Zillow Group Inc (Z) 2018 Q3 法說會逐字稿

Good day, ladies and gentlemen, and welcome to the Zillow Group Q3 2018 Earnings Call.

(Operator Instructions) As a reminder, this conference call is being recorded.

I would now like to introduce your host for today's conference, Mr. RJ Jones, Vice President of Investor Relations.

Sir, you may begin.

Thank you.

Good afternoon, and welcome to Zillow Group's Third Quarter 2018 Financial Results Conference Call.

Joining me today to discuss our results is Zillow Group's Chief Executive Officer, Spencer Rascoff.

Interim CFO and Chief Accounting Officer, Jennifer Rock, is on parental leave and is not participating on today's call.

During the call, we will make forward-looking statements regarding future financial performance, operations and events.

Although we believe the expectations reflected in the forward-looking statements are reasonable, we cannot guarantee these results.

We caution you to consider the risk factors described in our SEC filings, which could cause actual results to differ materially from those in the forward-looking statements made on this call.

The date of this call is November 6, 2018, and forward-looking statements made today are based on assumptions as of this date.

We undertake no obligation to update these statements as a result of new information or future events, except as required by law.

This call is being broadcast on the Internet and is accessible through the Investor Relations section of Zillow Group's website.

A recording of the call will be available later today.

During the call, we will discuss GAAP and non-GAAP measures.

We encourage you to read our financial press -- financial results press release, which can be found on our Investor Relations website, as it contains important information about our GAAP and non-GAAP results, including reconciliation of non-GAAP financial measures.

In our remarks, the non-GAAP financial measure adjusted EBITDA is referred to as EBITDA, which excludes other income, depreciation and amortization expense, share-based compensation expense, impairment costs, acquisition-related costs, interest expense and income taxes.

We will open up the call with brief remarks, followed by a live Q&A.

We have published our detailed quarterly update letter on our Investor Relations website.

To save more time for Q&A, Spencer's remarks will only focus on highlights.

We encourage you to read the full letter with more detail.

In addition to taking questions from those dialed into the call, we will answer questions asked via Slido.

We encourage you to visit www.slido.com, where you may submit questions by entering the event code #ZEarnings.

On Slido, you may vote on which submitted questions you want us to answer.

This will ensure that we prioritize the questions that you consider most important.

You may begin submitting questions and voting now.

I will now turn the call over to Spencer.

Thanks, RJ.

Zillow Group reported third quarter 2018 total revenue of $343.1 million, which was in line with our guidance and up 22% year-over-year, driven primarily by growth in our Premier Agent, Rentals and New Construction marketplaces as well as our first quarter of revenue contribution from the Homes segment, which represents our Zillow Offers business.

On a consolidated basis, GAAP net loss for the quarter was approximately $0.5 million, and adjusted EBITDA was in line with our guidance at $66.2 million or 19% of revenue.

We ended the quarter with $1.6 billion in cash and investments on our balance sheet and added cash from operating activities of approximately $102.5 million on a year-to-date basis, which positions us well to fund our next stage of growth.

Zillow Group has entered a period of transformational innovation.

The residential real estate industry is also changing.

To maintain a competitive advantage in today's environment, we must continuously evolve our business as consumer expectations change.

When we launched 12 years ago, the ability to access real estate listings and home valuations online was revolutionary.

Today, consumers' expectations are shaped by the mobile on-demand world.

We can control many aspects of our lives: transportation, commerce, food delivery, car shopping, even dog walking with an end-to-end app on our phones.

Those digital experiences have shaped expectations for the real estate industry.

Consumers tell us that they want to control more of their home shopping experience and to do it from the palm of their hand.

To that end, we are creating new businesses and evolving our current businesses to give consumers what they want.

By prioritizing the consumer experience, we will create long-term value for our users, customers, shareholders and employees.

People come to Zillow Group's mobile apps and websites expecting excellent service and we're focused on ensuring that they get what they want 100% of the time.

In 2017, Zillow Group surveyed consumers who submitted a request for information on a for-sale listing from a Premier Agent to measure how much of a "push button, magic happens" experience we were delivering.

Consumers reported that they never heard back from a Premier Agent 49% of the time.

The Premier Agent business has been a fast-growing successful business for many years, so we knew that by making changes to meet or exceed consumers' evolving expectations would only add to the success of the business.

This survey identified a huge opportunity for us to improve both the consumer experience and generate more Premier Agent revenue.

To address these opportunities, we began rolling out our new lead validation and distribution process in April 2018 that first vets consumers before connecting them with a Premier Agent, and then significantly improves the connection rate by contacting the next available Premier Agent in an automated queue if the first does not answer the phone.

We are now delivering nearly instant service that consumers deserve.

In fact, home shoppers say that they are 3 to 4 times more likely to work with the agent they connect with in this improved process versus through the old system.

Our early estimates of the overall impact of the new lead validation and distribution system indicate that we may be able to support doubling of the number of transactions and the amount of commissions from the same level of consumer traffic.

However, we faced some challenges during the rollout of these changes.

Our third quarter Premier Agent revenue was lower than our guidance because of higher-than-expected advertiser churn.

Not all of the changes we made were well received by our advertisers.

First, we introduced these changes to our model just as agents began feeling a high amount of strain from 3 factors: cost-per-lead increases resulting from growing demand in the auction-based pricing model; pressure on home sales from a slowing market; and seasonality in the back half of the year when real estate slows.

