Real Estate Website SEO (2026): Checklist + Service-Area Pages

Only 12 percent of agents say their own website generates leads. That is from NAR’s 2025 technology survey, and it is the most useful number in real estate SEO, because it tells you how low the bar actually sits. This guide is the on page and service area page system behind my real estate coaching.

What real estate website SEO can and cannot win

Start with the ceiling, because most agents waste a year discovering it the slow way. Zillow reported 220 million average monthly unique users in the first quarter of 2026. It ranks first or second for “homes for sale,” a term with roughly 450,000 US searches a month, and for “houses for sale near me” at about 550,000. Realtor.com, Redfin, and Homes.com occupy most of what is left. There is no version of your website, your content calendar, or your budget that takes those positions.

That is not a reason to skip search. It is a reason to be precise about which searches you are going after. The portals dominate inventory queries, meaning searches where the person wants to browse listings. They are noticeably weaker on agent selection queries and on hyper local questions that require someone to actually know the place. They are weaker again on the long tail of neighborhood and situation specific searches that no national database answers well.

Look at the shape of the demand. “Realtors near me” runs about 74,000 searches a month in the United States at a $7.30 cost per click. The results page there is not a wall of portals, it includes independent brokerages and single agent domains. Drop to a real market and the numbers get small fast. “Homes for sale in fairfax va” is roughly 1,300 a month. “Fairfax va real estate agent” is about 10 a month at a $10.35 cost per click. “Best realtor in fairfax va” is around 20.

Ten searches a month sounds like nothing until you price it. At a $10 cost per click and a realistic click share, that page might send you two or three visitors a month. If one of them becomes a client every eighteen months, and your average net commission is $9,000, the page has paid for the four hours it took to write it many times over. The mistake is not that the numbers are small. The mistake is building forty of these pages as though they were free.

The second reality check comes from the demand side. NAR’s 2025 Profile of Home Buyers and Sellers covers transactions from July 2024 through June 2025. It found 52 percent of buyers found the home they purchased on the internet and 27 percent found it through an agent. But when the same buyers were asked how they found their agent, 43 percent said a friend, neighbor, or relative, and 15 percent had used the agent before. Referral still dominates agent selection by a wide margin.

So the honest job description for your website is narrow and worth doing. It has to convert the referral who is checking you out before they call, which is most of your traffic and the part nobody measures. It has to catch the minority who genuinely start with a search box and no name in mind. And it has to give your other channels somewhere credible to land. If you approach it expecting to replace your lead generation, you will be disappointed and you will quit at month five.

NAR’s 2025 REALTORS Technology Survey, which drew 1,241 responses, asked which technology tools generated the most quality leads in the prior twelve months. Social media came first at 39 percent, CRM second at 23 percent, MLS at 17 percent, brokerage website at 13 percent, and the agent’s own website at 12 percent. Twelve percent is a low number, and it is precisely why the opportunity is real. Your competition in local organic search is a field of abandoned template sites with three pages of stock copy on them.

The advice almost every real estate SEO guide gets wrong

I read twenty of the currently ranking articles on real estate website SEO before writing this one. Every one of them tells you to build location pages. Not one of them quotes the Google policy that governs whether those pages are allowed to exist. The phrase “scaled content abuse” appears in zero of the twenty. The phrase “crawled, currently not indexed” also appears in zero of the twenty. That is remarkable, given it is the single most common outcome for the pages they tell you to build.

The advice on how many pages to build is not just missing, it is wildly contradictory. One guide says a minimum of 300 words per area page. Another says 800 to 1,200. One vendor advertises automatically generated location pages as a product feature. Another tells agents to build pages down to specific roads or blocks. None of them cite a source for any of it, and none of them acknowledge that the others exist.

The result is a very specific and very common failure. An agent buys a platform and uses the location page generator. It produces thirty or fifty pages titled “Homes for Sale in [City]” with the same four paragraphs and a swapped city name. Then he waits. Six months later, ten of the pages are indexed, twenty sit in Search Console under “crawled, currently not indexed,” and the site’s older pages have quietly slipped too. Nothing announced itself. No manual action arrived. The site simply got heavier and less trusted.

What follows is the part the vendor guides leave out: the actual policy text, what it means in practice, and how to build area pages that survive it.

