SEO for real estate: what the portals own and what an agent can win

Search "homes for sale Denver" and count how far down the first agent appears. On 18 August 2026 the answer was position ten: Trulia and Zillow held the top of the results, Homes.com and Redfin followed, national franchise sites filled the middle, and the only domain belonging to an agent team closed the page. The same day, "sell my house Denver" returned a different market entirely. Not one listing portal appeared: two national cash-offer platforms led, and five we-buy-houses operators filled the rest, several of them single-market Denver sites. Most real estate SEO budgets are spent on the first page, which is closed, and almost none on the second, which is not.

Quick answer

• The inventory searches are gone. Every variant of "homes for sale" belongs to Zillow, Trulia, Redfin and Realtor.com, which publish every listing in the market, including yours.
• The seller side is open. Local seller queries return cash-offer operators with template pages, and local domains already rank there, which is the proof the ground can be held.
• The neighborhood page is the defensible unit. The portal has every listing and no knowledge: it cannot say which blocks carry the arterial noise or where the school boundary actually runs.
• The agent is the entity being searched. In NAR's 2025 profile, 88 percent of buyers purchased through an agent, and the search that follows a referral is a name, not a keyword.

The model that burns the budget

The plan most agents are sold treats Zillow as the competitor to outrank, and the site it produces is a small inventory portal: an IDX search box, listing pages generated from the MLS feed, and "homes for sale" in every title. The structural problem is that the portal publishes the same inventory, plus everyone else's, through the same kind of feed. Competing with it on listings means competing against a database that already contains everything the agent has to show.

The portal is not a competitor. It is a distribution channel that took the results page, the same pattern that governs SEO for therapists, where the directories own the generic queries and the specific ones stay open. The difference in real estate is where the open half sits: not in specificity but on the other side of the transaction. A buyer search asks to browse inventory, which is exactly what an aggregator does better than any single agent. A seller search asks what a house is worth and who to trust with it, and no national database answers either question about one specific house on one specific street.

The first page for this query does not say any of this. The guides I could open in full teach keyword research, backlinks and profile optimization as if every query were equally available, name the portals in passing or not at all, and never turn the buyer-seller split into strategy, which is how budgets end up on the closed half.

Does SEO for real estate work for an agent?

SEO for real estate works when it concedes the inventory searches to the portals and targets the seller side: valuation, neighborhood and name searches. A solo agent can win the handful of neighborhoods they farm; a team adds a page per area and per agent; a brokerage competes on district-level market content the portals cannot write.

The demand is real and measured. In my own Ahrefs study of 17 August 2026, "seo for real estate" records 1,600 monthly searches in the United States with a keyword difficulty of 10 and a cost per click of 7 USD, and "seo for real estate agents" adds 1,300 more. Two things follow. The low difficulty says the pages ranking today are not heavily defended, which matches what they leave out. And the 7 USD says what marketers pay to reach an agent asking this question, which is a useful measure of how much advice this sector gets sold.

Two sides of one transaction, two different results pages

The buyer side rewards aggregation. NAR's 2025 profile reports that what buyers wanted most from their agent was help finding the right home, at 50 percent, and finding the home is precisely the job the portals industrialized: every listing, every filter, refreshed continuously. The Denver page observed above is the consequence, portals first, franchises second, one agent domain in tenth place. Holding a position there means out-aggregating an aggregator, and no budget an agent controls gets there.

The seller side rewards trust, and its results page shows it. "Sell my house Denver" returned cash-offer operators, several of them single-market sites, with pages built on the same three promises: fast, as-is, no fees. Thin pages hold that ground because almost nobody contests it. An agent's page that does the one thing those pages will not, explaining what the speed costs, what a cash offer gives up against a listed sale and when taking one is still rational, competes against template text, not against Zillow.

The valuation query splits the same way. Nationally, "how much is my home worth" belongs to automated estimate tools: Zillow, Redfin, two major banks and iBuyer platforms held the first results on 18 August 2026, so that form is closed. The local form is not, because an automated estimate is exactly as good as its data, and the data does not know what was renovated, what backs onto the freight line, or which side of the street floods. A valuation page for one market, written by someone who prices houses there every week, answers the question the tool approximates.

The two sides are different markets. Ninety-one percent of sellers sold with an agent in NAR's 2025 profile, up from 90 percent the year before, and for-sale-by-owner fell to five percent, the lowest recorded. The seller demand did not migrate to the portals, and the seller search did not either: it sits in queries the portals answer generically or not at all.

The neighborhood is the unit, because knowledge does not syndicate

Listing data flows through the feed to everyone. What does not flow is the reason a buyer picks one street over the next, and NAR's 2025 profile says that reason now dominates the decision: 59 percent of buyers ranked neighborhood quality as a top factor in choosing where to live, 47 percent convenience to friends and family, while convenience to the buyer's job has fallen from 52 percent in 2014 to 31 percent. The unit of the housing decision moved from the commute to the neighborhood itself. The portal's neighborhood page, meanwhile, is a template: a median price, a school list pulled from a database, a map.

A real neighborhood page carries what the template cannot. Where the school catchment boundary actually runs, against where listings claim it runs. What the same floor plan sold for on each side of that boundary this spring, sourced to the local MLS by name. Days on market for the neighborhood against the metro, with the month the numbers were pulled. Which blocks absorb the arterial noise and which are shielded from it. None of it syndicates. Every line is checkable, and a page holding ten lines like that eventually outranks a template because nothing else on the query answers them.