Second, we prioritized quality of leads over quantity.

Regardless of lead quality, though, many agents tell us that they also value a higher quantity of leads.

Both the timing and effects of our changes drove higher advertiser churn, despite continued strong sales.

We believe the issues in the Premier Agent business are temporary and solvable, and in fact, we've already made changes to improve advertiser retention and drive acceptance of the new model.

First, we streamlined the way we validate leads in order to increase the number of connections from our system by reducing our screening questions.

Second, in the next few weeks, we will increase lead volumes further by returning to our old system of providing up-funnel leads to our Premier Agents.

These are leads which we had previously screened out of the immediate connection process that likely require more time before being ready to transact.

Third, we implemented auction pricing caps in certain ZIP Codes to help improve agent ROI.

We believe that Premier Agent churn will return to normal levels in early 2019.

Also, next year, we anticipate consumer connections received from the second half of 2018 will start converting into transactions.

That, plus the associated commissions and the seasonal uptick in traffic and consumer connections at the start of the year, will help improve agent receptivity.

At a high level, our Premier Agent business connects consumers with agents, and agents earn commissions from those connections.

The more commissions agents earn from these connections, the more Premier Agent revenue Zillow Group will earn.

Historically, we have provided annual backward-looking estimates of commissions earned from our connections.

Our last update was approximately $6.5 billion of commissions from our connections in 2017, prior to any of the changes that we've made over the last few months, which are designed to increase connections and commissions, and the results of which look good so far.

Zillow Group is moving further down the funnel and closer to the transaction during this period of transformational innovation.

In 2019, Zillow Group will be a larger business that is much more integrated into the consumer's entire home life cycle.

Our addressable market today is an order of magnitude larger than it was just a year ago.

We have the potential to transform the largest asset class of our economy with an end-to-end transaction.

Within that transaction, Zillow Group can be involved on multiple levels, from agent connections to buying and selling a home, to mortgage lending and the potential for even more adjacencies in the future.

Zillow Group's mission remains the same: to build the largest, most trusted and vibrant home-related marketplace in the world.

In 2018, we pledged to create better experiences for consumers as we accompanied them further down the funnel and closer to the transaction.

We also said that we would evolve our business models to better align our results with our industry partners.

These were big promises, and we kept them.

Consumers using our platforms deserve a better experience.

And now we're ensuring that, from searching listings all the way through closing on their dream home.

Our industry partners deserve high-quality validated leads.

The changes we're making to our Premier Agent program are already dramatically improving connection rates, and we're now shifting our focus to helping agents increase conversion of those connections into transactions.

Many homeowners want an easier way to sell their home.

The Zillow Offers service relieves them of the hassle and stress of selling the traditional way.

And some agents prefer flexible options for advertising on our platforms.

With Flex Pricing, we can attract more agents to advertise with us in a way that fits their marketing budget.

The future for Zillow Group is bright.

We are just getting started and are excited about what's next.

With that, I'll now open up the call to questions.

(Operator Instructions) Our first question comes from John Campbell with Stephens.

This is Carter taking the place of John.

Do you guys have any sense of the impact that expanding into Canada can have on traffic growth?

And can you keep the pace of kind of that run rate there for year-over-year growth for traffic growth?

Yes.

So 2 questions there.

First, on Canada.

Canada represents about 10% of the market opportunity of the U.S. in terms of, really, most major metrics.

For now, we're focused on improving the product quality in Canada.

We already have a lot of listings, but we'll be adding more and more listings over the next couple of months and then adding in more features so that in 2019, the Canadian product matches the experience that U.S. home shoppers have come to appreciate on Zillow.

And then at some point in 2019, we'll focus on monetization in Canada.

In terms of overall traffic growth, in the shareholder letter, we provided more detail on traffic growth.

I didn't go into detail in the remarks I just read.

But Zillow traffic was up, I think, 22% year-over-year on visits in September.

And Zillow traffic continues to grow nicely.

Trulia traffic continues to be a challenge, although it's worth noting that Trulia is basically maintaining its audience market share in the category despite us spending quite a bit less advertising the Trulia brand than we spend advertising the Zillow brand or than other competitors spend advertising their brand.

So Trulia is holding its own even with a much lower spend level.

Our next question comes from Jason Helfstein with Oppenheimer.

I guess, a question around your long-term outlook around Homes.

How do you think your outlook around Premier Agent, the macro housing environment and interest rates play into the outlook for the Homes segment as we're thinking about next year?

And knowing you're not going to give guidance, how does all that play out into that segment for next year?

Yes.

So thanks, Jason.

I mean, Homes is growing extremely quickly, and so it's very difficult to forecast something growing as quickly as Homes is.

Here's what we know so far.

We know that sellers are really responding to this method of selling their home.

I'll give you some additional data points on that.

I think on the last earnings call, I said that 15% of homes that -- of the home volume that's sold in Phoenix, we got a chance to bid on during that time period.

That 15% in June maps to a 25% in September and a 35% in October.

So specifically, what that math is, is the numerator is the dollar value that we were asked to bid on in Phoenix, and the denominator is the value of all homes that actually sold in October.

So it is very appealing to home sellers.

And to give you another sense of how quickly it's ramping, in the shareholder letter, I noted that in the third quarter, we bought 168 homes and we sold 36 homes.