What Google’s spam policies actually say

Google publishes its spam policies at developers.google.com/search/docs/essentials/spam-policies. Two sections govern location pages. The first is doorway abuse. Google lists as one example “having multiple domain names or pages targeted at specific regions or cities that funnel users to one page.” It lists as another “creating substantially similar pages that are closer to search results than a clearly defined, browseable hierarchy.”

Read that first example again slowly, because it describes the standard real estate city page almost exactly. A page targeted at a specific city whose purpose is to funnel the visitor to a contact form or an IDX search is, on its face, the thing the policy names. What saves a page from that description is not the disclaimer at the bottom. It is whether the page has independent value to someone who lands on it and never converts.

The second section is scaled content abuse, which Google added on March 5, 2024. The text reads: “Scaled content abuse is when many pages are generated for the primary purpose of manipulating search rankings and not helping users. This abusive practice is typically focused on creating large amounts of unoriginal content that provides little to no value to users, no matter how it’s created.”

The last five words are the ones that changed the game. The policy this replaced was called spammy automatically generated content, and it was scoped to automation. Google removed that scope deliberately. Google said it was “strengthening our policy to focus on this abusive behavior.” The target is “producing content at scale to boost search ranking, whether automation, humans or a combination are involved.” Writing your fifty city pages by hand does not exempt you. Neither does having a contractor write them, and neither does editing an AI draft.

There is a third piece of context that matters. In March 2024, Google folded its Helpful Content system into its core ranking systems, which the company confirmed in its ranking systems documentation. Before that, a site could be hit by a discrete helpful content update and recover when the next one ran. Now the assessment is continuous and site wide. A block of thin pages is not quarantined to itself, it becomes part of how the whole domain is evaluated.

Google employees have been unusually direct about this specific tactic. In December 2019, responding to a plan to build roughly 1,300 city landing pages, John Mueller said “that sounds like doorway pages, not something I’d recommend.” In July 2023 he said “programmatic SEO is often a fancy banner for spam.” Neither statement is policy, but both tell you how the pattern is regarded internally.

Now the complication, because it would be dishonest to leave it out. In a mid 2024 office hours session, asked whether localized content counts as duplicate content, Mueller answered plainly that it does not. Both things are true at once, and the space between them is the entire subject of the next two sections. Localized content is fine. Localized templates are not. The difference is whether the local part is the content or just the variable.

Doorway page or real service area page: the five questions

The line between a legitimate service area page and a doorway page is not a word count and it is not a checklist item. It is whether the page would still be worth publishing if search engines did not exist. That sounds abstract, so here is how to make it operational. Run every proposed page through five questions before you write a word of it, and build only the ones that clear all five.

Question three is the one agents argue with, so let me defend it. The reason it matters is not ethical, it is practical. If you have not worked in a place, you cannot write the paragraph that makes the page unique. You will write the generic one instead, and the page fails questions one and two automatically. Experience is not a moral prerequisite, it is the raw material. Without it you have nothing to say that a competitor with the same MLS access cannot say faster.

This is also how the apparent contradiction in Google’s own commentary resolves. Localized content genuinely is not duplicate content, exactly as Mueller said. A page about Vienna, Virginia that reports Vienna’s price bands, Vienna’s absorption rate, and a Vienna transaction you closed is localized content. A page about Vienna that is your Reston page with three nouns replaced is a template with a variable in it. Google is not contradicting itself. It is describing two different things that look identical in a site map and nothing alike on the page.

How many service area pages should you actually build?

Nobody will give you a number, so here is mine, with the reasoning, so you can adjust it rather than obey it. The constraint is not Google’s tolerance. It is your capacity to make each page genuinely different, which is finite and much smaller than agents assume.

Assume a real service area page takes four to six hours to build properly the first time, because you are pulling market data, writing original commentary, and sourcing photographs. Assume it takes another forty five minutes per quarter to keep the numbers current, since stale statistics are worse than no statistics on a page whose entire claim is local expertise. Forty pages is therefore roughly 200 hours of construction and 120 hours a year of maintenance forever. That is a part time job you did not intend to take.

Notice that even the largest tier tops out well below what platform generators produce by default. A generator that offers to create a page for every municipality in a metro area is offering you a liability with a progress bar. If the tool builds them, the tool cannot make them different, and different is the only thing that matters.

There is a sequencing rule too. Build one page. Get it indexed and ranking. Only then build the second. Agents who launch twenty pages simultaneously learn nothing from the launch, because if the batch underperforms they cannot tell which variable failed. Agents who launch sequentially find out within six weeks whether their template is working, and can fix it before it is replicated nineteen times.

