How does Zillow’s accuracy rate work? Let’s talk about two important points. I’m not writing because I’m angry, but only because the public places enormous trust in this brand, and I see this accuracy rate quoted everywhere. Well, let’s at least know the fine print so we can have informed conversations.
Two quick points on Zillow’s accuracy rate:
1) Median makes all the difference: Zillow says their median error rate is 5% for the United States. That sounds impressive, but we have to realize this is a MEDIAN error rate, which means the Zestimate is within 5% of the purchase price only half the time. So Zillow’s slogan could be, “We’re within 5% half the time.” But that doesn’t sound as polished as “Our median error rate is 5%”.
2) Zillow uses the most recent Zestimate: This is such an important point – even if it feels a little technical. I have an example to help show what I mean too. Zillow states “Zestimate accuracy is computed by comparing the final sale price to the Zestimate on or before the sale date.” Let me show you why this matters.
An example: Imagine a property listed for $380,000 and the Zestimate was also $380,000. But then after a series of price reductions, the home sold for $350,000. That would be 8 percent lower than the original Zestimate. That’s not very good, right? Well, now imagine as the list price was reduced, so was the Zestimate, and the most recent Zestimate was $353,000, which is only 1 percent higher than what the property sold for. Here’s the kicker: The home may have sold for 8 percent below the original Zestimate, but for its accuracy rate Zillow can claim 1 percent because it uses the most recent Zestimate instead of the original one.
In short, the most accurate stat here would be 8%. After all, the home literally sold 8% lower than what Zillow said it was worth. Thus when Zillow uses the most recent Zestimate instead, it’s a clever way to get much sexier stats.
One more thing.
Chasing the List Price & Fixer Upper TV show: What sort of effect does the list price have on the Zestimate? In other words, does the list price actually change Zillow’s value? This might not happen in all cases, but here’s a glaring example of the “shotgun” house from the TV show Fixer Upper. It’s a one-bedroom that was listed for $950,000 and didn’t end up selling.
In this example the list price was outrageously high, but Zillow’s algorithm looks to have simply matched the Zestimate to the list price. This might not happen in all cases, but it certainly did here.
I hope that was interesting or helpful.
Article I wrote: Here’s a piece I wrote about online valuation sites for Comstock’s Magazine last month.
CLASS I’M TEACHING: I’m doing my favorite class at SAR on May 28th from 9am-12pm called “How to Think Like an Appraiser.” Sign up here.
Questions: What do you think? Anything else to add?
If you liked this post, subscribe by email (or RSS). Thanks for being here.
A house I am selling had a zestimate of 151k right before it was listed on zillow at 250k then the zestimate went to 255k. its been in contract for an extended time for several reasons and i think due to teh time in contract zillow is bringing their zestimate down to anticiapte a lower sale price so they have good stats.
its going to close for 200k… follow the zestimate logic.
address;
901 w 12th ave chico,ca
BTW-at one point in history-the house next door at 897 was worth less according to zillow even though it wasnt a house-it is a triplex.
I also have seen zillow be over 300k on 901 and showing worth over 100k more than 901 but its not showing in the history-I think they go back and edit historical data. I distinctly remember the MH at 901 being worth more than the triplex at 897.
yep-901 is a MH in the middle of town on a LARGE lot.
Thank you Mark. I appreciate your commentary very much. Thank you for posting the address too so we can check in on this property from time to time.
Oh boy, you’re opening a can of worms to the commenters here! Lol. No, seriously, thank you for posting this, and I truly hope it goes out far and wide. The public need to understand Zillow’s misleading advertising. Like you, I’m not mad at Zillow. Zillow serves a purpose to the general public. But Zillow should NEVER be taken by any party as being anything other than what it is – computer generated algorithms based on paper data and statistics. It is under no circumstances a substitute for a real valuation aka Appraisal.
