Industry: Real estate brokerages

Find out which brokerage offices and agent sites earn your search traffic

Your brand site, your office sites and dozens of agent team sites all chase the same buyer and seller searches. Each one reports to a separate GA4 property and Search Console site, often under someone else's Google login. Connect them once and ask your AI agent which one is actually winning.

Continue with Google

Free for 2 sites · No credit card

Or have your agent set it up

Set up DuckDive for my website. Follow https://analytics.duckdive.ai/agents.md

Updated 16 min read

The short answer

Add your brand site, each office site and each agent team site to DuckDive as separate websites, and attach whatever analytics, search and Local Services Ads data each one runs on. Your AI agent can then rank them on search clicks, valuation requests and leads, and tell you which layer of the brokerage is earning the demand.

  • Judge home valuation pages on conversions and listing pages on search reach. They do different jobs.
  • Keep careers and join-our-team pages out of your lead totals, or they quietly inflate them.
  • Office pages in a folder of the brand domain count as one website. Separate domains or properties split cleanly.

Brand site, office sites or agent sites: who earns the traffic

Most brokerages past a handful of offices end up with three layers of web presence nobody planned: the brand domain with the IDX search, office pages or microsites, and agent and team sites built on whatever platform each agent liked, often on domains the agent registered.

That third layer is bigger than most marketing directors assume. In NAR's 2026 Member Profile, 73% of REALTORS® kept a business website, 82% of those sites featured the agent's own listings, and only 62% linked to their firm's site, as HousingWire reported in June 2026. So many of your agents' sites show your listings without linking back to the firm.

So the real question is which layer should win which search, and whether it does. Adjust this split to fit how your brokerage pays for and owns each site.

Which layer of a brokerage should win which search
LayerSearches it should winJudge it on
Brand siteBrokerage name, broad metro listing searches, careersBranded clicks, listing sessions, recruiting inquiries
Office site or office pagesTown and neighborhood listing and selling searchesLocal query clicks, valuation conversions
Agent or team siteThe agent's name, their farm area, their listingsName searches, contact form conversions, LSA leads
Shared riskTwo layers ranking for the same town queryWhich one gets the click, and who owns that domain

In DuckDive, every brokerage site you add becomes a separate website record, carrying the analytics property, the search site and any ad accounts that sit behind it. How your layers are built decides how cleanly they split.

  • Separate domains. Each office or team domain with its own GA4 property stands alone. Clean comparisons.
  • Office folders on the brand domain. One website. Your agent compares offices by grouping landing pages by path, such as everything under /offices/southbank/.
  • Office subdomains. Watch Search Console. Google's help says a Domain property "includes all subdomains", so a brand Domain property already counts every office subdomain's clicks. Link URL-prefix properties to office websites, or you count those clicks twice when you add up the brokerage.
  • One GA4 tag across many domains. If brand and agent sites share one tag through GA4 cross-domain measurement, which needs the same tag ID on every domain, they are one property. You can split them by landing page, not by site.
Ask your agent
Which agent team sites got more search clicks than their own office site last month?

Tools it calls list_websites get_search_console_summary

Seller-lead pages versus listing pages

A brokerage site carries two kinds of pages that look alike in a GA4 report and do opposite jobs. Listing pages, the IDX search results and property detail pages, pull large volumes of buyer sessions and churn as homes sell. Seller-lead pages, such as "what is my home worth" and valuation request forms, get a fraction of the traffic and are where listing inventory starts.

Sellers matter because they are scarce. NAR's 2026 Profile of Home Buyers and Sellers found sellers had owned their homes for a median of 12 years, and 92% sold with an agent or broker. A homeowner who fills in your valuation form may not be back in the market for over a decade. Most also arrive with someone in mind: Real Estate News reported that 63% of sellers in the same survey used an agent they had worked with before or were referred to. Your valuation page competes for the rest.

So judge the two groups on different numbers:

Sessions (GA4)
Visits that start on a page. The volume measure for listing pages.
Conversions (key events in GA4)
A valuation request only shows up here if someone has flagged the form-submit event as a key event inside that office's own GA4 property. Check every office's property, not just the brand's; Google's key events help says the marking happens on the Key events page in Admin. DuckDive's tools call the result conversions.
Conversion rate
Valuation requests as a share of the visits that landed on that valuation page, both counted over one window. This is the number that separates a busy page from a productive one.
Clicks and average position (Search Console)
Clicks from Google results to the site, and where it typically showed. They reveal which layer is ranking.

