• Homes with more page views and more pictures sell more quickly.
• Homes priced higher than their estimated market value sell more slowly.
• In most cities, mid-tier homes sell fastest and high-end homes slowest.

“List on major websites.” “Add plenty of photos.” “Don’t overprice your home.”

If you’ve sold a home, you’ve likely heard all or some of this advice. But while it’s almost always well-intentioned – and offered in hopes of avoiding a long, protracted sale – is it really true?

Actually, yes.

Zillow analyzed ten key variables, such as number of photos included in a listing, page views in the first week after listing and the percent difference between a property’s list price and its Zestimate, to determine their impact on the probability a home will sell in a given amount of time (figure 1).

How To Read These Graphs

The curves in figure 1 represent the probability of selling a home in a certain number of days in a selected city, with the colors representing different ranges for the selected variable.

We break each variable into two to five buckets. When the variable has natural divisions, like with bedrooms (1, 2, 3, 4, 5+), we use those divisions for the buckets. When the variable is continuous, like with square footage, we break the variable into ranges in which each bucket gets roughly the same number of homes. For variables where it’s useful to compare exact values across cities, like the list-price-to-Zestimate percent difference, we kept the bucket ranges the same for all cities; otherwise, the bucket ranges are calculated separately for each city.

Each curve has error bars (confidence intervals) indicating where we believe (with 95 percent confidence) the true probability of a sale falls. Small sample sizes generally mean wider intervals.

For example, select “Seattle, WA” for the city and “Page views in the first week” for the variable. The yellow curve represents homes with 690 or more[1] page views in the first week of being listed. If we want to know the probability of one of these homes selling in 60 days, we find 60 on the “Days on market” axis, follow it up to the yellow curve and see that the “Probability of selling” axis has a range of 64 to 75 percent. Hovering your mouse over each point will also give this information.

To put all homes in the same bucket for a particular city, choose “[None]” as the variable.

So, what did we find?

It’s All About Eyeballs

Generally speaking, a home with more page views in its first week on Zillow sells more quickly. This isn’t a surprise: More page views means more interested buyers, which in turn leads to a faster sale. But the magnitude of the difference may come as a shock. Nationwide, a home in the lowest bucket of page views (fewer than 100 page views) has an estimated 12 percent chance of selling in sixty days. A home in the highest bucket (280 or more) has around a 36 percent chance – or three times more likely.

For the page views variable, we excluded homes with a pending offer in the first week; otherwise, the bucket with the lowest number of page views would include homes with both the lowest demand and the highest demand. However, you can view the probability of sale for these homes by choosing the variable “Page views in first week (w/ early pending sales).” Almost by definition, they’re much more likely to sell: Nationwide, there’s a 73 percent chance they’ll sell in 60 days. (The reason it’s not 100 percent is that some sales may take longer than 60 days, or even fall through.)

In strong sellers’ markets like San Francisco and Seattle, the smallest and largest buckets aren’t separated by as large of a gap. This is likely because more homes in these hot markets are getting a healthy number of page views, so they effectively behave like the high-bucket homes in cities more favorable to buyers. The best example of this is Seattle, where the lowest bucket includes homes with fewer than 390 page views; at the U.S. level, homes with 390 page views are found in the highest bucket.

Take a Picture, It’ll Last Longer?

Well, actually, it won’t: More pictures means a faster sale.

In particular, homes with fewer than nine photos fare worst. Nationally, a home with fewer than nine photos is about 20 percent less likely to sell in sixty days than a home with 22 to 27 photos.

Homes with the highest number of pictures (28 or more), interestingly, sell less quickly than those in the fourth bucket (22 to 27). This is likely because other factors begin to outweigh the small marginal effect of each additional photo when there are already a lot of photos to begin with. For example, homes with 28 or more photos are 12 percent more expensive, and garner 5 percent fewer page views than listings with 22 to 27 photos.[2] Homes with 28 or more photos are also listed at a price a median 4.9 percent above their estimated market value, which is high compared to the 3.5 percent listing premium for the U.S. as a whole. Higher home values, fewer page views and higher list prices relative to estimated value are each correlated with slower sales.

The Price Must be Right

Homes priced around (and sometimes well below) estimated market value sell fastest. Most reliably, homes priced too high (at 12 percent or more above the estimated market value) sell slowest. In fact, these homes are almost 50 percent less likely to sell in 60 days than homes with list prices closer to the estimated value.

You may expect heavily discounted homes to sell fastest – and sometimes that’s the case, like in Chicago and Philadelphia. But in cities like Los Angeles and San Diego, homes could be discounted for a reason (like disrepair), in which case unobservable factors may be preventing these homes from selling.

Mid-Tier Moves, Luxury Lags

Within cities, homes valued at each end of the spectrum – those in the lowest and highest buckets – often sell slowest, with high-end homes most reliably lagging the pack.[3]

A notable exception is Detroit, where homes in the highest-value bucket sell quickly. The catch, of course, is that the range of home values in Detroit’s highest bucket is enormous, including all homes over $56,000. This bucket likely captures the relatively few homes that are “move-in ready,” whereas the lower buckets include less-expensive homes that require substantial renovation before they’re deemed livable by many buyers.

Your turn

We’ll leave it up to you to explore the relationships between some of the other variables such as home size, home value per square foot, year built and the ratio of lot size to finished square feet.

As you explore, it’s helpful to keep in mind that correlation isn’t causation. For example, nationwide, the largest homes sell less quickly. Of course, if the Real Estate Fairy added 1,000 square feet to your home, odds are it would sell more quickly, not less – holding everything else constant. The reason it sells less quickly in the first case is that square footage is correlated with other variables (e.g., home value, geographical factors) that contribute to a slower sale. However, given that square footage of a home is high, it makes sense to say that it will sell less quickly. The difference is subtle but important!

Methodology

To create the curves, we conducted a survival analysis, which is traditionally used to estimate and compare the cumulative survival probabilities of different groups of patients (e.g., ones that receive a treatment versus ones that don’t) at different points in time.[4] Instead of surviving patients, we have homes that don’t sell. And instead of patients who survive over the course of the medical study (which are “right-censored” observations), we have homes that don’t sell by the end of the analysis timeframe.[5]

The data are for all single-family residences, condos and co-ops listed for sale in 2014 Q4 and sold before March 31, 2015. We only included properties listed for a week or more, except where the home had a pending offer in the first week.

[1] This appears in the legend as 690 to 999 since the highest number of page views after removing extreme values was around 1,000.

[2] When comparing the medians of the buckets.

[3] Nationally, high-end homes have the highest cumulative probability of selling between roughly 30 and 60 days, relative to the other buckets. This is because the high-end bucket nationally is over-represented by homes located in sellers’ markets (think San Francisco, where homes are expensive and sell quickly), which explains why the trend isn’t seen at the city level.

[4] The statistic used to estimate the survival function was the Kaplan-Meier estimator.

[5] Note that although patient survival is the focus of a traditional survival analysis, the probability that a home doesn’t sell isn’t the focus of our analysis. So we calculate one minus the probability that a home doesn’t sell to find the probability that it does sell.