Close icon to exit the modal
Search icon
Close icon to exit the modal
Sign up to Publift Scoop
Subscribe to the Publift newsletter for all the latest news, updates and insights from across the industry.
Play video
Play icon
Play video

Layout Teardown: Episode 4 - Zillow.com's Unique Advertising Strategies

Layout Teardown: Episode 4 - Zillow.com's Unique Advertising Strategies

Next arrow icon
Back

Layout Teardown: Episode 4 - Zillow.com's Unique Advertising Strategies

Ben and Brock explore how Zillow optimises its ad strategy without affecting its core business, offering insights into both desktop and mobile layouts for improving ad visibility and user engagement.

‍

Key Takeaways

  • Programmatic is a rounding error on the balance sheet. Ads bring in just 2% of Zillow's total revenue. That shapes every decision. Nothing in the ad stack is worth touching if it puts the leads and listings business at risk.
  • The map creates a hidden waste problem. Zillow's desktop search is built around a map, and every time a user drags or zooms, the page reinitialises and fires off a fresh round of ad requests. Most of those go nowhere. A smarter setup that recognises when the page is actually reloading versus just updating would cut a lot of that waste.
  • Mobile details has no ads at all. That is a deliberate call, and probably a sensible one given how protective the business is of its conversion funnel. But desktop details does carry ads, which suggests there is at least some appetite to test the idea carefully on mobile too.
  • Logged-in users are the real unlock. Someone who is actively looking to buy deserves a lighter ad experience. Someone browsing million-dollar homes on a Sunday afternoon for fun is fair game. Splitting the two and treating them differently is where the real revenue opportunity sits.
  • The easiest desktop win costs almost nothing. The bottom 300x250 only accepts one size. Making it multisize and letting a 728x90 compete for the same slot would lift revenue with almost zero product effort and nothing visible changing for the user.
  • Two obvious gaps on mobile. A sticky footer is straightforward, high-volume inventory that Zillow simply is not running right now. And with mortgage, finance, and insurance brands all desperate to reach this audience, native in-feed formats are an easy fit that would not feel out of place alongside the listings.

What makes Zillow's business model different, and why does it shape their ad strategy?

Ben: Hey, and welcome to episode 4 of Layout Teardown, where we go through some of the layouts and ad strategies of some of the biggest websites globally. I'm Ben.

Brock: I'm Brock.

Ben: And today we're going to bring you through Zillow, which is the largest property marketplace in the US. They are a mega publisher — I think one of the top 20 globally — and we're going to bring you through the strategy that they will use for advertising, why that's different to a kind of standard publisher, where they see that revenue stream sitting, and what we would do differently too. So Zillow was founded back in 2006, quickly became the biggest property marketplace in the US. Today they sit just behind Reddit within their kind of top publishers globally, getting about 105 million unique visitors every year. They are three times bigger than their next competitors in market. So really a beast of a business that's been going for a very long time. From what we can see in their investor deck, programmatic advertising is only 2% of total revenue. So this is definitely going to change how they run their advertising strategy — what we need to look at and what we need to take into account. Brock, what do you think is going to be different for them versus your standard kind of big publisher?

Brock: Yeah, I think the major piece for them is really going to be the impact ads have on the rest of their business. So if you look at the pie chart there, we can see that the traditional buying and selling of property is going to dwarf — and continues to outpace — everything else on that chart. But of course, you have to consider cases where a highly engaging ad might drag someone away from the website, and therefore they don't go through for two hours or, you know, apply for a contract or other functions that the website provides. On the other hand, however, we would definitely be looking at ways that we can leverage their massive reach to build up an engaged audience — to get log-ins, particularly now as cookies start to decline — and really drive revenue through all channels because of their huge presence in this market.

Ben: Yeah. One of the things that makes Zillow interesting is it's free to put a listing on the site. A lot of property marketplaces charge quite a bit for that. And they've definitely diversified all the different revenue streams that they have. Even still, programmatic advertising will be seen as a nice-to-have that's adding some value. So our job is to make sure it's not impacting anything in the business widely, and that it can actually maybe facilitate some of the other areas of the business in different ways to drive more outcomes and generate more revenue.

Brock: Yeah. To add to that — if you look at the second smallest section of the pie chart, mortgage being 5% — I'm guessing that's a new area of business for them. One easy approach might be to leverage the aforementioned logged-in data, assuming that they have it, and really start to use segments of users across Zillow to be pitching them mortgage products through advertising or otherwise.

Ben: Yeah, nice. Okay. So we're going to jump into the layout. For really any marketplace, you've really got two core parts of the website: you've got search, which generates about 70% of traffic, and then you've got the details page, which is where you go and look into the photos and get some more information on what you're going to buy — about 30% of traffic. So we're going to go through those in order, look at them on both desktop and mobile, give you our opinions on how it is today, what we might change conservatively, and — let's say best case, we got to do what we wanted — what would we do to really drive outcomes from a programmatic standpoint?

What does Zillow's traffic data tell us, and how should it shape audience strategy?

Ben: So we are jumping into the layout. Before we do that, we're going to have a look at some of the data that we have access to. Now, we haven't spoken with Zillow — we don't have too much insider information. So a lot of this is just stuff we're grabbing from the investor deck and from SimilarWeb, a Chrome extension we use. Within the investor deck, we can see they're getting about 100 million unique users. Really chunky in terms of traffic. Looking into SimilarWeb, we can see most of this is all in the US, as we'd expect. Generally, these kinds of sites are pretty localised to where the property is, which makes a lot of sense. Then it's about 60% coming from direct and 40% coming from search. What would that tell you normally, Brock? What are you getting from that?

Brock: They're getting engaged users. I'm thinking a brand that is well known in the market — the overwhelming direct-over-search number makes me think people are going to Zillow because Zillow is the brand that gives them what they want, and that will be buying houses or, like me, looking at all the photos of all the expensive houses.

Ben: Yes, definitely. And we can see that in the deck — they're three times bigger in terms of traffic than their competitors. People know Zillow. They're coming here to look at the properties that they're interested in.

What does the desktop search page look like, and how is the ad stack set up?