The service area page blueprint

Every competitor guide offers a template, and a template is exactly what gets you in trouble. So what follows is a set of modules with a requirement attached to each. The module list can repeat across pages. The content inside each has to be independently researched, or the module should be dropped from that page rather than padded.

The order matters as much as the contents. Put the IDX search module below the written substance rather than at the top. This runs against the instinct to show inventory immediately. It is also the single change that most reliably moves a page from the doorway category into the resource category, because it changes what the page is for.

On length, ignore the 300 word floor and the 1,200 word target you will read elsewhere. Both numbers are invented. A page that fills all twelve modules honestly will land somewhere between 900 and 2,000 words without you counting. A page that cannot fill them should be shorter, not padded to a target.

Writing the parts nobody can copy

Three modules on that list carry almost all of the defensive weight, and they are the three that a competitor with the same MLS access and the same website platform cannot reproduce. Get these right and the rest of the page can be ordinary.

The first is the transaction paragraph. Not a testimonial, a case description. Something like: a 1980s colonial on a cul de sac that had been sitting at the wrong price for ninety days, repriced and restaged, sold in eleven days for $12,000 over the reduced ask. Anyone reading that learns something about the local market’s tolerance for pricing errors, and no competitor can write it because it did not happen to them.

The second is the surprises paragraph. Every market has two or three facts that consistently blindside buyers who are new to it. A septic system where everyone assumed public sewer. A school boundary that splits a subdivision down the middle. A commute that is fifteen minutes at ten in the morning and fifty five at eight. Front loaded assessments in a newer development. Writing this paragraph well requires having watched a deal nearly die over it, which is why it almost never appears on competitor pages.

The third is the pricing reality paragraph, and it is the one most agents refuse to write because it involves saying something unflattering. What does not sell here, and why. Which product type sits. Where sellers are consistently a full price band optimistic. A page that says the split levels on the north side are taking twice as long to move as the colonials is doing something no brochure page does. It is telling the truth at a small cost to itself. That is also the paragraph readers remember and forward.

Photographs deserve their own note, since the effort to value ratio is unusually good. Twenty minutes with a phone camera at the neighborhood entrance, the community amenities, the school, and the main commercial strip gives you five original images per page. Name the files descriptively, write alt text that describes what is actually shown, and you have both an accessibility win and an originality signal that no template can generate. This is the same discipline that makes geographic farming work offline, applied to a web page.

Page viability check

Before you commit four hours to a page, score it. Answer for one specific place, and check only what is true right now rather than what you intend to arrange later.

If you scored a page at three to five, the useful move is not to write it anyway with the gaps papered over. It is to go get the missing inputs, which usually means an afternoon in the MLS and an hour driving the area. That is a much better use of the time than publishing something that will sit unindexed.

The 12 point on page checklist

Each row pairs the rule with the mistake that undoes it, because the rules are not where agents go wrong. Everyone knows to write a title tag. Fewer people know that using the same title pattern across thirty pages is what tells a classifier those pages are a set rather than individual resources.

Item eight has changed materially and is covered in its own section below, because most of what you will read about schema for real estate is now out of date. Item nine is worth a specific audit on any IDX powered site, since incorrect canonical tags are one of the most common and most damaging defaults in the category.

Title tags and meta descriptions that survive rewriting

Google frequently rewrites title links in search results, using page content, headings, and anchor text when it judges the supplied title to be a poor match for the query. Your title tag is therefore a strong suggestion rather than an instruction, and the way to have it respected is to make it accurate and specific rather than optimised.

Three habits get titles rewritten. Keyword stuffing is the obvious one, as in a title carrying the city name three times. Boilerplate is the second, where every page ends with the same forty character brand and location suffix that pushes the actual subject out of view. Mismatch is the third and most common, where the title promises something the page does not deliver, so Google substitutes an H2 that describes the page more honestly.

For area pages specifically, vary the construction deliberately. If every page is “Homes for Sale in X, VA | Firm Name,” you have built a pattern. Some pages can lead with the neighborhood guide angle, some with the market report angle, some with the buying process angle, depending on what each page actually contains most of. This is not a trick to evade detection. The titles differ because the pages differ, which is the entire point.

