How RadPad is using bad data to help you find your next apartment.
In the spring of 2012, I started looking for a new place to live. Not because I didn’t like my current apartment, I actually lived in this really awesome condo in the heart of West Hollywood. It also had nothing to do with my landlord, whom I had become good friends with while renting from him over five years.
I wanted to move because I was ready for a new adventure.
What I love about renting is the freedom it gives you to come and go, live in new places and experience new things that come from meeting new people in new neighborhoods.
I was ready to do this. I had made up my mind to move. Now I just needed to find my next apartment. I knew exactly what I wanted.
I used Craigslist and one other service, and while Craigslist in particular had a lot of listings, the experience was a fucking mess.
A lot of the apartments on Craigslist were duplicates or had already been rented and when I would come across an apartment that was ‘the one’ the landlord either wouldn’t respond to my messages or it ended up being a scam.
I was frustrated, discouraged and ready to give up on the hunt. The real problem was that there were all these bad listings mixed in with the good ones and figuring out the good from the bad was a nightmare.
It was exhausting.
That’s when I got mad and decided to do something about it. Six months later Tyler, Tim and I launched RadPad.
For the last two years we’ve been building a marketplace focused almost exclusively on renters.
Most renters know RadPad by our mobile experience, but so much of what goes into creating this experience has to do with what goes on under the hood.
Last month we quietly released a significant update to RadPad called PadRank. PadRank impacts every listing and landlord on RadPad.
Every month, RadPad processes more than one million listings with more than six million photos. We’re talking about a ton of data coming at us from a multitude of sources.
One of the biggest problems in the entire rental industry is the quality of the listing data. The rental industry is plagued with bad data and this comes in the form of shitty photos, inaccurate listings, incorrect address locations, listings that have been rented but still show as available, duplicate listings and of course, scams.
I believe that the biggest problem facing the rental market today is the sheer volume of bad data that’s being passed around from one rental site to another.
A main source of stress, frustration and anxiety experienced by renters in their search for a new apartment is directly correlated to the amount of bad data the renter experiences during their apartment hunt.
When we started RadPad two years ago, we took a small but important step by requiring at least three photos for each listing
By deliberately keeping listings off of RadPad because they didn’t meet our three photo requirement, we risked the perception with renters that we had less listings than our competitors, which might make us less desirable to use when searching for an apartment.
Even though we considered changing the photo requirement several times early on, we kept it because we noticed a correlation between the quality of the listing and the number of photos it had.
Better photos help renters make better decisions, and if renters can make better decisions, landlords should get better leads. It’s a win/win for both the renter and the landlord.
We’ve downplayed all that we do behind the scenes to clean up the vast amounts of listing data that ends up on RadPad, because even with all that we do to ensure we have the most accurate listings in the world, bad data still makes it onto RadPad.
Bad data = a bad user experience = frustration and stress = lack of trust in RadPad = failing our customers.
Accurate data = more informed decision making = trust = happier renters = higher quality leads = happier landlords.
With more than two years of experience learning all we can about the listing data in our industry, we’ve made a major step forward in how RadPad handles all of this data.
PadRank is an algorithm that was developed by one of our engineers Justin Jia in conjunction with our engineering & product teams to evaluate the quality and accuracy of all our listings, in real-time.
PadRank’s job is to assign every listing on RadPad a score, and decide if the score meets a threshold that is high enough to show the listing to renters on RadPad.
PadRank scores listings on several criteria, some of which includes:
- Quality of the listings photos
- If we’ve been able to verify the landlord
- The time it takes the landlord to respond to renters
- If the listing has a valid street address
Here are a two examples of listings that were scored by PadRank.
PadRank Example 1:
PadRank Example 2:
PadRank favors landlords & data partners who provide better content and respond more quickly to renters.
We’ve already noticed a significant improvement to the quality of our listings, including the reduction of duplicate & inaccurate listings on RadPad.
PadRank, in its first version, doesn’t solve all of RadPad’s problems. We still have inaccurate listings, renters will still experience radio silence from landlords and unfortunately, scams will still make it on RadPad.
However, PadRank is a big step forward to improving the integrity of our entire marketplace while upping the experience for renters and landlords.
The algorithm behind PadRank will continue to evolve over time with our goal being able to say that 99.99% of all listings on RadPad are 100% accurate.
But until we get there, we’ll keep our heads down and continue working hard; because we owe it to ourselves and all of our customers to solve this problem once and for all.