Well, in October, the first month of Q4, we bought 130 homes and sold 32 homes.

So we're now doing in a month what we basically did in all of Q3.

And soon, we'll be doing in a week what we did in all of Q3.

So Zillow Offers is growing very quickly.

Obviously, I don't have 2019 numbers for it yet, and I don't have an answer to your question on long-term potential, except to say that I'm very excited about where it can go from here.

Can I have a follow-up?

Sure, go ahead, Jason.

Just your comment on this.

So there's some discussion that this market will be entirely just -- consumers will just choose on price.

Just any thoughts around price versus service for a product like this?

Sure.

I mean, there are competitors in this space, clearly, which is implied in your question.

Thus far, we have found the market appeal of this method of selling to be large enough that we haven't felt the impact of competitors in the space.

With respect to your specific question about price versus brand or service, anecdotally, we have a number of sellers that -- where we -- who have told us that when they have had offers from other companies, they've chosen our offer because they trust the Zillow brand.

But that's a very small sample size and anecdotal.

So I think it depends on how predominant a method of selling this becomes.

My expectation, hope, hypothesis is that it becomes a very appealing method of selling your home and that the market ends up supporting multiple companies in the space, including, of course, us.

Our next question comes from Tom White with D.A. Davidson.

Just another follow-up on Homes.

I think last quarter, you guys talked about kind of underestimating the number of days between contracting with a seller and the closing of that transaction.

Just curious whether you're seeing kind of any compression in that time spend now that there are, I guess, maybe some signs that the supply of for-sale homes is freeing up a bit.

And just on a related topic, just I guess I'm trying to understand kind of any trade-off there between the timing of your revenue recognition versus maybe the impact on your return from maybe holding the homes for longer, if that's the case.

What we discussed last quarter, I would describe as a forecasting error out of the gate when we didn't have any data on which to judge the delta between handshake and purchase of a home.

We updated you last quarter after having had a quarter of data on that with better information, and there's no change since then.

Every month that passes, we develop more knowledge about how the business looks, and we get a little bit better at predicting it.

RJ, anything to add on that, or...?

No, no.

And as far as like the question on revenue recognition, I mean, what we're seeing is the model working like we expected it to.

And things are tracking from a performance perspective very well.

I mean, we're very excited about how things are going.

In the market, as Spencer talked about, through demand and then outflows through to results.

Our next question comes from Brent Thill with Jefferies.

This is Alex Giaimo on for Brent.

Spencer, you mentioned higher-than-expected advertiser churn as a result of the changes to Premier Agent.

How fixable do you think those problems are near term?

And then maybe in the past when you've experienced user churn, how successful have you been in winning back those accounts?

And then quickly, is there any update on the future opportunity to monetize the Homes business via sell or lease?

Two good questions.

So we think the Premier Agent issues are very solvable and are already in the process of being solved.

The changes that I described for -- from what I'll call PA 4, were really 2 sets of changes.

The first was it qualified leads by asking consumers qualifying questions; and the second was it distributed the lead sequentially, so that if the first agent didn't answer, the second agent's phone rang.

And the -- what we really did was we shifted from a high volume of low-quality e-mail leads to a low volume of high-quality phone connections from an agent's perspective.

What we heard back from agents was that they were pleased with the low volume of high-quality phone connections, but they also wanted the high volume of low-quality e-mail leads.

And so PA 4.1 is what I would call the new changes to Premier Agent, which are underway right now.

And those are twofold: firstly, we're reducing some of the screening questions; and secondly, we're giving agents the consumer up-funnel leads who are not ready to transact yet.

And we're letting agents follow up on those consumers and incubate them and nurture those leads.

And that was based on our quick response to advertiser feedback over the last month or 2 as we started rolling out PA 4 in more places.

Only about 70% of the country is on PA 4, and so the next 30% will move straight to 4.1.

So they'll get the improvements of the real-time phone calls, but they'll never experience the drop in total number of leads that the PA 4.0 rollout encountered.

The consumer experience from all of this is very important to understand.

I gave you some data about the improved response rate from about 49% to around 100% from consumers being able to get agents on the phone.

I'll give you one other data point to try to dimensionalize why these PA 4 and 4.1 changes are so important.

Consumers that were contacting an agent through PA 3 gave us a 2.5 out of 5-star rating on their experience.

Consumers that go through PA 4 or PA 4.1 give us a 4.1-star rating out of 5. So that's a 60% improvement in consumer satisfaction with this new lead connection model.

That's huge.

That's why I know that the changes that we're making with PA 4 and PA 4.1 are good for the business, good for the company and good for the user.

To answer your second question about seller leads from Zillow Offers, we have just started in the last 2 or 3 weeks testing, monetizing seller leads through Zillow Offers.

We're now passing those leads to agents and brokerages on a success fee.

It's too early to share results because we just started.

We think there's a lot of potential in this business.

And to give you a data point or 2 to understand my interest in this, about 45% of consumers that go through the Zillow Offers funnel end up listing their home overall.

So about 20% of them end up listing in the first 2 months after asking for us to make an offer, and another 25% end up listing a couple months later.

So these are serious home sellers.

And if we can succeed in connecting those whose homes we don't buy with real estate agents that this should, over time, become a really nice revenue stream for Premier Agent that hangs off of Zillow Offers.

But we're only 2 or 3 weeks into it.