Valuation pages usually fail for dull reasons: the form sits on a vendor's domain and the key event never fires, or each office named its page differently. If one ranks well and still sends nothing, the walkthrough on ranking pages that send no leads is the next stop.

Ask your agent
On each office site, split the last 90 days of landing page sessions into listing pages and seller pages (valuation, sell, what's my home worth), then compare their conversion rates side by side.

Tools it calls list_websites get_ga4_landing_pages

Five offices, one valuation form that stopped counting

Settle this before next quarter's seller campaign gets its budget, because a broken form and a weak page look identical in a total. The brokerage and its numbers are invented.

Ask your agent
Rank our home valuation landing pages across all office sites by GA4 conversions over the last 90 days.

Tools it calls list_websites get_ga4_landing_pages

EXAMPLE DATAFictional business: Clearmont Realty Group

Rank our home valuation landing pages across all office sites by GA4 conversions over the last 90 days.

  1. list_websitespicks out the five Clearmont office sites by name
  2. get_ga4_landing_pageseach office's entry pages over 90 days, filtered down to the valuation pages

Southbank leads with 56 valuation requests. Beckford Rise is the one to look at: it drew the second most visits to its valuation page and converted the lowest share of them (1.1%).

OfficeSessionsConversionsRate
Southbank1,240564.5%
Linden Quay880374.2%
Tamsworth Green420194.5%
Beckford Rise1,050121.1%
Fairhollow61091.5%

All five together: 4,200 sessions and 133 conversions, or 3.2%. I matched pages by path, and the offices use three different ones: /home-value at Southbank, Linden Quay and Beckford Rise, /whats-my-home-worth at Fairhollow, and /sell/valuation at Tamsworth Green. Tell me if I missed one.

Beckford Rise has the traffic and not the requests. Check that its form still records a key event in GA4 before blaming the page itself. Fairhollow has both problems, so it is the better place to test a new seller offer.

Google Analytics 4

Always ask which paths were matched, and compare rates rather than raw counts: Southbank's 56 partly reflects its traffic. If office properties count different key events, line them up first using the steps for lining up key events across separate GA4 properties.

Rank your own offices' valuation pages

Link one office site free, then ask which seller pages pull requests and which only pull visits.

Continue with Google

Free for 2 sites · No credit card

Local Services Ads for real estate agents

Local Services Ads are pay-per-lead ads above the regular results, and real estate has its own rules there:

  • Every agent is checked. Google's US screening requirements list "license checks for each real estate agent in the firm", and for some firms professional liability insurance too. Add new agents to the account early, because each one needs a license check.
  • Listing ads ride on the same account. Google's Home Listings Ads show for-sale homes on Search and let buyers call or message a buyer's agent. They need a verified Business Profile, a linked Local Services Ads campaign, and the Buyer's agent or Seller's agent job types. Google says they are "available in United States markets on mobile" and do not serve on Maps.

If your offices also run Google Ads search campaigns, Google's housing policy rules out ZIP code and demographic targeting in the US and Canada. Its restricted targeting page says ZIP code location targeting "cannot be used for Housing" ads there, and age, gender, marital status and parental status are off the table as well. That applies to your Google Ads campaigns, separate from the pay-per-lead units above.

Some brokerages hold one account per office; others let strong teams run their own. Either way, link each account to the site it sends people to, so its leads sit beside that site's search and GA4 numbers.

Per account, get_lsa_summary hands back three figures (lead count, charged-lead count, spend), and your agent works out cost per lead from those. It does not split buyer's agent leads from seller's agent leads, and it does not show call recordings or where an agent placed in the ad. The Local Services Ads source page covers the math and its traps, and if Google moves your agents' accounts into Google Ads, read how to keep LSA history through the move first. Multi-office law firms run into the same one-account-per-office pattern; the law firm page shows how it plays out with practice areas instead of towns.

Ask your agent
Pull the past 90 days of Local Services Ads numbers for every office and team site with an account linked. Work out our cost per charged lead and flag any team paying more than its own office.

Tools it calls list_websites get_lsa_summary

Recruiting pages, the other audience on a brokerage site

A brokerage also markets to agents deciding where to hang their license. In the same Member Profile, HousingWire reported a median of six years with the current firm and 53% of members at independent brokerages. Agents move, and they research firms first.

Recruiting pages cause two reporting problems.

  1. They inflate lead totals. If "join our team" form submits are a key event, they sit in the same conversions count as valuation requests. A good recruiting month then looks like a strong seller month.
  2. They pull branded traffic. Agents who search your brokerage's name plus "careers" or "commission" show up as branded clicks, which makes consumer brand demand look higher than it is.