Meta descriptions do not influence ranking, and Google rewrites them even more often than titles. They still matter, because they influence whether the person clicks. Write them as the two sentence pitch you would give if you had one chance. Include the specific detail that distinguishes your page from the four portal results above it, and keep them unique. A duplicated description across an area page set is another set signal.

Internal linking that consolidates authority

Internal linking is where most real estate sites leak the value they have earned. The typical structure is a home page, a blog with sixty posts pointing nowhere in particular, and a set of area pages linked only from a footer menu. Everything is technically reachable and nothing is emphasised, so the site tells Google that all pages are equally important, which means none of them are.

The structure that works is a shallow hierarchy with deliberate emphasis. Three or four top level sections. Service pages for what you do, area pages for where you do it, and guides for the questions people ask before they are ready to call. Every guide links up to the relevant service or area page. Every area page links to two or three guides that its readers would plausibly want. Nothing sits more than three clicks from the home page.

Two rules keep it from degrading. Never more than one internal link per paragraph, because a paragraph with three links reads as a link farm to a person as well as to a classifier. And no orphans: if a page cannot be reached from anywhere except the sitemap, it is telling Google nobody on your own site thinks it is worth mentioning.

Anchor text should describe the destination in the words a reader would use. Exact match repetition across dozens of links is a pattern worth avoiding. The bigger problem in practice is the opposite, where every link says “click here” or “learn more” and passes no information. If you are building this alongside a broader content plan, the sequencing in my real estate content marketing strategy guide covers which pages to publish in what order.

The IDX problem nobody selling you an IDX will explain

Your IDX feed puts MLS listing data on your website. The same feed puts identical data on every other participating agent’s website in your market, and on the portals, which have vastly more authority than you do. From an indexing standpoint you are publishing hundreds or thousands of pages that exist in near identical form elsewhere. Most of those other places are stronger than yours.

This is not a penalty situation. Google is clear that duplicate content across URLs does not trigger a manual action. Its own starter guide says plainly that if content sits under multiple URLs, “it’s fine; don’t fret about it,” while noting that it is inefficient. Inefficient is the operative word. The pages will generally not rank, and in volume they change the ratio of thin to substantial pages on your domain.

NAR’s IDX policy is worth knowing here because agents routinely assume it forbids indexing, and it does not. Policy Statement 7.58, effective January 1, 2026, states that MLS participants may not use IDX provided listings for any purpose other than IDX display. It then says explicitly that this “does not require participants to prevent indexing of IDX listings by recognized search engines.” Whether to index is your decision, not your MLS’s. Your specific MLS may layer its own rules on top, so check.

The more damaging problem is faceted search. IDX search interfaces generate URLs for every combination of filters a visitor can select: price band, bedrooms, bathrooms, property type, sort order, page number. Google’s own documentation on faceted navigation warns about overcrawling, noting that crawlers “will typically access a very large number of faceted navigation URLs.” It warns separately about slower discovery of the pages you care about.

The fix is unglamorous and effective. Let the crawler reach your area pages, your service pages, and your guides. Keep it out of parameterised search result URLs, either through robots.txt disallow rules on the parameter patterns or by making those states non crawlable in the interface. Keep individual IDX detail pages out of your sitemap. Make sure canonical tags on IDX pages point somewhere sensible rather than defaulting to the home page, which is a live defect on a surprising number of platforms.

None of this makes IDX a bad purchase. It makes it a tool with a specific job, which is serving inventory to people already on your site, not acquiring them from search. If you are still choosing a platform, the trade offs between them are covered in my comparison of real estate website builders.

Whether the IDX itself earns its subscription is a separate question, worked through in my analysis of whether IDX websites generate leads. The short version is that the feed is a retention tool, not an acquisition one.

Crawl budget, and whether you actually have one

Crawl budget is discussed constantly in real estate SEO and applies to almost none of the sites discussing it. Google’s own documentation sets the thresholds plainly. Crawl budget is a consideration for sites with more than a million unique pages where content changes weekly. It also applies to sites with more than 10,000 unique pages where content changes daily. A 40 page agent site with a blog is nowhere near either line.

The exception, and it is a real one, is the site whose page count is inflated by machine generated URLs. An agent site with 40 written pages can easily expose 200,000 crawlable URLs once faceted IDX search combinations are counted. That site does have a crawl efficiency problem, and it did not get there by publishing too much. It got there by leaving a URL generator open to crawlers.

