, and steering means directing homeseekers towards or away from areas on such grounds. A neighbourhood page is advertising, so an AI draft that calls an area "family-friendly" or "safe" becomes your published statement. Some places protect further characteristics, so ask your broker what applies to you. None of this is legal advice.
Why neighbourhood pages are where AI drafts go wrong first
Assistants learned real estate copy from the open web, where "family-friendly", "safe and quiet", "perfect for young professionals" and "top-rated schools" appear on millions of pages. Ask for a neighbourhood description with no further instructions and the draft will describe a lifestyle and the kind of person who leads it. That is exactly the style the advertising rules push against. The trade association's magazine put it plainly in April 2022: avoid describing the buyer, so "ideal for empty nesters" or "perfect for students" may imply a preference for one group, whereas phrases such as "easy walk to train" describe a feature and are likely acceptable. It added that ads are judged on a reasonable-person standard that can change over time, so wording that passed once may not pass later.
The statute's advertising provision is broad. It covers any notice, statement or advertisement that indicates a preference, limitation or discrimination on the listed grounds, and it reaches anyone who makes or causes such an ad to be published. A page written by a tool and published by you is yours. A neighbourhood page is also a long-lived asset, which means a risky phrase can sit on a site for years and be copied into many other pages. The remedy is not to avoid neighbourhood pages, because they answer questions buyers really ask. It is to build them from facts and write them about places.
The facts table comes first: what you can verify, and where from
Before any drafting, fill in a facts table for each neighbourhood. Every row needs a value, a source with a date and a note of how you checked it. This is the filled-in table for an imaginary neighbourhood called Alder Park, with every figure invented for the example.
| Fact | Value | Source and date | How it was checked |
|---|---|---|---|
| Closed home sales | 86 in 12 months | MLS report to 31 October 2026, residential, inside the boundary on the page's map | Boundary and property types confirmed on the report |
| Median sale price | $412,000 | Same report | Median, not average; recalculated from the export |
| Median days on market | 19 | Same report | Definition checked: listing date to contract date |
| Home types sold | 61 single-family (71%), 25 townhouses (29%) | Same report | Counts add to 86 |
| Rail station | About a 12-minute walk from the centre | Walking route on a map, measured on 30 October 2026 | Route spot-checked on foot |
| Library, supermarket | 9 and 14 minutes on foot | The same map, same method | Opening hours taken from the operators' own pages |
| Alder Park (the park) | 14 acres, three tennis courts, a playground, a Saturday market | The city's parks page, checked 30 October 2026 | Market dates confirmed on the organiser's page |
| School assignment | Depends on the address | The district's boundary tool (link only) | Link tested; no ratings quoted |
Three habits keep this table honest. First, take every statistic from your MLS and read its definitions, because "days on market" can mean different things and a median is not an average. Check your MLS rules on publishing statistics and sold data before it goes on a public page. Second, measure distances yourself: assistants are unreliable about walking times, and a "walkable" claim needs a measured route behind it. Third, link to official sources for anything that changes or that you should not characterise, such as school boundaries, planning proposals and transit timetables, instead of copying their content onto your page. The market update tutorial shows how to pull and verify the sales figures.
The Complete Excel Course with AI Integration
From formulas and pivot tables to Power Query, DAX, VBA and AI with Copilot, ChatGPT and Claude — hands-on from the first lesson.
By Sagnik Bhattacharya and Virginia Thorn, Primetime Emmy Award winner
Buy nowA page template with a job for every section
A template stops each page from becoming a different experiment, and it is what makes review fast. Give the assistant these sections, in this order, and a word budget for each, so a 900-word page does not grow into 2,000 words of adjectives.
- Opening paragraph (60 to 80 words). What the neighbourhood is and where its edges are, with the sales count and the date range from the facts table.
- The market in three numbers. Median price, median days on market and number of sales, each with its period and source. Add one sentence on what a median is.
- The homes. Property types by share of sales, size and age ranges drawn from your own records, and the price range of recent sales.
- Getting around. Walking and driving times to named stations, stops and major roads, measured on a map, with links to the operators' timetables.
- Parks, shops and services. Named places with distances and facilities. Describe what is there, not who it is for.
- Schools and public information. Links to the official boundary tool and to published local data, with no commentary of your own unless your broker allows it.
- Current listings and recent sales. A feed or search link, which keeps the page current without rewriting it.
- Questions people ask. Four to six property-based questions, covered further down.
- How this page was made. One line giving the data dates and the name of the agent who reviewed it.
The last section helps readers and it helps you. Google's guidance on generative AI content suggests that explaining how content was created can give readers useful context, and a dated, named review line is also proof of the manual check that the same guidance calls critical.