Ben: So if we then jump into our search page, we'll do this on desktop first. So Zillow have gone for a slightly different first version on their desktop site — we've got the map in full view. A lot of other property sites like Rightmove, Realtor, REA in Australia will generally do the list and you can move to a map, but they've gone straight for the map, which is quite interesting. And then we've got our listings on the side. We can see that we've got some version of programmatic running — six partners running in the auction. We saw an ad, it was pretty thin, but we can see they've definitely got some kind of programmatic advertising running, ads serving through Google Ad Manager, as you would expect. So we've got a little bit here to work with and to look at. Passing it to you, Brock — what are you seeing? What does it look like at first glance?

Brock: Yeah, I see confirmation of the investor deck. There aren't thousands and thousands of requests flying out to all ad tech vendors in the ecosystem — looking at the network file, it's only sort of 10 or 15 rows long. It looks like it's all server-side or built in-house. So that gives me a quite simple, I imagine fairly conservative, approach in the way that they're rolling out these changes — because 98% of their revenue is not that. So this is a very standard approach, I would say, for a business in their position.

Definition — Server-side header bidding: an alternative to client-side Prebid where the auction runs on a server rather than in the user's browser, reducing page latency but often providing less transparency into the bidding process.

Ben: Yeah. One key piece to note is that even within that programmatic advertising piece, there'll be parts carved out for open auction — which is your straight auction — as well as quite a lot of direct deals that they're setting up with advertisers, agencies, realtors, etc. They also have their own self-service platform where you can go in and set up campaigns and deals. So they've really tried to scope that out — it's self-serve, you can get it set up yourselves without too much effort on their side. So actually looking for some ads on the page — as we start to scroll, we can see that we've got ads loading in straight away. Generally from what we can see, they're fairly hidden — or I would say pushed down to the very bottom of the page. What do you think the performance is going to be? Thoughts on that?

Brock: I think the good is putting it near pagination. So you can see here on the first page of this search listing — usually putting ads around areas where people are going to interact helps from a viewability and interaction metrics point of view. The bit that I'm not liking as much is how far it is down the page. But I am definitely biased — considering the state of where they are from a programmatic growth point of view, this is totally normal. I would not expect to see anything really different yet.

Ben: What I have noticed though is when you play around with the map — as you do so, the page starts to recalculate. And you can kind of see there on the screen that we haven't actually added much in terms of rich information about the properties at this point, but we've made extra requests and we're reloading the units over and over and over again. So my expectation, because they've gone map-first — a user is very likely going to make two, three, four, five zooms or drags. Reinitialisation of that sidebar.

Brock: Yeah, which makes me think it might have a fair bit of wastage in terms of requests right now. If that is really a concern, that's to be seen — because I believe they've also added some sort of on-scroll loading capability, just to ease the tension a little bit. From a metrics point of view, yeah — high floor — and that makes complete sense. So what I'd like to see is one of two things. One: can we get the ad into a position where it's more likely to be interacted with? Now that is a huge ask considering what they're trying to do here, with multimillion-dollar properties. So A/B testing and running it in a structured way is very, very important.

Definition — Price floor: the minimum CPM a publisher will accept for a given ad impression. Setting a high floor on premium inventory prevents it from being bought cheaply in the open auction.

Brock: That's why pushing the log-in strategy is also very important — because you can very easily find out: here's my segment of users that are logged in, they show propensity to buy, probably don't want to block them in any way or distract them with advertising. Whereas across 100 million uniques, there must be at least a million — I'd say a lot more than that — who are there just to look at the properties or to initially scope out where they might want to live. They'd be a perfect audience to start trying to get a bit of extra revenue out of through programmatic.

Ben: Right. So the key takeaways there were: we've got an ad loading in at the bottom, generally we're not seeing a huge amount of lazy loading, so performance probably isn't amazing on your key viewability and CTR metrics. As we move the map around, we're getting a refresh, but that ad unit is still there — so that could also be impacting the general performance we've got. You generally want to move that ad somewhere that's in a better location, but we're going to target specific segments of audiences.

Brock: Yeah, that was definitely one way to do it. The other way is: do we actually need to refresh the ad units when we make these adjustments to the map? It looks very SPA/PWA-style from the implementation.

Definition — SPA/PWA (Single-Page Application / Progressive Web App): web architectures where content loads dynamically without full page reloads, which requires special handling for ad refresh logic to avoid excessive or wasted ad requests.

Ben: Yeah. I haven't dug in enough to tell you exactly the answer, but the way to tell is — we're interacting with the map, we're loading alternative content without refreshing the whole page. The list to the right-hand side is being recycled. It looks like that list widget is recycling. The ad itself doesn't need to do that.

Brock: Maybe not. Of course, you'd take a hit on volume if you do it that way. But the question is: what kind of interaction are you getting? Are we sending requests out to the void? Can we improve on other metrics by doing it that way?

Ben: Yeah, definitely. It's an interesting point about deciding who we're going to show different ad strategies to as well. We've talked to a few different online marketplaces recently and what Brock described is pretty common — you've got a smaller group that are highly engaged and ready to purchase, so whether that's renting a new property or buying a house. And then you've got actually a fairly significant number of users that are browsing and just kind of looking around — that's what they do on the weekends. I definitely look at other properties on the weekends. Targeting them differently makes sense because they're less likely to generate revenue on other parts of the business, so we can monetise them from a programmatic standpoint. So if you've got your CDP or DMP set up with these segments in it, we can basically pick up those interactions with the page and show a different ad strategy to those different users — customise it, make sure it fits with the wider business model and generates revenue.

Definition — CDP/DMP (Customer Data Platform / Data Management Platform): platforms used to collect and activate audience data for ad targeting. A CDP typically handles first-party data; a DMP handles broader third-party audience data.

What does the desktop details page look like, and what are the ad opportunities?

Ben: Okay, let's click on a property. So clicking in here brings us to the details section. Again, this feels like a SPA because we're not really loading a new page here. And to get an ad unit, we've got to scroll down quite a bit — and I think there's two in a row. Yeah, there's a double-stacked 300x250 on top of each other there. What's your first thought here?