So the diagnostic is simple. Open Search Console, look at crawl stats, and compare the number of URLs Google is requesting against the number of pages you have actually written. If those numbers are within an order of magnitude of each other, crawl budget is not your problem and you should stop reading about it. If Google is fetching thousands of URLs a day on a site with sixty real pages, you have a faceted navigation problem to close.

Two related myths deserve killing here. The first is that submitting URLs through Google’s Indexing API speeds up indexing for ordinary pages. It does not, and it is not permitted. Google’s documentation states the Indexing API “can only be used to crawl pages with either JobPosting or BroadcastEvent embedded in a VideoObject.” Real estate pages qualify as neither. The second is that a large sitemap helps. Sitemaps aid discovery, they do not confer quality, and Google’s own guidance suggests they become genuinely useful past roughly 500 pages.

“Crawled, currently not indexed”: what it means and what to do

This is the status agents see most often and understand least, and none of the twenty competitor guides I read mentions it at all. Google’s definition in the Search Console documentation is precise and worth reading literally: “The page was crawled by Google but not indexed. It may or may not be indexed in the future; no need to resubmit this URL for crawling.”

Three things follow from that sentence. Google found the page, so this is not a discovery or a technical access problem. Google made a decision, which means resubmitting will not change the answer. And the decision is revisable, which means fixing the page genuinely can move it. What it will not respond to is repeatedly clicking request indexing, which is the first thing almost everyone does.

Why does it happen? Google does not publish a reason per URL, so anything specific is inference. But John Mueller addressed the pattern directly in a mid 2026 Search Off the Record episode. He said that where Google’s systems are “seriously worried about the quality of a website, that they will reduce the number of pages that they index.” Non indexing is being described there as a quality response, not a capacity limit.

The remediation ladder runs in this order, and the first step is the one people skip. Compare the unindexed page against the indexed ones on your own site and identify what the indexed ones have that it lacks. Usually the answer is specific data, original imagery, or first hand detail. Add those things substantively, not cosmetically, then update the page’s modified date honestly and link to it from a page that is already indexed and getting traffic.

If a page remains unindexed after two months of that treatment, accept the verdict. Merge its useful content into a stronger neighboring page, redirect it there with a 301, and reduce your page count. A site with eight indexed, ranking area pages outperforms a site with thirty pages of which eight are indexed. In the second case the other twenty two are still counted when your domain is assessed.

Core Web Vitals: the exact numbers, and how much they really matter

The metrics and thresholds are unchanged going into the second half of 2026, and they are measured at the 75th percentile of real user visits, split between mobile and desktop. Interaction to Next Paint became a stable Core Web Vital on March 12, 2024, replacing First Input Delay, which was retired from Chrome tooling in September 2024. If a consultant is still reporting FID to you, their toolset is two years stale.

Now the part vendors overstate. Page experience is a modest ranking input, not a lever. Google’s current documentation says there is “no single signal,” and that its core systems look at a variety of signals aligning with overall page experience. Then it adds the line that settles most arguments: “Google Search always seeks to show the most relevant content, even if the page experience is sub-par.” Relevance wins. A faster page does not outrank a better one.

Which does not make speed unimportant, it relocates why it matters. A six second load on a phone loses a meaningful share of visitors before they see anything, and that happens whether you rank third or thirteenth. Treat Core Web Vitals as a conversion problem that has a small ranking side effect, and you will make better decisions about how much to spend fixing them.

On real estate sites the causes are predictable. Unoptimised hero photography is the usual LCP offender, since a 4MB image of a listing will fail the threshold on mobile data no matter what your host charges. INP problems come from script load, typically chat widgets, multiple analytics tags, and IDX bundles all competing on the main thread. CLS comes from images without dimensions and from ad or widget containers that inject themselves after the text has rendered.

Fix them in that order and stop. Compress and correctly size images, set explicit width and height attributes, and lazy load anything below the fold. Audit third party scripts, remove the ones nobody looks at, and defer what remains. That sequence resolves the large majority of failing real estate pages without a rebuild. If your platform makes those changes impossible, that is information about your platform.

Mobile only crawling, and what quietly breaks

Google completed its move to mobile only crawling on July 5, 2024. There is no longer a desktop crawl to fall back on. Whatever Googlebot sees on a smartphone is the entirety of what it knows about your page. Anything your responsive theme hides at narrow widths is, for practical purposes, not on the page.