Wording that implies who lives there, and what to write instead
Most compliance trouble in AI drafts comes from a short list of habits. The table pairs each with the problem and a property-based rewrite. Your brokerage and MLS may keep their own word lists, which can be stricter, so treat this as a starting point rather than a complete list.
Not every phrase in the table is unlawful in itself. Official advertising guidance issued in 1995 under the Act treats phrases such as "quiet streets", "family room" and "walk to bus-stop", and calling a neighbourhood desirable, as neutral descriptions, and says an advertisement describing a facility, such as an apartment complex with a chapel, does not state a religious preference on its face. The rewrites below are a cautious house style that leaves nothing for a reader to misread, and they also produce better pages, because each one swaps an adjective for a fact a buyer can check.
| Typical AI phrase | The problem | Write instead |
|---|---|---|
| "A family-friendly neighbourhood" | Describes familial status, a protected characteristic | "Alder Park has a playground and three tennis courts, and the library is a 9-minute walk" |
| "Safe and quiet" | "Safe" is a claim you cannot verify and is often read as a comment on the residents; "quiet" says nothing a buyer can check | Drop "safe"; replace "quiet" with the reason, such as "a cul-de-sac of 14 homes with no through traffic" |
| "Perfect for young professionals" or "ideal for empty nesters" | Describes the buyer, as the trade magazine's example warns | "Townhouses of 1,150 to 1,400 sq ft, a 12-minute walk to the rail station" |
| "Top-rated schools" | An unverified opinion that can stand in for who lives there | "School assignment depends on the address; check it on the district's boundary tool" |
| "A close-knit community" | Describes people, not the place | "A Saturday market in the park, on the dates listed on the organiser's page" |
| "Walk to church on Sunday" | Pictures the buyer's religious practice rather than describing the property | Use civic landmarks: "A 9-minute walk to the library" |
| "An exclusive enclave" | Can signal who is wanted or kept out; subjective | "Detached homes on lots of 0.2 to 0.3 acres, built between 1988 and 2004" |
| "Up and coming", "in transition" | Speculative, and can read as a comment on who is moving in | "Median sale price was $389,000 in the 12 months to October 2025 and $412,000 in the 12 months to October 2026 (MLS)" |
| "A diverse, multicultural area" | Describes residents by background, even when meant kindly | Remove; describe amenities and events by name and date |
One test sorts nearly every case. Read the sentence and ask whether it would still be true if the residents were completely different. If the sentence is really about the people, delete it. If it is about the property, a street, a distance or a service, keep it and attach a source. The test does not make you timid. It leaves a page that is more specific than the usual copy, with numbers, names and measured times where others have adjectives, and that specificity is also what makes the page useful to a buyer who is comparing areas.
A drafting prompt: facts in, rules fixed, gaps marked
Paste the facts table, then fix the rules in the prompt itself. Keep the prompt, with your banned-phrase list, in a Project (ChatGPT or Claude) or a Gemini skill (Google is replacing Gems with skills, which began rolling out in October 2026), so every page starts from the same rules.
Write a neighbourhood page for [neighbourhood] of about 900 words, using the
template below and ONLY the facts in the table I paste after it.
Rules:
1. Describe the property, the streets, distances and services. Never describe
the people who live there or imply who would like it or belongs there.
2. Do not use: family-friendly, safe, quiet, desirable, exclusive, up and coming,
close-knit, diverse, perfect for [any group], top-rated, good schools,
or any reference to religion, age, family status or background.
3. Every number needs its period and source in the sentence.
4. If a sentence needs a fact I have not given, write [VERIFY: what is needed]
instead of guessing. Do not estimate distances or travel times.
5. For schools and public safety, give only the official link I supply.
Template: opening paragraph; market in three numbers; the homes; getting around;
parks, shops and services; schools and public information; questions people ask;
how this page was made.
Finish with a list of every sentence you were unsure about.
[facts table]Even with those rules, expect slips, and read for them. An invented first draft and its edited version show the pattern.
First draft: "Alder Park is a vibrant, family-friendly neighbourhood with a safe, quiet feel. Young professionals love the trendy cafés, while empty nesters enjoy the peaceful tree-lined streets. With top-rated schools and a close-knit community, it is the perfect place to raise a family. Walk to church on Sunday or stroll to the park."
Edited: "Alder Park had 86 closed home sales in the 12 months to 31 October 2026 (MLS data), with a median sale price of $412,000 and a median of 19 days on market. Of those sales, 71% were single-family homes and 29% townhouses. The rail station is about a 12-minute walk from the centre of the neighbourhood, and the library and a supermarket are 9 and 14 minutes on foot. Alder Park itself covers 14 acres, with three tennis courts, a playground and a Saturday market. School assignment depends on the address: check it on the district's boundary tool."