Operator, I think I'm going to go to Slido for the next couple questions.

And the first one is, why implement auction caps?

If the agent wants to spend more money, why restrict them?

So this refers to our market-based pricing system in Premier Agent.

The switch to market-based pricing was designed to create price discovery so that agents who wanted to spend more could bid up the cost per impression and cost per lead in a particular ZIP Code.

And it worked.

It drove price increases and even price decreases in low-demand ZIP Codes.

The reason to add pricing caps is to protect ROI.

Because without price caps, new agents come in, bid up the price in a certain ZIP Code above a profitable level.

And agents churn, and then we have to keep reselling the same inventory on sort of a hamster wheel or a treadmill replacing lost revenue.

And so it's much more effective to cap pricing in certain high-demand ZIP Codes, protecting agents' ROI and focus our selling efforts on other areas where we can improve monetization like under-monetized ZIP Codes with higher ROI which haven't experienced auction dynamics.

So that's the explanation as to why we implement auction caps.

The next question from Slido was Opendoor is launching a new agent partnership for its listings.

Redfin employs its agents full time.

Does it make sense for ZG to have some agent presence?

Does an acquisition by ZG of an asset-light agent business like RE/MAX or EXPI make sense?

I think we have a terrific model right now with advertising Premier Agents.

For us, agents are a revenue line, not an expense line.

Tens of thousands of Premier Agents pay us rather than us paying thousands of employee agents or tens of thousands of independent contractor agents.

So I liked our -- I like our model the way it is.

Should I keep going on Slido, RJ, or back to you...

I can -- there's one point I'd like to clarify.

So on an earlier question, Spencer, you were asked about traffic and you cited our revenue growth number.

The traffic growth number is 7% year-over-year, and growth in visits is 14% year-over-year.

Our next question comes from Matthew Brooks with Macquarie.

I just wanted to ask, is there a reason why home purchases are still accelerating in Phoenix, but it seems like Vegas is ramping more slowly?

Sure.

Each market is different.

We're seeing a faster ramp in certain markets than others.

Some of it has to do with the size of the market.

Some of it has to do with receptivity to the offers.

Some of it is self-imposed, where we control the throughput of our funnel by increasing and decreasing seller demand by different levels of site merchandising to keep our assembly line of inspections, renovations and resale flowing appropriately.

So that's why you see different paces of different markets.

RJ, I want to clarify your clarification.

I know I wasn't crazy.

So in the shareholder letter, it says visits on our flagship Zillow brand were up nearly 22% year-over-year in September.

That was what I was referencing.

So I was right.

Apologies.

Next question comes from Ron Josey with JMP Securities.

Just a few follow-up questions on PA 4, Spencer, an easier one maybe.

For those agents that churned off, can you talk about the type of agent?

Are these the newer ones that perhaps you're bidding up in the auction or longer term in general?

And then also from a PA 4 question, just the timing and understand 4.1 and the offer, the option to be nurturing versus qualified.

But as we head into the all-important, call it, the first half of the year, how do you think -- does this change anything in terms of the outlook of how Zillow is prepared into the home-buying season?

And then second question, just on Flex Pricing, just bigger picture.

You talked about seller leads going into sort of Flex Pricing benefit as well.

It was success-based, but then you also did Premier Broker launch on Flex Pricing.

So can you talk about the progress there and you talked about expanding that later in new markets here.

So very curious on Flex Pricing.

Sure.

Thanks, Ron.

So the changes that we made to Premier Agent with 4.0 and 4.1, although painful this quarter, and you see it in our guidance, position us very, very well for 2019 and the buying season because we have just solved the leaky bucket of our lead connection rate from a 49%, I think, it was response rate to nearly 100% response rate.

And that -- and we've improved the consumer satisfaction of the experience dramatically.

So that, to me, is the big story here, which I know in the handwringing around Premier Agent revenue in Q3 and Q4, might get lost.

But for me, it's hard to lose sight of the importance of that.

In terms of giving you a little bit more clarity on advertiser responsiveness to PA 4 and 4.1, I'll give you an example.

The average Premier Agent was getting around 10 leads a month through PA 3, 10 e-mail leads a month.

So if you're an agent in PA 3, you're getting 10 e-mails into your inbox a month.

With PA 4, with our screening questions and our -- and the other lead validation system, you switch from 10 e-mail leads a month to 4 live phone connections a month.

With PA 4.1, where we changed our screening questions and we're giving them back the up-funnel leads, it goes to 4 phone connections plus another 4 up-funnel leads a month, e-mail leads.

So you're back up to about 8. That's 4 phone calls and 4 up-funnel leads.

So to your question of what type of people were churning, this is an average Premier Agent.

Imagine if you're an average Premier Agent and you were used to 10 e-mail leads and you go down to 4 or fewer phone connections, and your phone is only ringing once a month -- sorry, once a week, it'd be easy for you to conclude that your ad spend was not being effective, even though, mathematically speaking, you're probably converting the same or more commission dollars at the bottom of your funnel.

But you just have fewer at-bats.

And so returning them to a higher number of at-bats, 8 in this example, which is for an average Premier Agent, is important to improve receptivity.

And in terms of what types of agents were reacting negatively, you had high-spending agents that had built up teams that had -- did a good job of lead conversion on those up-funnel leads.

And they were the most vocal advertisers who said, Give us those up-funnel leads back; I can convert them.