One more thing – have you noticed how the infamous Zestimate changes based on a property’s listing? Property A’s Zestimate could be $469,321 one day, and the very next day it could be raised to $569,898 because the homeowner listed it for sale for $569,900 and Zillow recognized the listing. Then when it sells in 90 days at $549,500 the Zestimate might change again to $550,000. I always tell people, you get what you pay for with Zillow. It’s not a bad starting point, but it’s also nothing more than that – a starting point. Go ahead and look at it, but take the information with a grain of salt. Because tomorrow it could be very different, either up or down. Lol.
Thanks Jacob. I appreciate your thoughts. That’s a common scenario. I’ve heard many real estate agents tell me the same thing. I have a fun post coming out eventually to talk about this dynamic. 🙂
I think understanding how the error rate works is really important. But I’m also realistic that the public might not really care. When a brand is trusted, the details don’t matter so much. But in this case the details absolutely matter.
You’re right, the details DO matter in an industry that has direct influence on the overall US economy – the real estate industry. When PIW or appraisal waivers are being issued on homes that were listed on Zillow based on a Zestimate, that’s a problem. When loan officers are using Zillow as one of their CMA resources for LTV and risk assessment, that’s a problem. Like I said, Zillow has its place in this world, but that place should be in advertising and NOT in the world of lending, valuation, or risk assessment.
I hear you. The big problem is Zillow calls itself a starting place or “ballpark”, but that’s not the way the public tends to see it. The public sees the number and sincerely struggles to budge from it in many cases.
Zillow has mastered advertising, but clearly they are thinking way beyond this. It’s only a matter of time before we see a local Zillow brokerage show up in many markets across the country. Then Zillow will have a local broker’s license and access to all the listings. It’s been understandable but alarming to see their business plan unfold. At some point I’m concerned about data integrity too if they start listing too many sales on their site and I am not able to verify details….
Ryan, I think the public does and would care if they were aware to the extent that housing prices have had the potential to be manipulated… they just don’t examine these details and also expect those representing them to be acting in their best interest.
The biggest problem here is the influence this has on market expectations. There are two sides to this equation, 1. Zillow’s responsibility to keep their algorithm unbiased from an agent’s list price and reductions… 2. The agents involved to represent their buyers & sellers appropriately by completing their own CMA… It should be demanded by ALL sellers and buyers! A good agent will do the work rather than defer to Zillow.
In the case of Zillow’s responsibility – by changing their Zestimate solely due to list price or reduction, they reinforce the agent’s opinion of value regardless of alignment to trend data. Unlike Redfin, agent’s do not have the ability to have the AVM estimate removed from their listing if the agent feels it isn’t accurate or “serves their listing well.” Instead, poof… you get a Zestimate that now reinforces the list price or at least close to it. From a buyer’s perspective this is extremely misleading.
Additionally, I wonder how many millions of dollars home owners have lost, who were represented by agents who listed their home at the high end simply because of the Zestimate? Only to result in a series of price reductions and most likely a final sales price lower than what could have been achieved if properly listed in the first place…
Thanks Tony. I appreciate hearing your take. I linked to a post above where I explain some huge issues with the way Zillow’s accuracy rate works. But unfortunately when we start talking about something technical, people can tune it. This is a paramount issue, but it’s not easy to have conversations with the public in mass when we’re talking about technicalities – even if they are crucial. So the struggle is real. The irony is in the meantime a tech brand is building their empire by manipulating numbers.
The property value is a problem, but I see a bigger problem with their agents and the rating they are given. I had three instances dealing with high rated agents that were incompetent. I wanted to give them a low rating on Zillow but another Realtor can’t do that because it is against their rules. In two cases I was the owner of the sold property but my opinion (along with complete documented evidence) wasn’t acceptable. Buyer beware.
The market capitalization of Zillow (ZG) is at $6.8Billion with over a $Billion in revenues. Where does that money come from? I’m guessing poor agents pay a good part of that revenue.