The first draft contains at least eight problems: a family reference, two claims about safety and quiet, two descriptions of buyers, an unverified schools claim, a description of the community's people and a religious reference. None of them informs a buyer. The edited paragraph is roughly twice as long, and every sentence either states a measured fact or points to an official source. When the assistant kept "walkable" in a draft despite the rules, the fix was the facts table: no measured route, no claim.
Questions and answers that stay about property
Question-and-answer blocks suit neighbourhood pages, because buyers type questions and answer engines quote short answers. Choose questions whose answers are facts. Google stopped showing FAQ rich results from 7 May 2026, so don't expect a special search listing for them; the blocks are for readers and for answer engines, and they should be short and sourced.
- "What types of homes sell in Alder Park?" Answer from the facts table: 61 single-family homes and 25 townhouses in the 12 months to 31 October 2026.
- "How long do homes take to sell?" The median was 19 days from listing to contract in that period, and say what the median means.
- "How do I get to the rail station?" About a 12-minute walk from the centre; link the operator's timetable page.
- "Which schools serve Alder Park?" Assignment depends on the address; link the district's boundary tool and add nothing else.
- "What are property taxes like?" Quote the published rate from the tax authority's page with its date, and say that the bill for a given home depends on its assessment.
Questions to decline are just as useful to write down: "Is Alder Park safe?", "Is it good for families?" and "What kind of people live there?" An answer to any of them is a statement about residents. The honest reply on a page is to say that you describe homes and places, and to point to the official sources, if your policy allows. The tutorial on listing descriptions that pass fair-housing checks applies the same thinking to the text on a single listing.
When a buyer asks "which area is best for us?"
The same rules apply in conversation, and the questions on your pages usually start there. A buyer writes: "We have two children and a dog. Where should we look?" The steering risk is in the reply. Telling that buyer which areas suit families, or nudging them away from others, is the kind of conduct described as steering, and familial status is a protected characteristic just as race is. The safer pattern is to ask for criteria and answer with options that match them: budget, bedrooms, lot size, commute, the features they have named, and the areas where homes meeting those criteria are for sale. Send every client the same kind of information.
An assistant can draft that reply, provided it is told the pattern. A reply along these lines, written for this tutorial, shows it.
"Thanks for getting in touch. To narrow the search, could you tell me your budget range, the minimum number of bedrooms, whether you need a fenced garden, and the longest commute you'd accept? I'll send every area where homes matching that are currently on the market, with recent sale prices, the home types sold and travel times you can check yourself. You can then decide where to look first, and I'm happy to arrange viewings in any of them."
Notice what is missing. The reply doesn't mention the children, doesn't rank areas by who lives there and offers every matching area. Ask your broker for the office's own script, and add it to the Project instructions so the assistant uses it. The tutorial on fair-housing-safe prompts for realtors holds more wording of this kind.
How many pages before it becomes scaled content
Neighbourhood pages tempt you to publish dozens at once, because an assistant can write them in minutes. Google's spam policy defines scaled content abuse as many pages generated for the primary purpose of manipulating search rankings, and lists using generative AI tools to produce many pages without adding value for users among its examples. Its guidance on using generative AI content doesn't ban the tools. It says to focus on accuracy, quality and relevance, to fact-check and review all AI-generated content before publishing, calling a manual fact-check critical, and it notes that generative models predict likely wording rather than retrieve facts.
A practical test follows from that. For each page, write down five facts that appear on no other page of your site: a sales count, a median, a measured walking time, a named amenity, a dated planning item. If you can't find five, the page doesn't deserve to exist yet. Start with the five to eight areas where you have real sales and local knowledge, publish them one by one after review, and add more only as you gather facts for them. Fifty near-identical pages with the place name swapped are the pattern the policy describes, and they also tend to carry the same risky sentences fifty times.
A review routine before anything is published
Review is where the compliance happens, and it works best when each check has an owner and leaves a trace. This table suits a small team; a solo agent can be both columns, with the broker as the second pair of eyes for the fair-housing read.
| Check | Who | What is kept |
|---|---|---|
| Every figure matches the MLS export | The agent | The export and a ticked copy of the page |
| Distances and walking times measured | The agent | Map screenshots with dates |
| No banned phrases, no descriptions of people | A second reader | The page marked up with the banned-phrase list |
| Links work and go to official sources | The agent | A dated link check |
| Fair-housing read against your brokerage policy | Broker or compliance contact | Written sign-off |
| Data dates and reviewer shown on the page | The agent | The live page |
| Next refresh date in the calendar | The agent | The calendar entry |
Keep the marked-up draft and the sign-off for each page in one folder. If a complaint ever arrives, the folder shows how the page was produced and who read it.