DuckDive's GA4 tools report conversions per landing page, not per event name, so the clean split is by page. Keep recruiting content under its own path, such as /careers/ or /join/, and have your agent set those rows aside before it totals anything for the seller side.

Ask your agent
On our brand site, how many sessions and conversions did the careers and join-our-team landing pages get over the last 6 months, and what share of the site's total conversions is that?

Tools it calls search get_ga4_landing_pages get_ga4_overview

Recruiting page check, once a quarter

  • Own path. Every recruiting page sits under one folder, so your agent can set it apart by path.
  • Own form. The recruiting form is a different event from the valuation form, even if both are key events.
  • Search terms. Ask for brand-site queries that contain "careers", "join", "commission" or "split" to see what agents look for.
  • Exit check. After an agent leaves, see whether their old team pages still rank, and where they now point.

When Elm Row Team outranks Fairhollow

Strong teams build good sites, and sometimes they outrank their own office for its town. Invented numbers again.

EXAMPLE DATAFictional business: Elm Row Team at Clearmont's Fairhollow office

Which Fairhollow team sites got more search clicks than the Fairhollow office site in the last 30 days, and on which searches?

  1. list_websitesfinds the Fairhollow office site and the three team sites named after it
  2. get_search_console_summary30-day totals for the office and its three team sites
  3. get_search_console_keywordsthe leading search queries on both the office site and Elm Row's, to see where they collide
  4. get_search_console_pagesElm Row Team's top pages in Google results

One team did. Elm Row Team got 2,010 clicks against the office site's 1,380, a gap of 630.

SiteClicksImpressionsAvg position
Elm Row Team2,01048,9009.8
Fairhollow office1,38061,20014.2
Rowanmere Group1,12039,80012.6
Copperline Team64022,70017.5

The office site is seen more often (61,200 impressions) but clicked less. Both sites have "fairhollow homes for sale" and "fairhollow neighborhoods" in their top queries, and Elm Row gets more clicks on both. Its top pages in search are its neighborhood guides. The office's own top queries are mostly the brokerage name.

Search Console

Not a problem by itself: whoever finds Elm Row still works with a Clearmont agent. It turns into one in three cases:

  • The team owns the domain. If Elm Row moves firms, its search traffic goes with it. Agree in writing who owns team domains, and get legal advice on the wording.
  • The team site never links back. Ask for a clear link to the office page. It costs the team nothing.
  • The office has nothing to rank with. If the office site is a contact page and a listing feed, it will lose town searches to anyone who writes about the town. Neighborhood content belongs on the office site too.

The narrower check below asks whether an office ranks for the buyer searches it should own. Only the 100 highest-click queries come back, so a query nobody clicked in the window stays invisible. Every query row carries its own average position, so you can see whether Southbank sits on page one for its town or just below it.

Ask your agent
For the Southbank office site, show Search Console queries containing 'homes for sale' and their average position over the last 30 days.

Tools it calls search get_search_console_keywords

Windows, not months

There is no calendar-month option: ask for October's listing traffic and you get the last 30 days. Rolling windows run from a week to a year. When the board wants a figure for one exact month, pull it from Search Console's own date picker instead.

Keeping agent sites from swamping your website list

Fifteen offices and sixty team sites make seventy-six websites with the brand. list_websites returns up to 100 per call and filters on name or URL text, so consistent names do the sorting.

  • Lead with the office. Name each site "Fairhollow office" or "Fairhollow, Elm Row Team". Then "every Fairhollow site" is a single filter.
  • Mark the layer. Use "office", "team" or "agent" in every name, so your agent can compare like with like.
  • Add only sites you can read. An agent site on a platform that will not give you GA4 or Search Console access adds a row with nothing behind it.
Ask your agent
Find every website with Fairhollow in its name and flag any that is still missing a Search Console site or a GA4 property.

Tools it calls list_websites

Agent sites, offices and access

At a brokerage, setup is mostly an access problem. Solve that and the rest is quick.

  1. Map every site to its owner

    List the brand domain, each office site or folder and each team site, who controls each one's GA4 and Search Console, and any GA4 tag shared across domains.

  2. Get your Google login added

    For sites you do not control, the agent or team lead adds your Google login as a user in GA4 and Search Console, and in Local Services Ads if they run it. Nobody shares a password.

  3. Sign in and add the websites

    Sign up with the Google account that now holds all that access. DuckDive shows every analytics property, search site and ad account it can reach. Create one website per office or team site, attach the matching property and search site, and attach ad accounts only where that office or team actually buys ads. Each link starts an import of the site's history; expect a few minutes, longer for an IDX site with years of pages.