Thanks Paul. It just goes to show vendors can pay for a premium spot, but they may not be premier. It’s the way the world works today, and certainly how it works on Zillow’s website.
By the way, I wanted to also comment from a different capacity as a Realtor. Having been a Realtor (and Appraiser with both licenses) for 15+ years, and having been under a high-volume Broker (has won many awards in both Arizona and California), I have been trained by said Broker that Zillow is nothing more than a lead generator. Realtors and Brokers do not care about how accurate the Zestimate is (or isn’t), or whether “listings” on Zillow are actually a real representation of the market – often “listings” on Zillow are in reality expired or sold and no longer available, but they generate leads. To brokers and realtors, Zillow is nothing more than a lead generator and that’s it’s primary purpose (for now, until like Ryan said they become their own brokerage…) So to confuse this with an actual semi-reliable valuation firm could be a fatal mistake to the economy. People must be smarter, and by people I mean the public, market participants, appraisers, realtors and brokers, and perhaps most importantly the policy makers.
Thanks Jacob. I appreciate your Realtor take too. 🙂
I’m reminded of big data and how we can so easily trust big companies without really thinking critically. It’s just how the world works today. But nobody gets a pass. Companies like Zillow absolutely serve a purpose in today’s world. And frankly their AVMs are likely only going to get better over time. Yet we still need to consider things I wrote about in this post and many other aspects of what is happening.
Hi Ryan,
Dan here in SoCal. You may recall we had a conversation on this very topic a year ago on this very subject while I was adding a HELCO and the lender was first using a Zillow-type of AVM service. We ended up actually paying for a ‘real’ appraisal. The AVM value was some 25% below the actual appraised value. There was no relationship between the comps used in the AVM vs the real thing. Your blog may want to consider a total debunk of AVM by studying the value relationship between AVM and traditional appraisal. My experience was frustrating because of several factors that introduce errors into AVM including; 1) The last sale was the raw land as we built, 2) the county records were incorrect regarding size by a substantial amount (understated by 800 sqft for a 3800 sqft house), 3) does not account for most amenities like pool, solar, well, condition, upgrades, features, etc. Maybe my situation is a one off but the October 2017 real appraisal was $1.15M and today Zillow shows $868,262. That’s a 24.5% differential and not realistic for much. Cheers, Dan
Hi Dan. Thanks for the follow-up. I remember talking about this. I’m so glad you got things sorted out too. This just goes to show in valuation we are only as good as the accuracy of our data and our understanding of how it all fits together. It’s crazy to me that the AVM would throw in the land sale as a comp though. I get considering the value of the land, but it shouldn’t be used as a comp. Thanks for the post idea.
I agree! I found this out personally when i sold my house last year. I built it myself and so there was no starting point for Zillow. My Zestimate was $175k before i listed it. Granted it we live in a very rural area and that messes with the algorithms. I put the house on the market at $365k and shortly after my Zestimate changed to $365K. We lowered the price to $345K and low and behold my Zestimate followed suit. It finally sold for $320 and you can guess what the Zestimate ended up as. Zillow is useless. I should have listed for $1.5M wonder were my zestimate would have gone? Shoot for the stars! : )
Thanks so much Mark. I appreciate your story. Online valuation sites have a huge incentive to chase the list price if the accuracy rate depends on it. I get it of course from a value standpoint because if something didn’t sell, maybe that means something for value… In this regard we cannot ignore the list price or list price changes. Yet chasing it down strongly benefits the “accuracy” rate, and that’s not good…. Thanks again.
When I brought this up on twitter, Spencer said that their zestimate isn’t adjusted to the list price. But it used to be, and I think they might have just changed it in the last few months on the few I’ve followed.
Redfin is still extremely guilty of adjusting their estimate to within 1% of the list price, however.
Thanks Jim. It might not be universally adjusted, but sometimes it is like the case above. Thus on some level we know the list price means something to the algorithm – or outright sways it. It’s like the trump card in some cases.
