In January 2025, I looked through 793 real estate marketplace websites from around the world to see what the standard is for property search, if indeed there is a standard experience. This report seeks to identify which features come as standard on these websites, and which remain nascent… (View Highlight)
30% of portals still don’t offer a map view of listings. Only 40% let users scroll listing photos directly from search results.
Less than 5% support commute time search. (View Highlight)
AI is Ubiquitous on Press Releases but Rare on Homepages While AI is frequently mentioned in PR, only 7 portals globally offer true AI-powered home search.
Business model friction, not tech barriers, is the main blocker.
Most “AI” implementations are chatbots or back-end functionalities, not search enhancements. (View Highlight)
AVMs Are Evolving into Data Capture Tools 60% of valuation tools now sit behind registration walls (up from 36% in 2021).
Portals increasingly use AVMs as a first touchpoint for long-term user profiling and seller lead nurturing. (View Highlight)
Leading Portals Restrict Search to Control Results Pages On average market-leading portals allow their users 6.5 fewer search parameters than their competitors, despite having 2.5 more fields of data available per listing. (View Highlight)
There has not been any great leap forward in the uptake of any UX tool on real estate marketplaces since 2023.
Any noticeable differences between the two datasets likely reflect a slight change in the sample size rather than any significant uptick in adoption. (View Highlight)
Portals need to be incremental in their approach to AI and make a clear distinction between AI for search and AI for enrichment. (View Highlight)
There is currently a divide between portals looking into new search technologies. There is a cultural hesitance among some portals to disrupt traditional search methods.
Many real estate portals started as classified directories and have been slow to adopt new technologies, partly due to their origins and existing business models. (View Highlight)
From 2021 to 2023 we saw a marked increase in the percentage of these sites that showed their listings on the map, but that trend seems to have hit its ceiling.
There are still many markets where agents don’t get exclusive mandates—where there is a risk of competing agents poaching listings if they know a property’s specific address. (View Highlight)
Perhaps the continued existence and commercial viability of real estate marketplaces without a map in mature markets speaks to the endurance of the basic classifieds model and the habits of unsophisticated users.
For example, the Adevinta-owned Dutch horizontal Marktplaats still has a real estate section despite the paucity of listings (308 for sale at the time of writing) and no ability to map them. (View Highlight)
Scroll Photos from Results Page The fewer tabs you have to open the better. That’s why we are big fans of real estate websites that let users look through a listing’s photos from the results page without having to open the listing page itself.
Unfortunately, this undeniably useful feature remains far from ubiquitous on real estate marketplaces, with only 40% of the websites we studied letting users scroll photos from the results page. (View Highlight)
It’s true that any portal wanting to implement this functionality does have to load more images on its all-important results pages but there is another reason that marketplace websites may not want to help the user here…
Making a user click through means more listing page views if not more leads. Many portals evidently still use the number of pageviews of an agent’s listings to justify their fees. (View Highlight)
Commute Time Search Letting users search for a new home by the time it takes them to get to a specified point of interest (workplace, school or family) is a feature we think all portals should have. (View Highlight)
Commute Time Search Letting users search for a new home by the time it takes them to get to a specified point of interest (workplace, school or family) is a feature we think all portals should have.
Unfortunately less than 5% of the websites we looked at have it. On the bright side, at least the figure is up from the 3.5% we observed in 2023. (View Highlight)
Unsurprisingly, portals that allow users to search by commute time are mostly located in developed markets like the United States, France and the UK.
There is a lot of data to crunch in the background when calculating the bounding boxes for a commute time search function. However, there is evidence that the uplift in conversions is worth it. (View Highlight)
Green-Acres is a France-based portal specialising in the second home market across Europe.
In 2023 the company launched a rebranding and product evolution to position itself as the portal for people looking to change their lives.
The user interviews conducted as part of this process revealed something interesting, according to Green-Acres Founder and CEO, Benoit Galy.
“Our users don’t start with a postcode. They start with aspirations — a lifestyle, a sport, a specific environment. So we built a search engine that allows them to find their future home accordingly.” (View Highlight)
After reviewing the interviews, Galy and his team set out to build their own lifestyle-based real estate search.