Brock: I think it's similar. We — being Zillow — are trying not to bombard users with ads in the key areas. I think it's a lot more understandable at this point in the journey. There is implied interest in the property. The question really for me would be: how often is a user going through and looking at photos relative to actually going through and making any other kind of interaction — be it requesting a contract or requesting a property tour, or whatever they'd be interested in?

Ben: So if you were to go through and look at the images — we now have this big screen view. This is where the user data becomes really interesting, because now our opportunity to monetise from an advertising point of view has decreased significantly. You could very easily shift the size of these images, deal with the pagination problem, and put a unit in there. So we'd have — basically, based on the user type, let's say unlikely to buy this $10 million house — we could have a billboard or a skyscraper in the side there and generate some pretty good revenue. Commonly it's a vertical unit, so about 600 pixels tall. It also opens up the opportunity for things like lead capture forms, native products that might be geo-targeted as well. This is a rich information page, even if it is inferred. And that, I believe, will give them a few extra strings in their bow and go to market with new advertising products.

Brock: Yeah, nice. And any other pieces you would go to if you could do anything here?

Ben: If I could do anything? I think having something that is visible on this screen is probably the biggest way you're going to change the game. That is a seismic change — yeah, that's not a beat around the bush. But when looking at some of the other competitors on the market, they very commonly have above-everything leaderboards. Alternatively, if you don't mind scrolling through the feed, you could use those positions for regular display. Even leaderboards.

Brock: Well, now if you can zoom into, say, New York or something, we'll get some more properties. Because one thing I did notice — we've got the 300x250 at the bottom of the page. It only accepts a 300x250, but on a fairly normal laptop screen — some of us, when we pretend to work on our big screens at the office — you would definitely be able to fit a 728x90 in there, even in the same position. So yes, if you were to ask me the opposite question — if I couldn't change anything, I had the space to work with — I think the easiest thing to do would be to make that unit multisize. So we'd have multiple size options other than just 300x250 able to fill there.

Definition — Multisize ad unit: an ad slot configured to accept multiple creative sizes (e.g. both 300x250 and 728x90), increasing the pool of eligible demand and improving fill rate and CPM.

Ben: Yeah. And you can see that they've kind of done this in other positions on the site. We saw the double-stacked 300x250 on the property details page — you could even do that. It's a much easier sell to the rest of the business since they're used to it.

Brock: But if you can fit a 728x90, you could probably do both. Which I think we'd just about have a bit of space to go for. So yeah, that would be a nice optimisation. GAM or otherwise lets you manage the size mappings very, very easily. So this is a pretty simple implementation — it really just depends on what we want from a consumer-facing point of view.

Ben: Yeah. My best case would be we'd put a 728x90 on the map so it's in view permanently. I'm pretty sure there's a pretty long visit duration on site — we'd get quite a lot out of it. And I'm sure a lot of product managers at Zillow would die before that ever happened. But yeah, a few options there.

What does Zillow's mobile search layout look like, and what's missing?

Ben: Okay, let's cut to mobile. So we are now going to look through the mobile layouts. We'll start off with search and then move into details. Our guess is this will be about at least 50% of traffic. And generally it's a little bit harder to monetise both from a programmatic perspective, but also just generally from an advertising standpoint. Ad units are a little bit smaller, a bit harder to get viewability and CTR on — smaller screen. The first thing I can see is a push towards downloading their app. Brock, why do you think that's there?

Brock: Yeah, I generally think this is very common practice in the industry. We're trying to move users into a much richer experience where the engineers building the native product are able to ship a cool fancy map, gallery features, all those kinds of things — you know, it's endemic to Android and iOS, they can build customisation around each. I also would say that they're more likely to get a log-in. It's fairly common that when you download any app — be it real estate or otherwise — you're presented with, particularly on iOS, iOS location data and all the various things you quickly click through. You'll be able to get a log-in, and that'll be very, very important and paramount to their strategy for building up their logged-in audience.

Ben: Yeah, generally that's a pretty big metric that a lot of these big publishers will track — logged-in versus logged-out user growth. Logged-in users are generally more valuable, and as cookies deprecate, we want to be able to target users on first-party data and generally build products for them in a good way. So that makes a lot of sense. In terms of the layout — they've gone for the list view this time versus starting off with the map. A bit of a change. My guess is this is just, you know, a little bit easier to use a list on mobile versus desktop.

Brock: Yeah, I would have thought so. We also get into scroll mentality — just keep on going, keep scrolling, browsing more images, probably some predictor of interaction there. Whereas on the other hand, map technology on an iPhone or Android is probably not amazing. You've got pinch and click, particularly for those that are not used to it. The list seems more accessible for the mobile user, I would say.

Ben: Yeah, agreed. From an ad perspective, we've seen what looks like two combinations — one is a 300x250 down the very bottom of the page, and then we see two sponsored units coming in that are kind of a native style that look like the property listings. What's your take so far on this setup? What would you do differently?

Brock: It makes a lot of sense. I think we've got a premium position relative to other listings. I believe it's third or fourth. It's marked as sponsored. It looks like the content. One thing to test would be how it responds on a horizontal mobile or on an iPad layout. But I think about those 100 million users — there has to be a significant amount viewing this site or even this list across regular mobile, so vertical, horizontal, then you've got your iPad users, and you move into your desktop. Right here is what I would call a vertical native design — we have image on top and call to action beneath. Generally what we see is as your screen gets wider, you get a much more distorted picture, or you compensate in ratios and have a really tall image, which is probably not what they're trying to achieve. You'd probably test that by dragging along your responsive bar.

Ben: Yeah. Because one thing — they've managed it well. One thing they could do is look to use horizontal native, and that's where you can move into spaces like a leaderboard through multisize, or rely on Google Ad Manager — the ad server — to serve up nice developer creatives, or whoever is selling it, ahead of something programmatic and horizontal. Going to double this view is a perfect way to manage it.