I’m following a higher-end property right now in my market because I recently measured the house. The Zestimate definitely didn’t match the list price, but it did go up $200,000 when the property was listed. So the Zestimate was at $2.2M, the property listed at $2.8M, and then the Zestimate in 30 days increased to $2.4M. So even though it wasn’t matched, in this example it seems the list price is being given strong weight.
I just checked another example of a $2.4M listing I measured for an agent. The Zestimate was at $1.3M before the listing and now it’s at $2.35M. Hmm…
Here’s one to watch that was just listed today (5/1/19): 4117 Horgan Way, Sacramento. It’s described as a 3/2/1328 “fixer-upper” built in 1960, which has “lots of potential.” Listed at $258,000. Zillow estimate is much higher at $365,242 (+41.6%) and Redfin estimate is significantly lower at $227,749 (-11.7%).
(MLS #: 19026646)
Thanks Terrence. I appreciate it. I’ll be curious to see if either estimate ends up matching the list price or final sales price. To be fair this sounds like a fixer, so it would be reasonable to see it coming in lower. Though clearly there is a big disconnect between Zillow and Redfin here.
Take a look at the Z estimate (and history) for a new listing at 4803 Cypress Ave, Carmichael. What’s going on here?
=========================
4 / 2 / 2,362 • Built in 2016
Listed 4/28/2019: $655,000
=========================
Zillow Estimate
• 5/1/2019: $586,873
• Last 30 Day Change:
+$99,451 (+20.4%)
• Z Est History:
Apr 2019: $586,900
Mar 2019: $487,600
Dec 2018: $478,800
Aug 2017: $528,800
=========================
Price History
Sold 8/25/2017: $540,000
Thanks Terrence. I pass by this property frequently. I’ll be curious to see what it goes for.
Great points about Zillow. I know Zillow is pretty bad, I just wish I had some stats on average appraisers who do pre-listing appraisals and the eventual sales price. Maybe I don’t want to know that stat ;-). Our company tracks our own staff pre-listing success and sometimes it humbling and sometimes it is gratifying.
Thanks Gary. I always appreciate your realism. This is so true. Value is not easy. I just finished an appraisal yesterday that was really complicated. It’s amazing what happens when you add a few additions through the years and all of the sudden the floorplan is awkward. Well, now we’ll see where the market bites as it hits the market….
On one hand it is easy to criticize the machine (Zillow) when there are no stats available for the humans. I agree with you on humbling and gratifying. 🙂
It’s mildly amusing to see how far a Zillow estimate can be off for any home with unique characteristics – like in my North Natomas community. Some houses are lakefront while others are not, with significant price differences between the two. Since more homes are sold off of the lake, the lakefront homes have Zestimates that are very low. The most recent model match to my home sold for $594k a few homes away, but my Zestimate is currently at $430k. One of my neighbors is a retired realtor and was bothered by his low Zestimate, so a couple years ago, he contacted Zillow and explained the lake vs off lake situation. Interestingly, they briefly popped up all of the values of the lake homes, but the algorithm has reduced them since then (my home value peaked in Oct. 2017 according to Zillow!)
Two larger homes were recently listed on lake in the high 700s, and their Zestimates went up $200k overnight.
Thank you Matt. I really appreciate your example. What I find striking in the midst of your commentary is how much your neighbor cared – and that’s not uncommon. People get emotionally tied to the Zestimate. It’s just true. This is the part where the Zestimate is more than just a ballpark or starting place. On some level people feel tied to it and expend a certain level of emotional energy on it. We know objectively it doesn’t matter what Zillow says, but we nonetheless find ourselves angry, bothered, etc… This story happens all the time. I don’t say that to criticize in any way either. This is what the world looks like today. I haven’t looked up my home Zestimate for a while, but I’ll admit I would be bothered if it was far off. 🙂
Great post Ryan! I love the example that you gave, which I believe is a very common situation in Zillow. As I’ve always said, when it come to value, Zillow is a toy, not a tool. It’s just not reliable.