  4. Or delegate the linking to your agent

    Your agent can create the office and team websites and link the obvious matches on its own, provided it uses an admin login and the "Also allow actions" box stayed ticked during the connection. If a property already belongs to some other website, it asks you first.

    Give your agent this line
    Set up DuckDive for my website. Follow https://analytics.duckdive.ai/agents.md
  5. Connect your AI app with the right login

    Sign your AI app into DuckDive as the marketing director, not as an office manager, so it sees every office and team site. Setup differs by app: see connecting Claude or connecting ChatGPT.

  6. Confirm the imports, then ask

    Ask your agent whether every office and team site has finished importing. get_sync_status flags any source still loading or failing. Run the valuation ranking first: it tests GA4, page naming and key events at once.

Hiring an outside consultant to audit the agent sites? They can read the same connected data; the first-week audit for consultants shows their usual method.

What this will not tell a broker

A brokerage's most important numbers sit outside Google, so know where the edge is.

  • No CRM, MLS or transaction data. It cannot say which valuation request became a listing, or which office closed the most sides.
  • No Google Business Profile data. Office profile calls and direction requests are not a source.
  • No lead detail from Local Services Ads. Counts and spend only, with no buyer and seller split.
  • Top 100 rows per call. A brand site with thousands of listing pages will show its top 100 landing pages, not all of them.
  • Twelve months through an agent. Fixed windows, with the last 12 months as the longest.

Something else fits better if everything lives on one domain and one GA4 property and page reports are all you need: GA4's explorations will do. Branded monthly PDFs per agent are also a job for a dedicated report builder. DuckDive pays off once you have many separate sites to question together. The multi-location overview covers brands outside real estate.

Questions brokerage marketers ask

Our office pages live in folders on the main domain. Can we still compare offices?

Yes, at page level. DuckDive treats the whole brand domain as a single website, so your agent splits offices by the folder in each page's path. Calls cap out at 100 rows, and thousands of listing pages can crowd a small office out of that top 100. Separate office domains or properties give a cleaner split.

Agents own their own domains. Do we need their Google passwords?

No. Each agent invites your Google login as a user in their own analytics and Search Console settings, just as they would invite any marketer. DuckDive never asks for a password, and what your AI app holds is a DuckDive token, never anyone's Google password.

Can DuckDive tell us which valuation requests turned into listings?

No. It reads Google data only, so the trail ends at the GA4 conversion. Which requests became listing appointments, signed listings or closed sides lives in your CRM or transaction system, and DuckDive has no connection to either.

Does it read leads from Local Services Ads for real estate?

Yes, as totals: lead count, charged-lead count and spend for every linked account, with one account per website. It does not split buyer leads from seller leads, show call recordings or show where an agent ranks in the ad unit. Those stay in the Local Services Ads screens.

Can individual agents see the whole brokerage's numbers?

An assistant connected to DuckDive sees only the websites that the signed-in DuckDive account can see, with that account's role. Admins can set things up; team members can only read. Decide who holds admin rights before you invite anyone.

How far back can we compare this year's spring market with last year's?

Not through an agent today. The widest range it can request reaches twelve months back, which holds one spring. DuckDive stores up to 2 years, but for last spring against this spring use GA4's or Search Console's own date comparison.

Sources

  1. Experienced Realtors hold the line in a tough housing market, HousingWire (on NAR's 2026 Member Profile), checked November 16, 2026.
  2. Add a website or platform property to Search Console, Google Search Console Help (Domain properties and subdomains), checked November 16, 2026.
  3. Set up cross-domain measurement, Analytics Help, checked November 16, 2026.
  4. Top 10 Takeaways from NAR's 2026 Profile of Home Buyers and Sellers, National Association of REALTORS®, checked November 16, 2026.
  5. Share of buyers, sellers using agents hits record high, Real Estate News, checked November 16, 2026.
  6. Conversions vs. key events in Google Analytics, Google Analytics Help (key event definition), checked November 16, 2026.
  7. Business screening and verification requirements, United States, Local Services Help (real estate agents and brokers), checked November 16, 2026.
  8. About Local Services Home Listings Ads, Local Services Help, checked November 16, 2026.
  9. Restricted targeting in Personalized advertising, Google Ads Policy Help, checked November 16, 2026.

Last reviewed . Numbers in EXAMPLE DATA blocks are fictional.

See which office really wins your town

Connect an office site, then ask whether its team sites, its valuation page or its ads are doing the work.

Free for 2 sites · No credit card · Working in an agent? Let it set DuckDive up