Brock: Yeah, nice. If you were to add to or adjust this — what are you recommending if you are in the room with the Zillow team? Yeah, I think particularly if they have a way to track it — moreover, if someone goes and closes the app download banner. If you want to have a go, it's gone. If there's space there that's already being used, that users are used to — you're able to supplement that with a fancy direct position like what they have in these content units, or even just something sticky with a high floor price like that. That's a very low-impact way relative to the rest of the site. We're talking that the ratio of content to ads — a 320x50 or something — will be an earner just on pure volume.

Ben: Yeah, always in view, sellable. If you only go direct you'd probably do well that way — because of the map and sort buttons that you have down the bottom, it'd be hard to put a sticky footer down there.

Brock: Yeah, it's fairly prominent down there. One thing you would have to manage is other fields up top. So right now as you scroll, you've got consistent UX elements at the top and the bottom. So one thing you might have to do is add space, or use a format that makes sense and isn't going to be accidentally clicked and send your click spam scores to the moon.

Ben: Yeah. The second part is what we're doing in content. Native first makes a lot of sense — I would imagine that mortgage brands, finance brands, insurance brands — all of those guys would be itching for that kind of content. One thing they could do, and they might do this — I haven't seen it — is use those native positions as multi-format positions. So have a nice, kind of informal in-feed video — that might be a way to break into the video space, considering real estate video on mobile is probably a bit of a hot topic. So they'd still maintain their nice image up top, buttons down the bottom, call-to-action style — but run a video player and maybe a taller unit altogether. An outstream-style player. So we're still getting that native feel, we're not breaking the UX, but we're getting different inventory that we can monetise in the same space effectively.

Definition — Outstream video: video ads that play within page content rather than inside a video player, allowing publishers to monetise with video without producing video content themselves.

Brock: Yeah, that's right. And it's more options, more products for their teams to go to market with. And that being said, you're always going to be able to backfill with programmatic — at a very high floor with pretty well-defined block lists and allow lists. If that's better for the Zillow team, they could really keep those units as a premium buy and then increase their volume by adding more products into the same position.

What does Zillow's mobile details page look like, and what would you change?

Ben: Yeah, definitely. Opening up then the details section — the first thing we noticed is there are no ads. So this is something that often happens in kind of online marketplaces. Details is the kind of — you know, you've now got the user through the funnel, they're definitely more interested, they are reading the details. And generally the agents and dealers and sellers are very conscious of any advertising here taking away from their opportunity to generate leads, and to get what they pay for when they come to Zillow. It's free, but most of their revenue is coming from kind of the paid versions. So getting ads in here is not super easy. We've done it with some online marketplaces, but you really need to show that there is no impact to that funnel and through the leads coming through to the buyers. So that's the key thing we're looking for.

Ben: So would any of the changes we'd be making with Zillow — the recommendation from us is always to A/B test, run around 1% of traffic. We want to see does it impact revenue. But really what we want to see is what's the impact on the wider business? Are we still getting our leads through the funnel? Are they still making it to the details page? And then are they still taking that apply button? Anything you'd add, or where would you be placing ads to run those kind of 1% tests?

Brock: Yeah, this one's particularly interesting because on desktop details we saw before they had the most ads in terms of density — the two on top of each other. And when I was looking at Zillow before on my large screen, they had two side by side as well, right at the bottom. So this infers that there's appetite — specifically on desktop — for it.

Ben: Yeah. But on mobile web, do we have less of a conversion perhaps?

Brock: Yeah. All the upheaval of just moving through the app and that's the goal. That also might be the case. So if mobile web is just a springboard to the app, then maybe the approach is just push more app download banners and then use positions that would go to programmatic or to native for browsing-type users.

Ben: Yeah. So that's one thing. Second part really is going to be getting units in where the people are. I think above the fold will be very tough in this case. Although going into the gallery would be the easiest way. Again, similar to desktop, you've got scrollable content, you're able to utilise fairly standard sizes. In this case it's squares — a 300x250 works perfectly — or native capture forms as well. You can kind of see there at the bottom that they're attempting something like that. The real question will be: could we A/B test an ad against a different kind of third-party or sign-up — for a quote on electricity or something — the direct placement versus what they have now?

Brock: Yeah. And then again, as you scroll through, you can see we've got Google Maps sections scoring. Basically what we want to be doing is figuring out — you can use things like heat map tools and there's plenty of things out there — to really understand where the attention is on this page. And that's where you'd want to start. I'd be testing, even if it's in small, very small ads, small-scale testing, small percentages. At 100 million uniques, that makes a big difference.

Summary: What are the biggest opportunities for Zillow's ad strategy?

Ben: Okay. So to recap — Zillow has a lot going on. There's a range of different business models and they're trying to diversify as much as possible. Ads are a part of that, but it's very small. So any changes we're going to be doing need to be brought in with a mind to the larger ecosystem that is Zillow, and measuring on small fractions of traffic and measuring the impact that has to those different funnels. We did notice a bit of inconsistency through that. On mobile, they've got a nice fancy interactive map — what happens is all the ads are pushed to the bottom. There are many ways that we can make this better: one thing might be optimising the position they're in, the other thing might be optimising when they're loaded and how they're loaded.

Brock: Yeah, and we looked at breaking out the advertising strategies based on segments as well. So lots of users will be in market to find a new rental or a new home, but lots of users will just be browsing. And so we want to show different strategies to these different users. So by picking up signals from the CDP or DMP, we can adjust the layout and the ad stack to allow us to monetise those different user groups in different ways.

Ben: App looks like a big thing, particularly for mobile — obviously continuing to push that stream makes a lot of sense, even if it is at the expense of some ad revenue on mobile web. So yeah, lots of room to grow — and doing that in a safe way that doesn't impact their wider business. But definitely some good opportunities there to increase yields, increase the inventory that they've got, and drive a bit more revenue.

Ben: Okay. So that's episode 4 of Layout Teardown. If you enjoyed the video, please like it, leave a comment if you learned something, subscribe — it helps us all out here. And any publishers out there that would like us to tear down their layout, please do reach out and click the link below. Thank you.

This is an edited transcript of Layout Teardown, Episode 4. The words are Ben and Brock's own — lightly edited for readability (filler words, false starts, typos, punctuation). No claims have been rewritten or generated by AI.