Ha. I’ve never heard someone say it that way. Thanks Jamie. The fascinating part to me is if Zillow continues to list properties and sell them, we may have to pay more attention to them as a data source. In this case I hope they publish data so we can track these sales. This is on my radar because the world is changing and data may be more diluted from the MLS at some point.
Agreed! I do think Zillow offers some useful data. I think they just need to stay within their area of expertise, which is not valuation. Definitely something to keep an eye on. Things are definitely changing in the way people buy and sell homes.
Thanks Jamie. Agreed. Zillow definitely has the best search out there too. Nobody beats them in terms of a mobile view of properties. Other companies would be wise to up their game and reinvent their websites so the consumer can easily find information, scroll through photos, etc… I find sometimes MLS systems or brokerage websites are clunky. So part of the allure of Zillow is not just the Zestimate or trends and such, but the way consumers are able to so easily interact with listings.
I totally agree! This is one reason why people love them. I have certainly used Zillow for obtaining information on some properties.
Thanks for pulling back the curtain on Zillow, Ryan. I think they are playing a shell game with the public when their Zestimate mirrors the list price. It doesn’t sound like they got their money’s worth when they recently paid someone $1 million to help increase their accuracy.
Thanks Tom. And just two days ago Zillow announced they’re entering the Sacramento market to buy and sell homes. It seems their plan is to buy privately from owners and then flip. They’ve built street cred with the public through their Zestimate, and now they’re looking to disrupt the traditional real estate model.
Relax, give Zillow a break. They do provide tax history and prior sales value, along with other general data, so it is an easy way to gather basic information. The vast majority of pro realtors do nothing more than guess about valuation trends, which can be just as wrong as Zillow or Trulia or any place. The one problem I have with any realtor is that they hope to make thousands of dollars off of idiots that depend on their guesswork.
Thanks Brave Chicken. No breaks here. I concur Zillow has its place in the world, especially when it comes to general data. But I don’t have much patience for stuff like this either though, and that’s an okay thing. Ultimately if a company is going to claim an error rate and use it to market themselves and earn the trust of the public, it’s prudent to understand what they are talking about and how they are working the error rate to their advantage. Of course it’s up to people whether they care or not. My goal here is to simply paint a picture of accurate information.
I listed my Northern California home on 05/14/2019 for $299,000. Just prior to listing the zestimate was $356,700. Almost immediately the zestimate dropped to $318,000, then it dropped again to about a few weeks later to around $298,000. On June 15, I lowered the price from $299,000 to $289,900 and within a day or so, the zestimate dropped to about $288,000. Yesterday, I did another price cut to $279,900 on 07/07/19 and predictably the zestimate is now about $10,000 lower than the day before at $278,400.
As a seller, this is very frustrating as the zestimate is clearly fluctuating to match the list price. However, what is most concerning, is that Zillow is marketing the zestimate as if it is an independent evaluation, where it clearly is not. Within a period of a month, the zestimate on my property slipped from about $356,000 to $278,000 which only goes to show that at least in some markets, for some homes a zestimate is less than worthless. It could cause a buyer too pay to much or a seller to sell for too little all so Zillow can advertise that it’s algorithms can accurately determine valuations 97% of the time, thus generating traffic and advertising revenue.
THIS!! Thank you Keith. I appreciate you sharing your story and I definitely share your concern here. I’ve seen this and heard of it time and time again. Zillow’s accuracy rate has a big-time asterisk, but nobody reads the fine print. They only look at the sexy number and then trust the figure. There is much more to the story though.