We actually like the change to the phone calls, but I also want the up-funnel e-mail leads.

And my inside sales associates will nurture those leads, and I'll be able to convert more of them.

So we reacted quickly to that feedback, and the changes that we made to 4.1, we think, keep the consumer benefits of PA 4, but improve the advertiser receptivity.

Your second question, Ron, was about Flex Pricing and Premier Broker.

So Flex Pricing allows an agent to pay us in arrears on a success basis once their leads close.

We're testing it right now in Florida, and we're rolling out 3 other regions.

We're testing it in under-monetized ZIP Codes where the ROI is very high.

And we're bringing on new agents.

On a success fee we'll improve monetization from our perspective and consumer experience from a consumer's perspective.

The receptivity on the part of agents and brokers has been excellent.

Why wouldn't it be?

There is really no reason for one not to want to participate.

There's no upfront costs, and they only pay us if the leads convert.

And so as you'd expect, agents have been extremely receptive to it.

But it's a very -- we've only just started testing it the last couple of weeks, so it's too early for us to assess the impact or what -- how it might play into 2019.

Our next question comes from Lloyd Walmsley with Deutsche Bank.

I have 2 questions, if I can.

Just first, on the PA revenue and kind of the impact on the IMT segment revenue and kind of margins, how should we think about, I guess, a time line of recovery in how you guys approach OpEx, and how we should think about margins in that segment as you go through the transition period?

And second question is really just, I guess, you're going to go through some agony, it sounds like, with PA 4 and 4.1.

Ultimately, you seem to be migrating more towards something like the Flex Pricing model anyway.

Why not just go straight to Flex Pricing?

Any thought there on moving straight to that or accelerating a move towards that nationally?

Sure.

Well, one of the reasons not to go straight to Flex Pricing is I would like to see transaction -- leads go through that system and us get paid at closing for at least a couple of months so we can have good data on lead conversion, on leakage, on breakage, et cetera.

So it's just -- it would be too big a risk, even for us, to migrate to something that radical that quickly without prudent testing.

The second reason is there'll be some areas where the auction-based pricing monetizes better than a success fee would.

And so I'd be reluctant to just flip the whole country over to Flex Pricing because of monetization differences between the 2 models.

But let's talk again next quarter once we start getting more data on Flex Pricing, agent receptivity and lead conversion.

In terms of OpEx and margin on IMT, we're deep in the midst of the budget process right now for 2019.

The company is really 3 different businesses now.

There's IMT, which is predominantly Premier Agent, but it's also Rentals, New Construction, Display, dotloop, New York City.

There's also -- that's number one.

Number two, there's the mortgage business, which is really 2 businesses.

There's the lead gen advertising business, and then now there's the mortgage origination business.

And then the third business is Zillow Offers.

And each of these 3 businesses have different near-term and long-term margin profiles, radically different TAMs and different growth rates.

As we go through the 2019 planning process and as we give you 2019 outlooks a quarter from now, we'll -- my goal is to give you better visibility into long-term TAM and long-term margin profiles of each of those 3 businesses.

I'll be doing that with our new CFO, who will want to be deeply involved in that also.

And I guess, the only other point worth mentioning on OpEx is we have -- we really -- we moved a lot of people from IMT into the Homes business to seed the Homes business over the last 6 months and a couple -- well, at least 100, maybe more, people on the product side.

And we then backfill those people with new employees in IMT.

And that drives OpEx and recruiting costs and wage costs because of it being such a competitive job market.

And so that's elaborating on it a little bit further in the shareholder letter.

Should I go back to Slido?

Okay.

So I'll go back to Slido for the next 1 or 2 questions.

Let's see.

Help us understand why the Homes business won't just be a capital war.

It's a very dramatic question.

I guess, it's sort of similar to the question that I answered previously, which is nobody yet knows whether 2% of homes will sell this way, 5% of homes will sell this way, 20% of homes will sell this way, 40% of homes will sell this way.

I think we've put in the shareholder letter that with our current buy box such as it is in terms of home vintage, square footage, price point, et cetera, that if we were in 200 metro areas, the addressable market would be about 2.75 million existing homes sold.

In other words, about half of all homes in the country that sell every year would be in our buy box if we were in 200 cities.

So if selling your home in this way becomes widely appealing, then I think there will be multiple winners.

I think our -- we have some serious advantages that we bring to this space, including our brand, including our very low seller acquisition costs because homeowners come to our sites to find out what their house is worth, including our analytics and our ability to merchandise these homes so we can sell them quickly.

So we have some advantages, but it's not yet clear to me that it's going to be a fight to the death in any case, because I think it's going to end up being pretty broadly appealing and have multiple winners.

Let's see.

Another question from Slido.

I think we answered that one, RJ?

Yes.

Okay.

Operator, we'll go back to the call, please.

All right.

Our next question comes from Maria Ripps with Canaccord Genuity.

Spencer, broadly, without getting into specifics, how do you view Premier Agent revenue growth next year?

And I guess, what are some levers, besides re-doing PA 4, that you can pull to accelerate growth from what you're targeting for Q4?

And then just on the high pricing from the auction-based model, any color you could share around how high prices are going?

And what types of markets are seeing this the most?

I'm sorry, what was just the very last part of your question?

Could you restate that?

Yes.

I was just wondering -- I was just asking about high prices.