‍

Publift Scoop

Publift Scoop

Sign up to the Publift newsletter for all the latest news, updates and insights from across the industry.
Sign up
Play video
Play icon
Play video

Layout Teardown: Episode 4 - Zillow.com's Unique Advertising Strategies

Layout Teardown: Episode 4 - Zillow.com's Unique Advertising Strategies

Contents
Contents

Layout Teardown: Episode 4 - Zillow.com's Unique Advertising Strategies

Ben and Brock explore how Zillow optimises its ad strategy without affecting its core business, offering insights into both desktop and mobile layouts for improving ad visibility and user engagement.

‍

Key Takeaways

  • Programmatic is a rounding error on the balance sheet. Ads bring in just 2% of Zillow's total revenue. That shapes every decision. Nothing in the ad stack is worth touching if it puts the leads and listings business at risk.
  • The map creates a hidden waste problem. Zillow's desktop search is built around a map, and every time a user drags or zooms, the page reinitialises and fires off a fresh round of ad requests. Most of those go nowhere. A smarter setup that recognises when the page is actually reloading versus just updating would cut a lot of that waste.
  • Mobile details has no ads at all. That is a deliberate call, and probably a sensible one given how protective the business is of its conversion funnel. But desktop details does carry ads, which suggests there is at least some appetite to test the idea carefully on mobile too.
  • Logged-in users are the real unlock. Someone who is actively looking to buy deserves a lighter ad experience. Someone browsing million-dollar homes on a Sunday afternoon for fun is fair game. Splitting the two and treating them differently is where the real revenue opportunity sits.
  • The easiest desktop win costs almost nothing. The bottom 300x250 only accepts one size. Making it multisize and letting a 728x90 compete for the same slot would lift revenue with almost zero product effort and nothing visible changing for the user.
  • Two obvious gaps on mobile. A sticky footer is straightforward, high-volume inventory that Zillow simply is not running right now. And with mortgage, finance, and insurance brands all desperate to reach this audience, native in-feed formats are an easy fit that would not feel out of place alongside the listings.

What makes Zillow's business model different, and why does it shape their ad strategy?

Ben: Hey, and welcome to episode 4 of Layout Teardown, where we go through some of the layouts and ad strategies of some of the biggest websites globally. I'm Ben.

Brock: I'm Brock.

Ben: And today we're going to bring you through Zillow, which is the largest property marketplace in the US. They are a mega publisher — I think one of the top 20 globally — and we're going to bring you through the strategy that they will use for advertising, why that's different to a kind of standard publisher, where they see that revenue stream sitting, and what we would do differently too. So Zillow was founded back in 2006, quickly became the biggest property marketplace in the US. Today they sit just behind Reddit within their kind of top publishers globally, getting about 105 million unique visitors every year. They are three times bigger than their next competitors in market. So really a beast of a business that's been going for a very long time. From what we can see in their investor deck, programmatic advertising is only 2% of total revenue. So this is definitely going to change how they run their advertising strategy — what we need to look at and what we need to take into account. Brock, what do you think is going to be different for them versus your standard kind of big publisher?

Brock: Yeah, I think the major piece for them is really going to be the impact ads have on the rest of their business. So if you look at the pie chart there, we can see that the traditional buying and selling of property is going to dwarf — and continues to outpace — everything else on that chart. But of course, you have to consider cases where a highly engaging ad might drag someone away from the website, and therefore they don't go through for two hours or, you know, apply for a contract or other functions that the website provides. On the other hand, however, we would definitely be looking at ways that we can leverage their massive reach to build up an engaged audience — to get log-ins, particularly now as cookies start to decline — and really drive revenue through all channels because of their huge presence in this market.

Ben: Yeah. One of the things that makes Zillow interesting is it's free to put a listing on the site. A lot of property marketplaces charge quite a bit for that. And they've definitely diversified all the different revenue streams that they have. Even still, programmatic advertising will be seen as a nice-to-have that's adding some value. So our job is to make sure it's not impacting anything in the business widely, and that it can actually maybe facilitate some of the other areas of the business in different ways to drive more outcomes and generate more revenue.

Brock: Yeah. To add to that — if you look at the second smallest section of the pie chart, mortgage being 5% — I'm guessing that's a new area of business for them. One easy approach might be to leverage the aforementioned logged-in data, assuming that they have it, and really start to use segments of users across Zillow to be pitching them mortgage products through advertising or otherwise.

Ben: Yeah, nice. Okay. So we're going to jump into the layout. For really any marketplace, you've really got two core parts of the website: you've got search, which generates about 70% of traffic, and then you've got the details page, which is where you go and look into the photos and get some more information on what you're going to buy — about 30% of traffic. So we're going to go through those in order, look at them on both desktop and mobile, give you our opinions on how it is today, what we might change conservatively, and — let's say best case, we got to do what we wanted — what would we do to really drive outcomes from a programmatic standpoint?

What does Zillow's traffic data tell us, and how should it shape audience strategy?

Ben: So we are jumping into the layout. Before we do that, we're going to have a look at some of the data that we have access to. Now, we haven't spoken with Zillow — we don't have too much insider information. So a lot of this is just stuff we're grabbing from the investor deck and from SimilarWeb, a Chrome extension we use. Within the investor deck, we can see they're getting about 100 million unique users. Really chunky in terms of traffic. Looking into SimilarWeb, we can see most of this is all in the US, as we'd expect. Generally, these kinds of sites are pretty localised to where the property is, which makes a lot of sense. Then it's about 60% coming from direct and 40% coming from search. What would that tell you normally, Brock? What are you getting from that?

Brock: They're getting engaged users. I'm thinking a brand that is well known in the market — the overwhelming direct-over-search number makes me think people are going to Zillow because Zillow is the brand that gives them what they want, and that will be buying houses or, like me, looking at all the photos of all the expensive houses.

Ben: Yes, definitely. And we can see that in the deck — they're three times bigger in terms of traffic than their competitors. People know Zillow. They're coming here to look at the properties that they're interested in.

What does the desktop search page look like, and how is the ad stack set up?