Green-Acres launched the feature in 2025. It asks the user a series of questions around the type of environment they’d like to live in and the points of interest or amenities they’d like to live near.
Although the feature itself is location agnostic, it does include a commute time feature linked to the ‘practical’, ‘leisure’ and environment’ variables users specify. So, for example, users can search for homes within 30 minutes drive of a tennis court.
Green-Acres’ lifestyle search sits alongside a traditional location-first search map on the portal’s homepage. (View Highlight)
Since 2021 we have been tracking every announcement made by a real estate portal company that mentions AI.
It’s fair to say that the companies behind real estate marketplaces are keen to let the market know when they have been experimenting with AI in their products.
While using AI to enhance listings is common, the most common category these portal AI announcements fall under is Discovery. (View Highlight)
Much has been said about the potential of artificial intelligence-powered real estate search at industry conferences over the last few years. We’re told that thanks to this technology, users will be able to find their dream home much faster Despite the hype, however, it seems true industry-wide adoption of AI-powered search is a long way off. Many of the announcements we’ve covered have been chatbots, virtual assisstants and matching systems rather than true AI-search. (View Highlight)
What doesn’t count as AI-powered real estate search? There are several portals out there that claim to have some kind of user facing AI on their platform, but it isn’t always search specific and it isn’t always available for everyone. Below are a few examples that we did not include as true AI- powered real estate search in the final tally… (View Highlight)
Flyhomes is a U.S. based tech-enabled brokerage and so- called ‘power buyer’ known for its cash-offer service for buyers and integrated mortgage services.
Thanks to the idiosyncrasies of the U.S. housing market dynamics, it also operates a popular web portal used for discovery (in much the same way as brokerage firm Redfin).
Having acquired ZeroDown, a real estate startup backed by OpenAI founder Sam Altman, Flyhomes launched its AI- powered home search platform in 2024. (View Highlight)
What makes Flyhomes’ AI search different from others we found is that it does not automatically land its users on a search results page.
Users selecting the AI-search option on the homepage are directed to a chat interface with a range of automatically generated prompts and a free text box. Once they input their search, they are presented with matching listings directly in the chat and the option of viewing a results page.
The user can also ask follow-up questions such as whether a particular listing represents a good deal compared to others or whether mobile coverage is good in the area.
The experience mimics that of popular AI search interfaces such as ChatGPT and Perplexity and gives the user certainty that the listings retrieved in the chat are good matches.
Flyhomes’ business model means that, unlike most portals around the world, it can afford to eschew search results pages. Nevertheless taking a step towards eliminating results pages feels like a big leap forwards and, for now, remains unique in real estate search. (View Highlight)
WHY HAVEN’T PORTALS EMBRACED AI-POWERED SEARCH?
So why have portals not adopted this technology that has been around for several years?
Let’s start with the reasons portals might want to implement AI-search. These can be boiled down to two assumptions: 1. In theory, AI-powered search is better for users and allows them to find their dream home faster.
2. Over time users will come to expect AI-powered search on vertical search platforms like portals as their general internet search experience shifts.
The nuances of these two points are obviously up for debate but they are both generally accepted to be true by the industry experts we have consulted for this report and interviewed on the PPW Podcast. (View Highlight)
As for the reasons militating against the adoption of AI- powered search in the industry, they were the topic of the author’s presentation at the recent Bangkok edition of PropTech & Portal Watch.
Business model friction (or the ‘Machiavelli Effect’): Portals are holding back AI-powered search because more precise search means fewer impressions for premium listings. For portals reliant on upsells and ad visibility, this is a problem. (View Highlight)
It’s not a lack of innovation: Portals are shipping more updates than ever. R&D budgets are up. The industry is not in an ‘innovation lull’. (View Highlight)
The technology already exists: Voice and AI-powered search aren’t new. Even smaller portals have deployed them successfully.
Affordability is a huge constraint: Most users are filtering by price, because that’s what’s determining their options. A prompt box can’t help if there’s nothing in budget.
Adoption hasn’t arrived yet: Many users are still used to dropdown filters and don’t expect or ask for AI search.