And any more color you could share around how high prices are getting?

And what markets are you seeing this the most?

Okay.

Sure, I'll answer that one first.

We've seen in some high-demand ZIP Codes, 40, 50, 60 agents bidding on impressions, cost per lead getting between $300, $400, $500 or more dollars per lead.

And the ROI, assuming some lead conversion rate, which is always a difficult assumption or a difficult estimate, the ROI getting below 1x.

In other words, the top bidders likely losing money on a per-unit basis.

Now there may be reasons why it will be economical to bid to that level if they attach a lifetime value to that transaction, if they think that, that consumer will then refer them other business in the future, et cetera.

But what we see from a selling standpoint, when ROI gets that low and pricing gets that high is agents churn.

And in that example, if there are 50 agents bidding on that ZIP, if 5 to 10 churn because they say, well, the auction has gotten too hot, then we have to go resell that or else we start losing revenue in that ZIP Code.

It would be far wiser to cap the ROI at a reasonable level and then focus our selling efforts elsewhere rather than continually reselling the same inventory.

Your question about Premier Agent revenue growth and levers, I'll point you back to the total commission dollars generated, the estimate that we produced in 2017 of around $6.5 billion.

That was with PA 3, with around half of our consumers not getting an agent when they sent an e-mail through Zillow or Trulia saying, I want to see this house or I have a question about this house.

Half of them were not getting responded to.

So if you take a giant step back, I think improving that response rate and making it much more realtime should allow us to generate a lot more commissions.

Now that is about recapturing the revenue available to us from the 4 -- the PA 4.1 changes, which now we have to go and do.

And we have to convince advertisers to accept the changes because change is hard and these advertisers are very fragmented, so it's difficult and complex for us to communicate our messages out to tens of thousands of them.

But as we go into 2019, I'm confident that we're going to be generating a lot more commission dollars in 2019 than we did in 2017, and that should position us well for Premier Agent revenue growth.

The other levers that we can pull are listing lead generation off of Zillow Offers, which we've discussed; an alternative payment model, Flex Pricing, which we've discussed; potentially, feature listing products, which we've discussed in the past and which remains interesting to us, although not as high priority as some of the other things that I've already discussed.

Our next question comes from Deepak Mathivanan with Barclays.

This is Mario Lu on for Deepak.

To the extent that you could, can you provide any thoughts on EBITDA for next year?

As you see the churn trend stabilize in the first half, should we expect profitability to improve as well?

And then related to that, what is driving higher hiring and retention costs incrementally over the last few months, specifically from 2Q to 3Q?

Is this something you expect to continue at elevated levels?

This is RJ.

I'll talk a little bit about the second part of your question.

Spencer can also chime in on that.

We're just seeing, in general, when you look at acquiring, retaining new tech talent, especially when you're a growth company growing into the opportunities that we're in, it's very competitive to bring on talented people given the choices that they have in the markets that we're in, especially Seattle, Phoenix, San Francisco, you name it.

There is tremendous amounts of opportunities for very talented tech people.

What that does from a cost perspective is you look at bringing them on at market rates that are being driven by the demand for that talent.

So that's one part of it.

The other part is that we've been successful in winning a lot of offers as far as new people coming on board.

So that's been coming in ahead of plan, both in the second and the third quarter, and we would anticipate that to continue into the fourth quarter.

The other question about EBITDA into 2019, looking at what Spencer talked through revenue, EBITDA, from one standpoint, will be impacted by what we're looking at in terms of revenue for next year.

That's one aspect that we've talked through.

The other is our opportunities to invest in the businesses, and that's another thing that we're going to talk about when we share with you our thoughts on the 2019 plan.

So as far as like specifics on what the 2019 outlook could be, which we're not going to get into today, the factors that have been there all along when we look at how do we plan and invest in the business are up to us.

That includes what we'll look at in terms of advertising into 2019, leaning into the opportunity for Zillow Offers and what that can mean as far as growth.

So there's a lot of things for us to consider given the exciting opportunities that are in front of us.

Spencer, anything else there as far as...

Those are -- the 3 big levers on margin for next year are revenue, headcount and ad expense, yes.

So we're deep in the budgeting process right now determining all 3.

Our next question comes from Brian Nowak with Morgan Stanley.

I'm just trying to maybe square the round peg a little bit.

In the past, I know you've talked about agent ROI somewhere in the high single, low double-digit range.

How do I think about that now at this point if agents were missing about half the leads?

You're talking about ROI sort of 1x.

So where are you now on an agent ROI perspective?

And then Spencer, when you talk about having confidence that you can double the number of transactions, can you just talk to some of the math that you're doing behind that to kind of give us the confidence you can double the transactions going to the platform?

Sure.

So the math on doubling transactions, it starts with connection rate.

It -- a lot of leads don't get responded to.

Now some of those could -- under PA 3. Now some of those consumers are double submitting, and we take that out -- right, so they'll submit, they don't hear anything back and then they go and submit on a second house, and then they do get connected with an agent in that example.

So we're taking those out from our math on doubling the likely total commission dollars.

We're also taking into account the consumer's self-reported likelihood of completing a transaction with the agent.

So I think in the prepared remarks, I said -- what was it, RJ?

Four times more likely.

Four times more likely through PA 4 than through PA 3. And the reason for that is that in PA 3, the consumer sends an e-mail, sits there and waits and hopes that the agent gets back to them.