What to expect after publishing, and what not to promise
Nobody can promise that a page will rank, and a page that does rank this month may not next month. Google's page on AI features says there are no additional requirements to appear in AI Overviews or AI Mode and no special optimisation or structured data needed, so the sensible plan is ordinary good practice: accurate, useful pages, clear internal links between neighbourhood pages and your live listings, and regular refreshes. The tutorial on being found in AI search covers what else is worth fixing on a real estate site.
Measure instead of hoping. Note the date each page goes live, then watch impressions and clicks for that group of pages in Google Search Console and count the enquiries that mention them. Judge over months, not days, and prune or improve any page that attracts nothing after a fair wait. A monthly routine, like the one in the small business AI SEO routine, keeps the work to an hour. Above all, never tell a client or yourself that a page will bring leads. The honest claim is that the page is accurate, current and written about places, which is what you can control.
Questions agents ask about neighbourhood pages and fair housing
Can I say a neighbourhood is family-friendly?
It is safer not to. The phrase describes who lives there or who should, and the fair-housing rules cover advertising that indicates a preference based on familial status among other characteristics. Describe facts instead: the parks, their facilities, distances and opening times. Ask your broker whether your brokerage policy allows any wording beyond that.
Can I link to school or crime information?
Many agents link to the official school boundary tool or published crime data rather than describing either themselves, because opinions about schools or safety can stand in for who lives somewhere. Whether your policy allows even a link is a question for your broker. If it does, link to the source and add no commentary.
Is a page written by AI penalised by Google?
Google's guidance says generative AI is not banned in itself. The risk is publishing many pages without adding value for users, which its spam policy calls scaled content abuse. A page with checked, unique local facts written for readers is a different thing from fifty near-identical pages with the place name swapped.
How often should I update neighbourhood pages?
Refresh the numbers on a fixed schedule, such as quarterly, and show the date range on the page. Check transit, planning and amenity links whenever you update. A page that quotes sales data from two years ago is both less useful and easier to get wrong, so put the refresh dates in your calendar.
Further reads
- Leasing Chatbot for Real Estate: A Fair Housing Checklist Before You Pay — The same fair-housing discipline applied to a leasing chatbot.
- Making a Website Using AI With Good SEO: A Step-by-Step Build — Build the wider website these pages sit inside.
- Best AI for SEO Content: A Refresh Workflow for Client Articles — A routine for refreshing pages as the data ages.
- AI Real Estate Social Media: A Month of Posts From Your Listings — Turn the verified facts into a month of posts.
- Home Inspector Website: Local Pages and Sample Report Copy With AI — Another trade's local-page template, for comparison.
- AI Disclosure in Real Estate: What Listing Photos and Copy Must Say — What listing photos and copy must disclose.
- Best ChatGPT Plan for Realtors: Free, Plus or Business? — Free, Go, Plus or Business for a solo brokerage: what each costs, which ones show ads, and what client data each plan can safely hold.
- Best Home Staging AI Apps: How to Compare Them on Your Own Rooms — The five rooms to test, a scoring sheet, verified prices per image for five staging apps and labelling wording that keeps listings honest.
- Real Estate AI Chatbot Demo: A Test Checklist Before You Sign — Five tests to run on your own listings before you sign: price accuracy, viewing bookings, fair-housing traps, handoff and lead export.
- AI Lead Generation for Real Estate Agents: No Bought Lists Needed — A consent-first plan for agents: nurture your sphere, publish local answers and capture website enquiries with AI, with no bought or scraped lists.
- AI Marketing for Real Estate Agents: A Four-Week Content Plan — Week by week: a listing launch, an area guide, a market update from your own figures and a proof week, each turned into posts, a script and an email.
- AI Prompts for Appraisers: Narratives, Neighbourhoods, Reconciliation — A five-prompt pack for appraisers with sample output: the assistant drafts from your verified facts, and you supply and check every conclusion.
- AI Real Estate Listing Presentation: Build the Seller Deck — How a realtor can build a listing presentation with AI: the slide spine, a worked comps page, a costed marketing plan and checks before the meeting.
- AI Tools for Real Estate Brokers: From Transactions to Agent Support — Five tool types for a small brokerage, each with published costs and watch-outs, a worked cost for 12 agents and a 90-day adoption plan.
- AI Policy for a Real Estate Brokerage: What Agents May and May Not Do — Turn a broker AI policy into rules an agent can follow on a Tuesday: do and don't lists, sample clauses and a sign-off sheet.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: the Fair Housing Act's advertising provision and an official fair-housing summary, NAR REALTOR Magazine's fair housing advertising article (April 2022), the 1995 official guidance memo on advertising under the Act, and Google's spam policies, generative AI content guidance, AI features page and Search Central changelog (all checked November 2026). The neighbourhood and every figure in the examples are invented.