Real estate templates break this in a small number of repeatable ways. Long descriptive text collapsed behind a “read more” control that loads content only on tap. Property detail tables hidden below a mobile breakpoint. Agent biography and credential blocks dropped from the mobile layout to save space, which removes precisely the expertise signals the page most needs. Structured data emitted only in the desktop template.

Google warns about the subtler version of this too. It adds that “even with the equivalent content, differences in DOM or layout between desktop and mobile page can result in Google understanding the content differently.” Content that is present but reordered or restructured on mobile can be weighted differently. The test is not whether the words exist somewhere in the file, it is what the mobile rendering actually presents.

Checking this takes ten minutes. Load your most important area page in a mobile viewport with a device emulator, and read what is actually rendered. Then run the same URL through Search Console’s URL inspection and view the rendered HTML. If the market data table, the transaction paragraph, or the author block is missing from either view, that is a content problem being reported as a design decision.

Schema markup in 2026: what still earns a rich result

This section corrects the most widespread factual error in current real estate SEO writing. Two of the strongest ranking competitor guides, both updated in mid 2026, tell agents to add FAQ structured data in order to earn a rich result. Google’s own documentation says the opposite, and has done since before either article was revised.

The FAQPage documentation states: “As of May 7, 2026, FAQ rich results are no longer appearing in Google Search. We will be dropping the FAQ search appearance, rich result report, and support in the Rich results test in June 2026,” with Search Console API support ending in August 2026. The restriction began earlier, in August 2023, when Google limited FAQ rich results to well known authoritative government and health sites. No agent website has been eligible since 2023, and now nobody is.

HowTo rich results were deprecated on August 15, 2023 and are similarly gone. There is no real estate listing rich result available to individual agents in Google’s search gallery. Google’s LocalBusiness documentation does not name RealEstateAgent as a supported type. Claims that Google “supports RealEstateAgent schema” for rich results should be treated with suspicion.

The review restriction catches people out, so here is the exact wording. Google’s rule is specific. If the entity being reviewed “controls the reviews about itself,” its pages using LocalBusiness or any other Organization type “are ineligible for star review feature.” Testimonials you collect and publish on your own site do not produce stars. This is not a technical failure, it is the intended behaviour.

One rule underpins all of it, and violating it is the fastest route to a structured data manual action: “Don’t mark up content that is not visible to readers of the page.” Marking up FAQs that do not appear, reviews that are not shown, or a business address that is nowhere in the rendered output is a policy violation, not a shortcut.

AI Overviews and real estate: the data that should calm you down

If FAQ rich results are gone and AI answers are eating clicks, the reasonable question is whether any of this is still worth doing. The data says yes, and more specifically it says the part of your search presence that matters most for getting clients is the part least exposed to AI answers.

The clearest picture comes from an Ahrefs analysis of 146 million desktop search results pages in September 2025. AI Overviews appeared on 20.5 percent of all results pages, but the distribution by intent is dramatic. Informational queries triggered them 21.4 percent of the time. Commercial queries 4.3 percent. Transactional queries 2.1 percent. Navigational queries 0.9 percent. The real estate category as a whole sat at 5.8 percent, and local “near me” style queries at 7.9 percent.

Read that against what you are actually trying to rank for. “How much does a real estate agent charge” is an informational query, heavily exposed. “Realtor in Vienna VA” is a local commercial query, barely exposed. Your service area pages target the second category almost exclusively. The blog posts you write to build topical coverage target the first, and those are the ones that will lose impressions to AI answers.

The damage where AI answers do appear is real and should not be minimised. Pew Research Center tracked 900 US adults across 68,879 searches in March 2025. When an AI summary was present, 8 percent of users clicked a search result, versus 15 percent when no summary appeared. Only 1 percent clicked a link inside the summary itself. Users were also more likely to end their session entirely, 26 percent versus 16 percent.

A separate Ahrefs study compared 300,000 keywords between December 2023 and December 2025. It found position one click through rate down 58 percent where AI Overviews appear, with position ten down 19.4 percent. Google’s own framing is more optimistic. Liz Reid wrote in August 2025 that “overall, total organic click volume from Google Search to websites has been relatively stable year-over-year.” She added that “average click quality has increased.” Google published no data supporting either claim, so weigh it accordingly.

The strategic conclusion is not complicated. Keep investing in the local commercial pages, because they are where the clients are and where AI answers are thinnest. Keep writing the informational guides, but expect them to earn fewer clicks per impression than they did in 2023. Judge them on the topical authority and internal linking value they provide rather than on their own traffic line.