The test for these pages is the one that governs contractor city pages, laid out in SEO for contractors: if the neighborhood name could be swapped and the text stay true, the page should not exist. The sustainable scale follows from structure. A solo agent maintains real pages for the three to five neighborhoods they actually farm; a team assigns areas to agents and maintains one real page per pairing; a brokerage can fund the district-level market reports no portal localizes. What no tier sustains is fifty pages generated from listing data, which is the portal's own game played without the portal's database.

The agent is the entity, and the site is the evidence

Eighty-eight percent of buyers purchased through an agent in NAR's 2025 profile, and five percent of sellers went without one. The industry reads those numbers as job security. The search reading is different: this business runs on referrals, and the client who arrives by referral does not type "realtor near me". The referral ends in a search box. They type the name they were given, and what comes back decides whether the recommendation survives contact with Google.

What the name search returns is territory the agent can control, starting with a complete Google Business Profile under the agent's own name rather than only the office's, because the license and the reputation belong to the person and both travel between brokerages. An agent who changes firms and left the entity in the brokerage's hands starts over; one who built it under their own name carries every review and every citation to the new desk, which is the compounding decision in this sector.

Reviews are the second half of that territory, and here the wording matters more than the count. "Sold our house in Park Hill in three weeks, two offers over asking" carries a neighborhood, a timeline and a transaction type, which is exactly what the next seller scans for; "great service" carries nothing. The mechanics of asking at the right moment are in how to get Google reviews, and in this sector the right moment has a date: closing day, when the result is concrete and the relief is real. A practice that asks then, every time, accumulates the one kind of proof this market cannot fake at scale.

The generic fraction of demand still resolves in the map. "Realtor [city]" and its variants produce a local pack, ordered by the factors covered in how to rank in Google Maps, and the agent who arrives with a complete profile and specific reviews competes there without doing anything additional, which is the sense in which the entity work pays twice. The pack rewards what the referral already required: a name that resolves cleanly, reviews that describe transactions, a category that matches the license. The rest of the local layer, citations, hours, consistency across directories, follows the general playbook in local SEO for small business, with one adjustment: the entity worth building is the agent, not the office, because clients hire and refer a person, not an address.

When NOT to invest in this

When the entire pipeline is sphere and repeat clients, and the agent wants it that way. A practice that closes from a warm list has outsourced discovery to its own clients, and the only piece worth securing is the name search, which is an afternoon of setup, not a retainer.

When the horizon is this quarter's closings. This channel compounds over quarters, and the comparison with paid channels, which can switch on the same Monday, is laid out in SEO vs Google Ads. An agent short on this month's pipeline is buying speed, and speed is the one thing this channel does not sell.

When there is no geographic focus yet. The defensible content is neighborhood knowledge, and an agent covering an entire metro without a farm area has none to write down. Choosing the ground comes first, and that is a business decision before it is a marketing one.

Mistakes that repeat

• Spending the budget on "homes for sale [city]". The observed page is portals plus franchises; the spend buys positions behind them, and those positions produce nothing.
• Building the site as a small inventory portal. Pages generated from the IDX feed duplicate what the portals publish with more authority, so the site inherits a comparison it cannot win.
• Generating a page per neighborhood from listing data. If the name swaps and the text stays true, the page fails the same test that kills contractor city pages, and it ranks accordingly.
• Leaving the name search to chance. The referral checks the name; a thin or absent first page turns a warm recommendation back into a cold comparison shop.
• Reviews that could describe any business. A review without a neighborhood, a timeline or a transaction type carries none of the information the next seller is scanning for.
• Judging the channel on buyer leads. The open half of this market produces listing appointments and name-search conversions, and a report counting only buyer leads cancels the part that was working.

Data and transparency

The three results pages described, "homes for sale Denver", "sell my house Denver" and "how much is my home worth", were observed on 18 August 2026 and reported as returned that day; they are observations, repeatable in a minute, not a study. The 88 percent of buyers who purchased through an agent, the 91 percent of sellers who sold with one, the five percent for-sale-by-owner share, the 50 percent who most wanted help finding the right home, and the neighborhood factors, 59 percent neighborhood quality, 47 percent convenience to friends and family, 31 percent convenience to work against 52 percent in 2014, all come from the National Association of Realtors' 2025 Profile of Home Buyers and Sellers, published November 2025 and read on 18 August 2026 in its highlights document and accompanying analysis. The demand figures come from my own Ahrefs study of 17 August 2026: "seo for real estate" shows 1,600 monthly US searches, keyword difficulty 10 and a 7 USD cost per click, and "seo for real estate agents" shows 1,300. The pricing reference in the FAQ comes from SE Ranking's survey published 13 December 2024, 260 agencies, whose authors warn the sample may not reach statistical significance.

No figure appears here for the share of closings that search produces, for agent website conversion rates, or for how far automated valuations miss: versions of all three circulate and none I checked traced to a primary source. The strategic ordering is operating judgment from audit work across a portfolio recording more than 300 million impressions a year in Search Console. Verified as of August 2026.

Primary sources, opened on 18 August 2026: NAR's 2025 Profile highlights (PDF); NAR's takeaways analysis; SE Ranking's pricing survey.

What this changes

The portals did not take the agent's market. Ninety-one percent of sellers still listed with an agent in NAR's 2025 profile, and for-sale-by-owner sits at its recorded low of five percent. What the portals took is one species of search, the browsing kind, and it was never the demand that paid an agent: a buyer scrolling inventory is not choosing representation, and the one who is ready to choose asks a different question or types a name. The expensive mistake in this sector is grieving the closed half of the results page and funding the grief, year after year, with the budget that the open half was waiting for: the valuation page, the neighborhood no portal can describe, the name search after a referral. The portal kept the searches that never chose an agent. The agent keeps the ones that do.

Frequently asked questions