Portals are hesitant to push something their audience might not understand or use. (View Highlight)
To get an outside perspective on the adoption of AI-search among portals we spoke to Mal McCallion. Mal was one of the founders of Prime Location and a former Sales Director at Zoopla. These days he runs a startup named ModelProp which helps agencies adopt AI (View Highlight)
In Mal’s view, the lack of adoption is “astonishing” with portals “wilfully restricting access to reasonably mature technology”.
Many portals already have the talent in-house to build AI- search functionality.
A lack of rich data associated with each listing coming into their systems is not an excuse.
Portal users will increasingly expect to be able to ‘vibe- search’ in real estate and portals may not have to educate them on new search functions.
Agents may adopt AI technologies faster than portals. (View Highlight)
For this year’s report, we wanted to dive deeper into real estate marketplace search and see just how many data fields portals have for each listing and whether they include them as search parameters. (View Highlight)
Some interesting differences between markets are reflected in the parameters that portals allow their users to search with.
Portals in the USA have the most search parameters, thanks to the rich MLS data associated with every listing.
House hunters in the UK can’t search by property size.
Both major portals in Singapore allow users to search by a property’s tenure.
In Germany users can filter for properties for which they wouldn’t have to pay a buyer’s commission fee.
French portals allow users to filter for properties sold on a life annuity basis.
Another interesting fact that the data turned up was that market leaders almost always allow their users fewer search parameters than number two or three players in the market. (View Highlight)
When it comes to the number of data fields shown for each property on individual listings pages, the U.S. portals again lead the way.
One interesting thing we noticed is that in some markets, portals can get away with displaying price cuts but not displaying a relative price (cheap or expensive compared to similar listings). Whereas in other markets, they can openly mark a listing as being relatively expensive but don’t get away with showing price history. (View Highlight)
You might be wondering why the market leaders allow their users to search with fewer parameters than challengers.
We initially thought that it could be because they have fewer fields of data associated to each listing. But our investigation reveals that’s definitely not the case.
On average market leaders have 6.5 fewer search parameters than number two players but have 2.5 more fields of data displayed on each property listing page. (View Highlight)
The fewer parameters a real estate marketplace allows its users, the more control it has over their search and, crucially, the listings it displays on the all-important results page.
For all their diversification and adjacent revenue streams, most real estate portals generate the bulk of their revenue from showing some listings above others on results pages. (View Highlight)
The more control a portal allows a user in searching for their next home, the fewer premium listings the user is likely to see and the greater the potential impact on the portal’s revenue.
However, there are ways that portals are getting around it.
One way is to demonstrate increased value as a transaction facilitator, software and services provider or data provider. This can mitigate any perceived drop in value agents feel if the premium listings they pay for start to yield fewer leads.
Another tactic that large portals have employed over the last few years is to move towards what former Property Finder CPO, Christophe de Rassenfosse called a ‘searchless experience’ (View Highlight)
“The game is really to build an engagement platform rather than a search platform. So there is a paradigm shift and we need to focus on how we engage with the user because the window [of looking for a property] is generally one to three months.” The terms ‘search’ and ‘discovery’ might often be used interchangably in the industry but they’re not the same thing.
Portals can still say they are adopting new ‘discovery’ technology while maintaining control over their results pages.
Applying machine learning to understand user behaviour and recommend properties rather than using AI to overhaul their traditional location-first search seems to be the approach being taken by the likes of Rightmove, REA Group, Property Finder and PropertyGuru. (View Highlight)
Automated Valuation Models (AVMs) have been around on real estate portals since Zillow launched its ‘Zestimate’ in 2006.
In 2021 we studied 50 leading real estate portals from aroud the world to see if they had an AVM, what information the user needed to put in, whether the estimate was behind a pay or email wall and their overall level of sophistication.
We repeated the study in 2025, widening our sample size to 69 portals, to see what’s changed. (View Highlight)
The first thing to note is that AVMs are becoming a must-have feature on real estate verticals in mature markets.
Even some portals with current or historic connections to agents associations (funda, OnTheMarket, Realtor.com) have tools that give consumers an idea of what their home is worth —a task traditionally closely guarded by agents. (View Highlight)
Also noteworthy are the hoops users have to jump through in 2025 to see their valuation compared to those from 2021.