Ben: So if we then jump into our search page, we'll do this on desktop first. So Zillow have gone for a slightly different first version on their desktop site — we've got the map in full view. A lot of other property sites like Rightmove, Realtor, REA in Australia will generally do the list and you can move to a map, but they've gone straight for the map, which is quite interesting. And then we've got our listings on the side. We can see that we've got some version of programmatic running — six partners running in the auction. We saw an ad, it was pretty thin, but we can see they've definitely got some kind of programmatic advertising running, ads serving through Google Ad Manager, as you would expect. So we've got a little bit here to work with and to look at. Passing it to you, Brock — what are you seeing? What does it look like at first glance?

Brock: Yeah, I see confirmation of the investor deck. There aren't thousands and thousands of requests flying out to all ad tech vendors in the ecosystem — looking at the network file, it's only sort of 10 or 15 rows long. It looks like it's all server-side or built in-house. So that gives me a quite simple, I imagine fairly conservative, approach in the way that they're rolling out these changes — because 98% of their revenue is not that. So this is a very standard approach, I would say, for a business in their position.

Definition — Server-side header bidding: an alternative to client-side Prebid where the auction runs on a server rather than in the user's browser, reducing page latency but often providing less transparency into the bidding process.

Ben: Yeah. One key piece to note is that even within that programmatic advertising piece, there'll be parts carved out for open auction — which is your straight auction — as well as quite a lot of direct deals that they're setting up with advertisers, agencies, realtors, etc. They also have their own self-service platform where you can go in and set up campaigns and deals. So they've really tried to scope that out — it's self-serve, you can get it set up yourselves without too much effort on their side. So actually looking for some ads on the page — as we start to scroll, we can see that we've got ads loading in straight away. Generally from what we can see, they're fairly hidden — or I would say pushed down to the very bottom of the page. What do you think the performance is going to be? Thoughts on that?

Brock: I think the good is putting it near pagination. So you can see here on the first page of this search listing — usually putting ads around areas where people are going to interact helps from a viewability and interaction metrics point of view. The bit that I'm not liking as much is how far it is down the page. But I am definitely biased — considering the state of where they are from a programmatic growth point of view, this is totally normal. I would not expect to see anything really different yet.

Ben: What I have noticed though is when you play around with the map — as you do so, the page starts to recalculate. And you can kind of see there on the screen that we haven't actually added much in terms of rich information about the properties at this point, but we've made extra requests and we're reloading the units over and over and over again. So my expectation, because they've gone map-first — a user is very likely going to make two, three, four, five zooms or drags. Reinitialisation of that sidebar.

Brock: Yeah, which makes me think it might have a fair bit of wastage in terms of requests right now. If that is really a concern, that's to be seen — because I believe they've also added some sort of on-scroll loading capability, just to ease the tension a little bit. From a metrics point of view, yeah — high floor — and that makes complete sense. So what I'd like to see is one of two things. One: can we get the ad into a position where it's more likely to be interacted with? Now that is a huge ask considering what they're trying to do here, with multimillion-dollar properties. So A/B testing and running it in a structured way is very, very important.

Definition — Price floor: the minimum CPM a publisher will accept for a given ad impression. Setting a high floor on premium inventory prevents it from being bought cheaply in the open auction.

Brock: That's why pushing the log-in strategy is also very important — because you can very easily find out: here's my segment of users that are logged in, they show propensity to buy, probably don't want to block them in any way or distract them with advertising. Whereas across 100 million uniques, there must be at least a million — I'd say a lot more than that — who are there just to look at the properties or to initially scope out where they might want to live. They'd be a perfect audience to start trying to get a bit of extra revenue out of through programmatic.

Ben: Right. So the key takeaways there were: we've got an ad loading in at the bottom, generally we're not seeing a huge amount of lazy loading, so performance probably isn't amazing on your key viewability and CTR metrics. As we move the map around, we're getting a refresh, but that ad unit is still there — so that could also be impacting the general performance we've got. You generally want to move that ad somewhere that's in a better location, but we're going to target specific segments of audiences.

Brock: Yeah, that was definitely one way to do it. The other way is: do we actually need to refresh the ad units when we make these adjustments to the map? It looks very SPA/PWA-style from the implementation.

Definition — SPA/PWA (Single-Page Application / Progressive Web App): web architectures where content loads dynamically without full page reloads, which requires special handling for ad refresh logic to avoid excessive or wasted ad requests.

Ben: Yeah. I haven't dug in enough to tell you exactly the answer, but the way to tell is — we're interacting with the map, we're loading alternative content without refreshing the whole page. The list to the right-hand side is being recycled. It looks like that list widget is recycling. The ad itself doesn't need to do that.

Brock: Maybe not. Of course, you'd take a hit on volume if you do it that way. But the question is: what kind of interaction are you getting? Are we sending requests out to the void? Can we improve on other metrics by doing it that way?

Ben: Yeah, definitely. It's an interesting point about deciding who we're going to show different ad strategies to as well. We've talked to a few different online marketplaces recently and what Brock described is pretty common — you've got a smaller group that are highly engaged and ready to purchase, so whether that's renting a new property or buying a house. And then you've got actually a fairly significant number of users that are browsing and just kind of looking around — that's what they do on the weekends. I definitely look at other properties on the weekends. Targeting them differently makes sense because they're less likely to generate revenue on other parts of the business, so we can monetise them from a programmatic standpoint. So if you've got your CDP or DMP set up with these segments in it, we can basically pick up those interactions with the page and show a different ad strategy to those different users — customise it, make sure it fits with the wider business model and generates revenue.

Definition — CDP/DMP (Customer Data Platform / Data Management Platform): platforms used to collect and activate audience data for ad targeting. A CDP typically handles first-party data; a DMP handles broader third-party audience data.

What does the desktop details page look like, and what are the ad opportunities?

Ben: Okay, let's click on a property. So clicking in here brings us to the details section. Again, this feels like a SPA because we're not really loading a new page here. And to get an ad unit, we've got to scroll down quite a bit — and I think there's two in a row. Yeah, there's a double-stacked 300x250 on top of each other there. What's your first thought here?