And half the time, they do.

And when they do, it's usually, I don't know, a couple of hours later.

Whereas in PA 4 or 4.1, they're getting on the phone with an agent instantaneously.

And so of course, they're going to be much more likely, self-reported 4 times more likely, to transact with that agent because they're getting immediate service versus PA 3. So combination of those things causes us to estimate that we think it generates twice the number of transactions on the bottom of the funnel.

In terms of total ROI, we know that, as I said, there are some ZIP Codes where a lot of the ROI has been competed away by the auction.

We know there are still a lot of other ZIP Codes where the ROI is very large.

We know that ROI is also highly dependent upon the individual agents and their own lead conversion.

I don't feel comfortable giving a systemwide ROI estimate right now because I'd be kind of -- it wouldn't be supported by enough math to give me confidence in giving it to you.

If -- let me try to put it from an individual advertiser's perspective, right?

So in the 10 e-mail lead model of PA 3 versus the 4 phone contact model of PA 4 and then compare that with the 4 in phone contact and 4 e-mail contact of PA 4.1, we think that will -- that 4.1 will create a lot more commission dollars than 4.0 or 3.0.

We need a couple of months to bear that out.

What we know so far is that the advertiser response to 4.1 has been excellent.

We announced it on stage at our Las Vegas forum event last week with over 2,000 Premier Agents, record attendance at our annual events.

And we received a huge ovation for those changes in 4.1.

We rolled it out at 3 -- or we pre-rolled it out at 3 other Premier Agent events in Chicago, Los Angeles and Washington, D.C. 2 or 3 weeks ago to about 500 total agents.

And again, received huge ovations for those changes.

So I feel good going into these changes of 4.1 that we've struck the right balance between the consumer benefit of the immediate response and the advertiser benefit of a high volume of leads and the immediate lead generation of the phone model of PA 4 and 4.1.

Our next question comes from Ygal Arounian with Wedbush Securities.

I guess I want to follow up on the buy box for Homes.

You talk about expanding it to the 200 largest metro areas in the U.S. And I want to ask how you think about how translatable Zillow Offers is in larger markets that are less homogeneous, where you could have really varied pricing even within like the same building, let alone the same ZIP Code.

So that was one question I had.

And then, I guess, I'll also ask about rental users, which was flat this quarter.

And last quarter, you talked about how it came in lower than expected and trying to drive some things to -- or trying to do some things like more SEO and SEM to drive users to the top of the funnel, while you were working on creating more opportunities in the lower end of the funnel.

And some of the things you implemented and if -- what didn't work and what did and your plans going forward there.

Sure.

So just to start with the buy box question.

We think Zillow Offers works best in areas with homogenous housing stock, as you point out.

These are places like Phoenix and Las Vegas, Atlanta, Houston, which is a market that we announced yesterday or today.

That math of being in the top 200 cities with our current buy box, meaning that about half of all home sales would be available to us, that takes that into account.

In other words, in that 200-city estimate, that would include New York City, for example.

And in New York City, if our -- our buy box to date has been kind of in and around the median home value and under 5-ish hundred thousand dollars.

And so we wouldn't be buying very many homes in Manhattan, probably no homes in Manhattan.

And so that would sort of still be included in that estimate, even though Zillow Offers probably will never work very well in Manhattan.

One of the goals over the next couple of months is to learn more as we roll into new markets to learn more about where Zillow Offers works best in terms of seller acquisition, in terms of quick renovation and relist and resale.

And I actually welcome us moving into some markets with slower home value appreciation, cooler housing markets so we can get more information and insights into how Zillow Offers will work in a slowing market.

Our hypothesis is that Zillow Offers will work even better in a slower market than in a hot market because the certainty and hassle avoidance that our selling process provides to a home seller is relatively more attractive when she has fewer alternatives to selling her home conventionally.

And as long as we adjust our offer prices and our fees accordingly, we'll still be able to build a very meaningful and profitable business even in those slower areas.

So diversity of markets is something that we will benefit from as we roll out more cities over the next couple of months.

So then regarding Rentals traffic, Rentals traffic, I think I put in the shareholder letter, was approximately flat.

Some of the SEO and traffic and e-mail initiatives that we talked about last quarter, we're still working on.

The building products that -- the products that we've launched around going deeper funnel had been very well received.

This is digitally scheduling a viewing of an apartment, digitally applying, digitally paying your rent, and soon, digitally signing a lease and something that we're rolling out state by state over the next couple of months.

So I'm very pleased with our adoption of the digitization of the rental transaction.

And just remember that we're moving into the slow rental season now in terms of traffic where, of course, new leases peak over the summer and in the fall, and first quarter lease origination is much smaller.

Our next question comes from Tom Champion with Cowen.

I think last quarter, about 25% of your footprint was converted to the new lead system.

And presumably you're beyond the worst of the churn there.

I'm assuming leads are down in those markets year-over-year, but just curious how conversions are tracking.

I am wondering if you could comment on that.

And then second, perhaps you could talk a little bit about the integration of MLOA.

I know the deal only closed just recently.

But can you help us think about how that will integrate into the Homes segment?

And maybe any thoughts on the EBITDA margin profile of that revenue?

Sure.

So yes, last time we spoke, we were about 1/4 of the way rolled out with PA 4. Today, we're about 70% of the way rolled out with PA 4. And then there are 2 changes in 4.1.