Local pack versus local organic: two different games

Agents routinely conflate ranking in the map pack with ranking in the blue links below it, and then optimise the wrong asset. They are separate systems with substantially different inputs, and knowing which one your work affects prevents a lot of wasted effort.

Whitespark’s Local Search Ranking Factors study, published in November 2025 with input from 47 practitioners, separates the two. For the local pack, the highest weighted factors are the primary Google Business Profile category and physical proximity of the searcher to the business. Keywords in the business title and an address in the city being searched follow. Almost none of that is on your website.

For localised organic results, the picture inverts. The top factors are dedicated service pages, geographic keyword relevance, and inbound link authority. That is your website, and specifically the service area page architecture this article has been describing. So the map pack is largely a Google Business Profile and location problem, and the organic results are a content problem.

The practical implication is that if you want the map pack, the work is in your profile rather than your site. That is a different discipline, covered in my Google Business Profile setup guide. If you want the organic results underneath it, everything in this article applies and the profile is a secondary concern.

The 90 day build order

Sequence matters more than effort here, because doing the right things in the wrong order means you cannot tell what worked. This is the order I would run if I were starting over on a site with a handful of thin pages and no rankings.

Ninety days will not produce rankings for competitive terms. It will produce an indexed, structurally sound site with three good area pages, which is the asset that produces rankings over the following six to twelve months. Anyone promising the compressed version is selling you the thing that gets sites suppressed.

The four numbers that matter

Real estate SEO reporting is usually a wall of vanity metrics that move independently of whether anyone called you. Four numbers tell you almost everything, and all four live in Search Console.

The first is indexed pages as a proportion of published pages. If you have published thirty and Google has indexed eleven, no amount of ranking work matters until that ratio improves, because nineteen of your pages are not competing at all. This is the number to check first and the one almost nobody looks at.

The second is impressions segmented by page group, comparing area pages against guides against service pages. Total impressions are noise. Impressions on the pages you intend to make money from are signal, and the two frequently move in opposite directions.

The third is the count of queries where you rank between position eight and twenty. That band is your pipeline. Those are pages Google already considers relevant but has not yet promoted, and a page at position eleven usually needs an afternoon of work rather than a rewrite. Working that list is the highest return activity in the whole discipline, and it is almost always more productive than publishing something new.

The fourth is what happens after the click, which Search Console cannot tell you. Tag your contact forms by source page and track which pages produce actual conversations. Agents are consistently surprised here. The page with the most traffic is rarely the page producing the most calls, and once you know which page converts you know which page to build more of. Conversion benchmarks by channel are covered in my breakdown of real estate lead conversion rates.

What not to measure. Domain authority scores from third party tools are vendor inventions Google does not use. Keyword rankings checked daily fluctuate meaninglessly at that frequency. Total organic sessions mixes branded traffic from people who already know you with the acquisition traffic you are trying to grow.

If you already built 50 thin city pages

A large share of the agents reading this have already done the thing this article warns against, usually because a platform made it a two click operation. That is a recoverable position, and the recovery is straightforward, though it involves deleting work you paid for.

Start with an inventory. Export every URL from Search Console with its index status, impressions, and clicks for the last twelve months. Sort by clicks. On a typical over built site, a handful of pages carry effectively all of the performance and the rest have never received a visitor from search. That distribution is the whole diagnosis.

Then sort every page into one of three piles. Keep and improve, for pages with real traffic or real potential in an area you genuinely work. Merge, for pages covering an area adjacent to a stronger page, where the useful content can be folded into that page as a section. Remove, for pages about places you have never worked and never will, which is usually the largest pile.

Expect the site to look worse before it looks better. Page count drops, total impressions drop, and if you are watching a dashboard weekly you will feel like you broke something. The recovery shows up in the ratio first, as the proportion of indexed pages climbs, then in impressions on the surviving pages, typically two to four months out. Assessment of a domain is continuous now, so the improvement is gradual rather than switch like.

Do the improvement work on the survivors before you build anything new. Adding a strong page to a site still carrying forty weak ones asks the new page to overcome the drag of everything around it, which is a slower path than removing the drag first.

If you have no time: the 20 percent that gets 80 percent

Most agents reading this will not run a 90 day program. So here is the honest short version, in priority order, for someone with roughly one day a month.