Brock: I think it's similar. We — being Zillow — are trying not to bombard users with ads in the key areas. I think it's a lot more understandable at this point in the journey. There is implied interest in the property. The question really for me would be: how often is a user going through and looking at photos relative to actually going through and making any other kind of interaction — be it requesting a contract or requesting a property tour, or whatever they'd be interested in?

Ben: So if you were to go through and look at the images — we now have this big screen view. This is where the user data becomes really interesting, because now our opportunity to monetise from an advertising point of view has decreased significantly. You could very easily shift the size of these images, deal with the pagination problem, and put a unit in there. So we'd have — basically, based on the user type, let's say unlikely to buy this $10 million house — we could have a billboard or a skyscraper in the side there and generate some pretty good revenue. Commonly it's a vertical unit, so about 600 pixels tall. It also opens up the opportunity for things like lead capture forms, native products that might be geo-targeted as well. This is a rich information page, even if it is inferred. And that, I believe, will give them a few extra strings in their bow and go to market with new advertising products.

Brock: Yeah, nice. And any other pieces you would go to if you could do anything here?

Ben: If I could do anything? I think having something that is visible on this screen is probably the biggest way you're going to change the game. That is a seismic change — yeah, that's not a beat around the bush. But when looking at some of the other competitors on the market, they very commonly have above-everything leaderboards. Alternatively, if you don't mind scrolling through the feed, you could use those positions for regular display. Even leaderboards.

Brock: Well, now if you can zoom into, say, New York or something, we'll get some more properties. Because one thing I did notice — we've got the 300x250 at the bottom of the page. It only accepts a 300x250, but on a fairly normal laptop screen — some of us, when we pretend to work on our big screens at the office — you would definitely be able to fit a 728x90 in there, even in the same position. So yes, if you were to ask me the opposite question — if I couldn't change anything, I had the space to work with — I think the easiest thing to do would be to make that unit multisize. So we'd have multiple size options other than just 300x250 able to fill there.

Definition — Multisize ad unit: an ad slot configured to accept multiple creative sizes (e.g. both 300x250 and 728x90), increasing the pool of eligible demand and improving fill rate and CPM.

Ben: Yeah. And you can see that they've kind of done this in other positions on the site. We saw the double-stacked 300x250 on the property details page — you could even do that. It's a much easier sell to the rest of the business since they're used to it.

Brock: But if you can fit a 728x90, you could probably do both. Which I think we'd just about have a bit of space to go for. So yeah, that would be a nice optimisation. GAM or otherwise lets you manage the size mappings very, very easily. So this is a pretty simple implementation — it really just depends on what we want from a consumer-facing point of view.

Ben: Yeah. My best case would be we'd put a 728x90 on the map so it's in view permanently. I'm pretty sure there's a pretty long visit duration on site — we'd get quite a lot out of it. And I'm sure a lot of product managers at Zillow would die before that ever happened. But yeah, a few options there.

What does Zillow's mobile search layout look like, and what's missing?

Ben: Okay, let's cut to mobile. So we are now going to look through the mobile layouts. We'll start off with search and then move into details. Our guess is this will be about at least 50% of traffic. And generally it's a little bit harder to monetise both from a programmatic perspective, but also just generally from an advertising standpoint. Ad units are a little bit smaller, a bit harder to get viewability and CTR on — smaller screen. The first thing I can see is a push towards downloading their app. Brock, why do you think that's there?

Brock: Yeah, I generally think this is very common practice in the industry. We're trying to move users into a much richer experience where the engineers building the native product are able to ship a cool fancy map, gallery features, all those kinds of things — you know, it's endemic to Android and iOS, they can build customisation around each. I also would say that they're more likely to get a log-in. It's fairly common that when you download any app — be it real estate or otherwise — you're presented with, particularly on iOS, iOS location data and all the various things you quickly click through. You'll be able to get a log-in, and that'll be very, very important and paramount to their strategy for building up their logged-in audience.

Ben: Yeah, generally that's a pretty big metric that a lot of these big publishers will track — logged-in versus logged-out user growth. Logged-in users are generally more valuable, and as cookies deprecate, we want to be able to target users on first-party data and generally build products for them in a good way. So that makes a lot of sense. In terms of the layout — they've gone for the list view this time versus starting off with the map. A bit of a change. My guess is this is just, you know, a little bit easier to use a list on mobile versus desktop.

Brock: Yeah, I would have thought so. We also get into scroll mentality — just keep on going, keep scrolling, browsing more images, probably some predictor of interaction there. Whereas on the other hand, map technology on an iPhone or Android is probably not amazing. You've got pinch and click, particularly for those that are not used to it. The list seems more accessible for the mobile user, I would say.

Ben: Yeah, agreed. From an ad perspective, we've seen what looks like two combinations — one is a 300x250 down the very bottom of the page, and then we see two sponsored units coming in that are kind of a native style that look like the property listings. What's your take so far on this setup? What would you do differently?

Brock: It makes a lot of sense. I think we've got a premium position relative to other listings. I believe it's third or fourth. It's marked as sponsored. It looks like the content. One thing to test would be how it responds on a horizontal mobile or on an iPad layout. But I think about those 100 million users — there has to be a significant amount viewing this site or even this list across regular mobile, so vertical, horizontal, then you've got your iPad users, and you move into your desktop. Right here is what I would call a vertical native design — we have image on top and call to action beneath. Generally what we see is as your screen gets wider, you get a much more distorted picture, or you compensate in ratios and have a really tall image, which is probably not what they're trying to achieve. You'd probably test that by dragging along your responsive bar.

Ben: Yeah. Because one thing — they've managed it well. One thing they could do is look to use horizontal native, and that's where you can move into spaces like a leaderboard through multisize, or rely on Google Ad Manager — the ad server — to serve up nice developer creatives, or whoever is selling it, ahead of something programmatic and horizontal. Going to double this view is a perfect way to manage it.

Brock: Yeah, nice. If you were to add to or adjust this — what are you recommending if you are in the room with the Zillow team? Yeah, I think particularly if they have a way to track it — moreover, if someone goes and closes the app download banner. If you want to have a go, it's gone. If there's space there that's already being used, that users are used to — you're able to supplement that with a fancy direct position like what they have in these content units, or even just something sticky with a high floor price like that. That's a very low-impact way relative to the rest of the site. We're talking that the ratio of content to ads — a 320x50 or something — will be an earner just on pure volume.