One is reducing the screening questions.

That is already complete, so all 70% of the country that is on PA 4 is, in that regard, is on that aspect of 4.1.

The second aspect of 4.1 is giving agents back those up-funnel leads.

These are consumers that say they're not ready to connect with an agent, but they will be soon.

We're doing that in the next month -- sorry, the next week or 2. And so within the next 2-ish weeks, 70% of the country will be on 4.1, and then we'll move the last 30% of the country before the end of the year from PA 3.0 to 4.1.

In terms of impact on lead conversion, I don't have data on that right now.

I do know that the agent receptivity to 4.1 has been very strong and that the switch from 3.0 to 4.0, the timing wasn't great because it passed peak seasonality.

And so their lead volume declined naturally at the same time that we also further reduced lead volume.

But it was the right thing to do.

We follow the consumer.

We know that this is a much better consumer experience.

And as soon as we saw that [FIFACT] data about how much more consumers preferred the immediate connectivity to an agent, we knew that it was the right thing to do to move to 4.0 as quickly as we could.

Now we think with 4.1 that we have a product that works with both consumers and for agents.

Regarding MLOA, the deal closed ahead of schedule, a huge thanks to our team that worked incredibly hard to get a lot of licensing documents completed rapidly.

We had to change mortgage licensing on a state-by-state basis post-acquisition, and so were able to close ahead of schedule, which was no small feat.

We will rebrand MLOA in the coming months, and the focus now in 2019 for MLOA will be to drive mortgage attach for our Zillow Offers business.

There are 2 opportunities for mortgage attach for Zillow Offers.

One is when we buy a home from somebody and then they go off and look for their next home, we want to originate their mortgage on their next purchase.

And the second is when we sell a home that Zillow Offers owns, we want to originate a mortgage for somebody buying a home from us.

And the focus of MLOA will be driving mortgage attach and supporting the Zillow Offers business in those 2 ways in 2019.

RJ, regarding EBITDA margin on MLOA?

Yes, there's going to be a difference between what MLOA's EBITDA margin profile was before coming into Zillow Group because we're going to be operating the business from a perspective of how does it support Zillow Offers, but also how do we invest in product to create innovative, new experiences for consumers and also to support what Zillow Offers can look like as we add or take away friction in the process and drive more speed and optimization into Zillow Offers.

So as far as that looks, we'll probably have to share more as far as our outlook when we look at 2019 and discuss that with you in February.

But the pre-acquisition profile of that business will be different as it's part of Zillow Group on a go-forward basis.

Our next question comes from Nat Schindler with Bank of America.

Two questions.

One, you mentioned in the note that there was -- a part of your guidance is being pulled down in Premier Agent from a little bit of weakness in the macro environment.

Can you help kind of quantify the difference between the churn?

Is it churn effect of the prequalification versus the macro environment and how you imagine the macro environment progressing from here and affecting your business?

Secondly, historically, you have said that your ROIs are in the 8x range.

And if that's true, why would there ever be a time that you need to hold down pricing to improve the ROI of your advertiser, if they're already getting such high average ROIs?

Sure.

Thanks, Nat.

So just on the ROI question, I think we've given a lower estimate than 8x more recently.

I think 8x was a couple of -- I don't know...

The comparison of that is the $6.5 billion estimated commissions to Premier Agent revenue last year, which is, top of my head, $768 million in revenue, which gets you that gross level.

What Spencer was talking about is that in different markets, in very, very active markets, you can see ROIs drop below 1. And so what we're talking about is something like when we launched market-based pricing, it's actually just managing it for the health of the auction and the health in terms of selling not only in the auction, but in the ZIP Codes around those really, really in-demand ZIP Codes.

And then the macro question.

It's very hard to disaggregate macro impact from changes that we brought to the market from moving from PA 3 to PA 4. If you think of the example of the average Premier Agent going from 8 e-mail contacts to 4 phone calls, the -- if macro slowed down in their particular city kind of around that same time and lead volume slowed a little bit and a homebuyer that they were working with started to get a little bit more cold feet and maybe said, "Hey, let's take it slow.

Let's maybe take 3, 4, 5 months to find a house that's right for us because it looks like the market is slowing and sellers might be willing to cut their price more." That started happening kind of around the same time that we reduced lead volume by switching from PA 3 to PA 4. So hard to know how much of it was macro and how much of it is self-induced.

What we do know is the switch to 4.1 gives them back a lot more leads.

So they just get more at-bats.

They get the phone calls, but then they also get the up-funnel leads and the lower -- the less qualified leads.

And so we think that helps restore the balance that is necessary for an advertiser to feel benefit from the PA 4 benefits, but also the higher lead volume from the older model.

Okay.

I think we're going to wrap.

We're at the hour.

Thank you for joining us.

We know that the changes we're making to the Premier Agent model are the right ones because they're improving consumer satisfaction.

Over the long term, I think changes are going to improve agent productivity and generate additional incremental transactions and commissions.

I will talk to you in February when we give our next quarterly update.

I will have with me on the call our new CFO, who I could not be more delighted, who will join us by Amazon in the next couple weeks, and I look forward to you -- to meeting him as well.

Thank you for your time.

Talk to you soon.

Ladies and gentlemen, thank you for your participation in today's conference.

This does conclude your program, and you may all disconnect.

Everyone, have a great day.