Fix your Google Business Profile properly, because it moves fastest and it drives the map pack, which is a different system from your website and much easier to influence. Then write one excellent service area page for the place you actually work, with your own data and photographs, and give it three months. Then write the second one. Then keep the numbers on both current.

Three genuinely good pages, maintained, will out perform thirty generated ones in every dimension that matters. That means indexing rate, ranking, conversion, and the amount of your life it consumes. That is not a compromise version of the strategy. On the evidence, it is the strategy, and the longer version simply repeats it more times.

Questions agents actually ask

How long does real estate SEO take to show results?

For a new site targeting local terms, expect first meaningful impressions around month three and the first consistent leads between months six and twelve. Pages have to be crawled, indexed, tested against real queries, and re evaluated as engagement data accumulates. Anyone quoting faster is selling something.

Can I rank above Zillow?

Not for inventory queries such as “homes for sale in [city].” Those results are dominated by portals with authority you cannot accumulate. You can rank above them for agent selection queries, hyper local questions, and neighborhood searches where a national database has nothing distinctive to say.

Are city landing pages against Google’s rules?

Not inherently. Google’s doorway abuse policy names pages “targeted at specific regions or cities that funnel users to one page.” A page with genuine local data, original photography, and first hand experience is not that. A cloned template with a swapped city name is, and a human writing it does not exempt it.

How many words should a service area page be?

There is no threshold and the circulating numbers are invented. Fill the modules above honestly and you will land between roughly 900 and 2,000 words. If you can only produce 300 words of substance about a place, that is a signal not to publish rather than an instruction to pad.

Should I use AI to write my location pages?

You can use it to draft and edit, but not to originate. Google’s policy language, “no matter how it’s created,” means the question is originality and purpose, not authorship. AI cannot know your transaction history, has never driven the neighborhood, and cannot take the photographs.

Do I still need FAQ schema?

Not for a rich result. Google discontinued FAQ rich results on May 7, 2026 and removed the supporting documentation the following month. The markup stays valid and machine readable, so it does no harm, but anyone promising FAQ rich results in 2026 is working from outdated information.

Should my IDX listing pages be indexed?

Generally not for syndicated listings from the shared feed, which appear near identically on hundreds of competing sites and go stale the moment a property closes. Your own listings, with your own photography and description, should be indexed. NAR’s policy does not require you to block indexing, though MLS rules differ.

How do I fix pages stuck in “crawled, currently not indexed”?

Not by resubmitting, which Google explicitly says is unnecessary. Compare the page against your pages that did index, find what they contain that it lacks, and add that substantively. Then link to it from an indexed page with traffic. Still nothing after two months, merge and redirect it.

Do Core Web Vitals actually affect rankings?

Modestly. Google states there is no single page experience signal and that Search “always seeks to show the most relevant content, even if the page experience is sub-par.” Speed matters far more for conversion than for position, which is still reason enough to fix it.

Will AI Overviews destroy my search traffic?

They will reduce clicks on informational content and barely touch your local commercial pages. Ahrefs measured AI Overviews on 21.4 percent of informational queries against 2.1 percent of transactional ones. Your blog will feel it. Your “realtor in [town]” pages largely will not.

Is a blog still worth writing?

Yes, but judge it differently. Individual posts earn fewer clicks per impression than they did three years ago. Their value now is topical coverage, internal link equity flowing to your money pages, and being retrievable by AI systems that cite sources.

How much should I pay someone to do this?

Pricing varies too widely for a useful benchmark, but the diagnostic questions do not. Ask what page count they intend to build and why that number. Ask how each area page will differ beyond the place name. An agency that cannot answer the second is selling you a template.

Does my site need to be on a specific platform?

No. Every major platform can be made to rank and every one can be configured badly. What matters is whether you control title tags, meta descriptions, canonical tags, heading structure, image attributes, and redirects, and whether mobile renders what desktop does.

What is the single biggest mistake agents make?

Building too many pages too fast. It comes from good intentions and a platform feature that makes it easy, and it produces a site where most pages never index and the ones that do are held back by the company they keep.

Educational content only, not legal, tax, or financial advice. Search engine behaviour changes without notice and no ranking outcome is guaranteed by any technique described here. Google policy language is quoted from Search Central and Search Console documentation as published at the time of writing, and policy pages are revised periodically. Search volume and cost per click figures are third party estimates. Study figures are attributed inline, and where two credible studies disagree both are reported. MLS and IDX rules vary by market. Verify your own before changing indexing settings.