Ben: Yeah, always in view, sellable. If you only go direct you'd probably do well that way — because of the map and sort buttons that you have down the bottom, it'd be hard to put a sticky footer down there.

Brock: Yeah, it's fairly prominent down there. One thing you would have to manage is other fields up top. So right now as you scroll, you've got consistent UX elements at the top and the bottom. So one thing you might have to do is add space, or use a format that makes sense and isn't going to be accidentally clicked and send your click spam scores to the moon.

Ben: Yeah. The second part is what we're doing in content. Native first makes a lot of sense — I would imagine that mortgage brands, finance brands, insurance brands — all of those guys would be itching for that kind of content. One thing they could do, and they might do this — I haven't seen it — is use those native positions as multi-format positions. So have a nice, kind of informal in-feed video — that might be a way to break into the video space, considering real estate video on mobile is probably a bit of a hot topic. So they'd still maintain their nice image up top, buttons down the bottom, call-to-action style — but run a video player and maybe a taller unit altogether. An outstream-style player. So we're still getting that native feel, we're not breaking the UX, but we're getting different inventory that we can monetise in the same space effectively.

Definition — Outstream video: video ads that play within page content rather than inside a video player, allowing publishers to monetise with video without producing video content themselves.

Brock: Yeah, that's right. And it's more options, more products for their teams to go to market with. And that being said, you're always going to be able to backfill with programmatic — at a very high floor with pretty well-defined block lists and allow lists. If that's better for the Zillow team, they could really keep those units as a premium buy and then increase their volume by adding more products into the same position.

What does Zillow's mobile details page look like, and what would you change?

Ben: Yeah, definitely. Opening up then the details section — the first thing we noticed is there are no ads. So this is something that often happens in kind of online marketplaces. Details is the kind of — you know, you've now got the user through the funnel, they're definitely more interested, they are reading the details. And generally the agents and dealers and sellers are very conscious of any advertising here taking away from their opportunity to generate leads, and to get what they pay for when they come to Zillow. It's free, but most of their revenue is coming from kind of the paid versions. So getting ads in here is not super easy. We've done it with some online marketplaces, but you really need to show that there is no impact to that funnel and through the leads coming through to the buyers. So that's the key thing we're looking for.

Ben: So would any of the changes we'd be making with Zillow — the recommendation from us is always to A/B test, run around 1% of traffic. We want to see does it impact revenue. But really what we want to see is what's the impact on the wider business? Are we still getting our leads through the funnel? Are they still making it to the details page? And then are they still taking that apply button? Anything you'd add, or where would you be placing ads to run those kind of 1% tests?

Brock: Yeah, this one's particularly interesting because on desktop details we saw before they had the most ads in terms of density — the two on top of each other. And when I was looking at Zillow before on my large screen, they had two side by side as well, right at the bottom. So this infers that there's appetite — specifically on desktop — for it.

Ben: Yeah. But on mobile web, do we have less of a conversion perhaps?

Brock: Yeah. All the upheaval of just moving through the app and that's the goal. That also might be the case. So if mobile web is just a springboard to the app, then maybe the approach is just push more app download banners and then use positions that would go to programmatic or to native for browsing-type users.

Ben: Yeah. So that's one thing. Second part really is going to be getting units in where the people are. I think above the fold will be very tough in this case. Although going into the gallery would be the easiest way. Again, similar to desktop, you've got scrollable content, you're able to utilise fairly standard sizes. In this case it's squares — a 300x250 works perfectly — or native capture forms as well. You can kind of see there at the bottom that they're attempting something like that. The real question will be: could we A/B test an ad against a different kind of third-party or sign-up — for a quote on electricity or something — the direct placement versus what they have now?

Brock: Yeah. And then again, as you scroll through, you can see we've got Google Maps sections scoring. Basically what we want to be doing is figuring out — you can use things like heat map tools and there's plenty of things out there — to really understand where the attention is on this page. And that's where you'd want to start. I'd be testing, even if it's in small, very small ads, small-scale testing, small percentages. At 100 million uniques, that makes a big difference.

Summary: What are the biggest opportunities for Zillow's ad strategy?

Ben: Okay. So to recap — Zillow has a lot going on. There's a range of different business models and they're trying to diversify as much as possible. Ads are a part of that, but it's very small. So any changes we're going to be doing need to be brought in with a mind to the larger ecosystem that is Zillow, and measuring on small fractions of traffic and measuring the impact that has to those different funnels. We did notice a bit of inconsistency through that. On mobile, they've got a nice fancy interactive map — what happens is all the ads are pushed to the bottom. There are many ways that we can make this better: one thing might be optimising the position they're in, the other thing might be optimising when they're loaded and how they're loaded.

Brock: Yeah, and we looked at breaking out the advertising strategies based on segments as well. So lots of users will be in market to find a new rental or a new home, but lots of users will just be browsing. And so we want to show different strategies to these different users. So by picking up signals from the CDP or DMP, we can adjust the layout and the ad stack to allow us to monetise those different user groups in different ways.

Ben: App looks like a big thing, particularly for mobile — obviously continuing to push that stream makes a lot of sense, even if it is at the expense of some ad revenue on mobile web. So yeah, lots of room to grow — and doing that in a safe way that doesn't impact their wider business. But definitely some good opportunities there to increase yields, increase the inventory that they've got, and drive a bit more revenue.

Ben: Okay. So that's episode 4 of Layout Teardown. If you enjoyed the video, please like it, leave a comment if you learned something, subscribe — it helps us all out here. And any publishers out there that would like us to tear down their layout, please do reach out and click the link below. Thank you.

This is an edited transcript of Layout Teardown, Episode 4. The words are Ben and Brock's own — lightly edited for readability (filler words, false starts, typos, punctuation). No claims have been rewritten or generated by AI.

‍

Publift Scoop

Publift Scoop

Sign up to the Publift newsletter for all the latest news, updates and insights from across the industry.
Sign up