While portals asking users to pay for valuations has all but disappeared—Property24 in South Africa and Realo in Belgium charge but market theirs as more comprehensive ‘Property reports’—more portals are putting their valuations behind a registration wall.
Only 36% of the portal AVMs we studied in 2021 required users to log in. In 2025 the figure had risen to 60% suggesting that AVMs are increasingly being recognised as strategic tools to generate future engagement and user data. (View Highlight)
While most portal AVMs nowadays require the user to make an account, only 19% of AVMs we looked at require the user to provide a phone number. (View Highlight)
Comparing data from 2025 and 2021 it seems clear that portal AVMs seem to be diverging down two distinct paths. (View Highlight)
One requires many fields of information from the user (often including a phone number) and often presents the AVM alongside an alternative option promoting a valuation direct from an agent.
A good example of this can be found on the valuations page of Italian portal Immobiliare.it where the agent option is unsubtly promoted over the portal’s own AVM.
The other type of portal AVM is a ‘logbook’ style model that already has information for all properties in the market and just needs the user to specify the address they’re interested in. (View Highlight)
Instead of serving a user up as a simple, unrefined seller lead to an agent as soon as they’ve generated a valuation, portals with trackable property pages get much richer information about a user through repeat visits to the page.
A user who comes back to the page increasingly often feeds a propensity model in the portal’s back end and is then categorized as someone who is much more likely to sell soon.
In many cases, the portals can gather unique data on properties alongside the data they are collecting on the user.
Editable fields on some portals’ property pages include the number of bedrooms, energy efficiency and whether the home has been renovated recently. (View Highlight)
A good example of the logbook style of product around a real estate portal AVM can be found in British number two player Zoopla’s My Home product.
Zoopla first built the tool in 2021 and is still the only UK portal to have this kind of product around its AVM. In a recent podcast interview, the company’s COO, Rich Hayes told us that it sees My Home as a “key competitive advantage”. (View Highlight)
My Home is a logbook-style product that lets users track the value of any home, on or off the market in the UK and, if they are the homeowner, edit the home’s details in Zoopla’s database.
The logbook for each home includes a confidence rating, gamified estimate of potentially competing listings at a given price, sales history, estimate history, equity estimate and a local market report. (View Highlight)
The tool has evolved from targeting homeowners in general to specifically being marketed to potential sellers.
Hayes told us that Zoopla utilizes propensity modeling to evolve the My Home experience based on both declared (user states they want to sell) and derived (behavior on the site suggests intent to sell) intent. The goal is a more personalized experience where Zoopla’s valuation follows the consumer across the website.
Together with some other site changes, the launch of My Home has led to a 225% increase in saved properties with four million homes tracked. Zoopla has also seen a substantial increase in valuation leads and a significant increase in view- to-lead conversions for sales and rental listings. (View Highlight)
To understand some more about the intricacies of AVMs on real estate portals we spoke to Hannah Parker who has worked in Product Management at Zoopla, Property Finder, Addland and the AVM provider, Hometrack. (View Highlight)
Hannah notes that despite the potential benefits of AVMs, their implementation on real estate portals remains limited.
She attributes this to challenges such as data quality and the complexity of accurately modelling property values.
Every AVM is inherently inaccurate due to market variability.
Launching a portal AVM via iterating with MVPs is a practical approach for introducing them.
Real estate agents often resist AVMs due to perceived threats to their roles. (View Highlight)
While those in charge of real estate marketplaces could rightly point to the fact that rarely has one portal prevailed over another purely because it has a better user experience, they would be missing the point.
At some point in the future real estate portals will not only be competing with other websites that look and work like theirs but also with some form of AI search. The form this new search will ultimately take and the time it will take to compete are up for debate but the competition when it comes will not be based on brand but on experience. (View Highlight)
The dichotomy between retaining control over search and providing a better experience for the user has no doubt been a source of friction for many years within portal companies.
I hope that this report throws into sharp relief how one-sided that argument has been until now and redresses the balance somewhat in favour of innovators.
Studying the industry and speaking to its more experienced participants about AI has turned me from a sceptic to a reluctant believer. (View Highlight)