We live in a southern California hillside neighborhood. We’ve renovated the house inside and out (kitchen, bathrooms, flooring, etc). Weirdly, the zestimate for our house does not budge, while neighbors’ zestimates have zoomed up, even for houses that have poorer views, smaller lots, no renovation, etc. Could our stuck zestimate be due to the fact that my spouse has owned the house for 50 years, there is no price history, and the property taxes are a tiny fraction of what the neighbors pay? While there is not too much churn in the neighborhood, none of the neighbors have lived in their houses more than 20 years. Could it be because our house cannot be seen from the street (it is behind huge hedges and a wall)? Is there anything we can do to raise the zestimate? I noticed a neighbor on another street was in a somewhat similar situation. The neighbor sold the house for 850K. The buyer painted it inside and out, installed an in-ground pool, upgraded the landscaping, renovated the kitchen and baths, and sold the house for 2.7 million. Shortly afterwards, that buyer sold the house for 3.1 million. Can we “flip” our own house as a way to raise the zestimate?
Hi there. Thanks for reaching out. Why is the Zestimate inaccurate? Because it’s Zillow. We have to be real about this. They don’t know the inside of your house and the particulars of your view either. It’s possible their algorithm may penalize you in some senses because there’s less data on your home. Since the house hasn’t transferred or sold there is less context (less data to consider). If you sold for $2M in 2015, for instance, maybe that would mean something to the algorithm. Maybe not though too.
I’m curious though. Why does it matter to you what Zillow says? What positive would there be for you if Zillow got it right? I’d love to hear your perspective. I’m genuinely curious. Thank you.
Thanks for your reply. Makes sense.
Why it matters: we might consider selling the house at some point. Online advice suggests the Z can affect buyers’ perception of a home’s market value and cause them to adjust their offers accordingly.
On the other hand, some online advice suggests the Z can change abruptly to align with a selling price.
We’re simply perplexed about nearby Zs compared to ours. E.G.: a house across the street sold for 650K in 2009 and its current Z is 1.1 million. The house is 1213 sq ft on a 4300 sq ft lot. The peculiarities of the neighborhood mean houses on that side of the street have no views and sloping back yards (in some cases a 1:1 slope). Our house is 1843 sq ft on a multi-level 9000 sq ft lot, half of which is flat (unusual in the hills). We have two decks, 3 separate entrances to the house, a detached garage, gazebo, flat basketball court, flat grassy yard, mature landscaping with live oaks, and the backyard adjoins an open space area. The interior has been completely renovated. Z says all that is worth 300K more than the house across the street. Something is not right.
I agree with you. Their $300,000 does sound a tad optimistic, doesn’t it?
Thank you. I can understand that. I think some buyers unfortunately do get hung up on Zillow. But frankly those buyers are misinformed and coming to the market with a skewed education. Buyers in touch with the market will be able to think past this and the market should prevail rather than what one website says.
Think about it this way. If the home is really worth say $1.1M, buyers will recognize that and offer accordingly. Most buyers won’t be getting hung up on the Zestimate when they know value is there. Will some? Maybe. But my guess is buyers are going to be paying way more attention to the comps rather than the Zestimates. After all, when buyers shop these days they know everything about the market. They know as soon as properties list, they’re in tune with price reductions, and they often get a sense for what is overpriced and underpriced (because they’re paying such close attention to all sales and listing activity in an area). Thus to think a buyer in today’s digital age would forget about all the obsessive research and get stuck on only one Zestimate is out of sync with the way buyers are scouring properties today. But I understand what you are saying and I’ve heard this sentiment before, so it is a real dynamic on some level. But in my mind comps and the obsessive detail to searching for homes is the trump card here. I would put all your attention there and give very little weight to the Zestimate. My two cents.
By the way, Jim commented here and he shared a recent story on his blog of a property where Zillow was just chasing the list price. This perfectly highlights how much weight a list price is given in the algorithm. I think this shows a lack of credibility too when it comes to the Zestimate. https://www.bubbleinfo.com/2019/10/02/zestimate-manipulation/
Thanks very much! Good insights, and thanks for the link!