Discover more from The Reflection Point
Finding the Perfect Place to Live
How and why I built RELO, the recommendation platform for where to live
Where to live is a massive question. In fact, it’s so overwhelming that most people seem to put more research into buying a TV than into where they will spend their next decade. Why? There is so much information and so few tools to cull it all down that we end up relying on clickbait-y Top 10 lists.
Which is nuts! Our physical location can determine who we meet, how we feel, and how our children evolve. It’s a lot like picking a college in terms of ramifications and process, but then imagine only visiting one or two campuses and then going “sure, this place is fine.”
As I’ve tried to figure out my own move after we hit our eventual Manhattan expiration date, I realized the research I was doing could help other people with the very same problem.
And so I built RELO (relome.io) to take everything I had been doing myself and streamline it into a recommendation and organization platform.
In doing this, there are a million different questions I could have answered. Given this is v1, I chose to focus on a key few. Below are the whats, hows, and whys of how it came to be, as well as dreams for what it might be in the future.
How do you curate an entire country’s worth of places?
Love it or hate it, there are places around the U.S. that plenty of people live in but not many people would voluntarily choose to move to. Curating the “viable” places down to 5,000 of the 33,000 zip codes could absolutely be controversial, and against my best efforts, I certainly may have missed some great places.
The options are based on factors such as proximity to major and minor airports, overall amenities, and even basic things like having a population of over a certain amount (there are some zip codes with a population of 10). Ultimately, it’s objective and subjective. It’s a tool for me specifically, so there is definitely me questioning “would anyone actually choose to move there?” when I look at a place I haven’t heard of before.
Some of these places are obvious – cities, suburbs – but others are more up and coming with the Covid diaspora. Las Cruces, NM. McCall, ID. Quechee, VT. Auburn, CA. All very cool, all very off your radar if you are on the other side of the country.
How can I create a ranking system?
Given Covid and the remote-work boom, there is a higher chance than ever that I’d consider moving far afield. I wanted to somehow rank places I was less familiar with. The final algorithm is based on the question “which are the towns that are a best fit for me?” It’s personal. My dad and I could do similar searches, but the small differences will impact in what order the results emerge.
How can I make 5,000 different zip codes feel similar?
All these websites – NYT’s Relocation Tool, Bestplaces.net, etc – just hoover up dirty, public data from the census and feed it back to us. But data without context is not helpful at all.
Population density was a big one that I felt no one did well. I knew it mattered but the numbers could be wildly misleading. For example, in massive, Western-state zip codes, the places people actually live can be pretty dense, but when divided by the total landmass of the zip code, it might look super rural.
So, again, through objective and subjective approaches, I aimed to categorize each zip code based on “feel”. Because in the end, that’s what I’m looking for as I search for a new place - feel. Labeling each area with “City Vibes”, “Tight Knit Neighborhoods”, or “Spread Out Properties” enabled me to find similarities that felt familiar as I jumped from San Diego to Omaha to Upstate New York.
Of course some places can have one side of town that have many homes close together and then another side that has big properties, but my goal was to provide the label that worked more often than not.
The problem and opportunity with zip codes.
Zip codes are the compromise between data availability and location specificity. There are some zip codes in cities that cover multiple neighborhoods, all unique in their own way. There are some towns that are very much just one zip code. Using zip codes allows for just enough specificity so I can ask “what area in Seattle is best for me?”
Neighborhood labels would be amazing, and sometimes zip codes actually do get it right – just look at SoHo vs. Tribeca vs. Battery Park City vs. FiDi. I suspect there is much more to be done here with new types of alternative data sets.
What factors matter?
This is a massive question, and in RELO right now there are certainly not all the potential factors that I want.
Things like Walk Score have done an ok job, but they still miss the fact that some areas have great, walkable downtowns, but it might take a quick drive to get there from your house. “Nightlife” is similar. Do you have to be within walking distance of a million bars? Or do you just want to be within driving distance?
Cost of living calculators are another broken feature everywhere I look. Cost of living is almost always heavily weighted towards housing prices, but if you already know the cost of housing, then really what you want is how expensive it is to live your life there. For example, how much does it cost to eat out in Charlotte, NC vs Manhattan? Anecdotally it’s around 25% less. Is milk 25% less expensive as well? And a Bud Light? I’m not sure.
Climate change is another one that I would love to incorporate. Is a location drought-ridden? How likely are you to face a hurricane or tornado? Where are the best places to ride out the next 10-20-40 years? Everyone’s pushing the Great Lakes region but data should be able to back that up.
The factors above take a ton of work and some money to figure out. Instead I left those questions to be answered post-RELO search. The things I could factor in now include age weightings, housing prices, region, political leanings, coastal/mountain proximity, weather, great public schools, and if it’s near a college town.
Again, it’s not just regurgitating that data; it’s providing context so I can actually use it. “Average home price” could be wildly misleading, as 4 bedroom homes are sometimes 3x the price of 2 bedroom homes. So I broke out those distinctions. “Days of Sun” is a figure that’s confusing as well, but armed with context about extremes and averages, I can quickly use analogy to get a read on a new place.
Can I save houses and notes for towns I like?
This was a big one for me. How many times did I end up with a million tabs open of different places and different houses? I built the favorites tab in RELO to make that part of my search much easier. I can quickly take notes on a specific place, plus easily save down any house from Zillow, Realtor, and Redfin into my profile. That way when I look at it next week, I can quickly see thumbnails and key data on houses that piqued my interest, plus see all my notes.
Connecting with an on-the-ground guide (a real estate agent) would be a great next step, but putting together a referral system is another whole thing entirely. And as RELO is just for me right now, I’m still left Googling what the best real estate agents are for each area I’m interested in.
User Experience Flaws
“If there’s a wrong way to do something, your users will find a way to do it.”
Unfortunately, if they are doing it “wrong”, I’m not designing it “right”.
Some things I’ve seen with people who have tooled around on RELO already:
Everyone wants to click everything:
If you give people the chance to click “low tax”, “best schools”, “college town”, and “Liberal”, they absolutely will, even if that is a fairytale situation that doesn’t exist anywhere. Same goes for hitting “Rockies” and “Coastal”. So far I’ve fixed this by showing more general, ranked results whenever a search doesn’t bring anything up. Better than that would be identifying what the limiting factor is and taking it off for them.
People say they want the “?” info buttons but no one actually clicks on them:
If I ask someone to walk through the platform with me, they’ll click everything, pretending that’s what they’d do in real life. Looking at user behavior anonymously shows they just dive right in and they assume they know what everything means. This points to making sure labels are are as clear as possible, because I’ll only get one shot.
I’ll only get one shot:
If someone goes on and doesn’t get results that feel right, they’ll move on. “It doesn’t work”. That means it’s on me to convey all the value they can get by going through the process (like saving homes) so they can understand it’s worth their time to play around with the search functionality.
There is so much RELO doesn’t do, but then there is so much it’s helped me with as I narrow in on where we should move.
There’s a long wish-list that may or may not come to fruition over time. For example, where your friends, family, and job are could be the main factor in where you go next, so maybe a radius tool would be a great addition. A quick tax comparison tool - income and property - would be super helpful as people realize that low income taxes can often mean very high property taxes.
In the end, as is, I hope RELO is as helpful to others as it has been for me! Play around with it and let me know! relome.io