Airbnb Case Study:
How To Reduce Churn
With Personalization

Use your keyboard arrows
to view the story!

Story Duration: 5 min

I've been travelling while working for a bit now.

My girlfriend and I were in a coffee shop in Lisbon, when suddenly…

Remember to use your keyboard arrows to navigate

…a last-minute event meant we had to book an Airbnb in Porto.

Despite the rush, I have to admit that—

—I was psyched to analyze Airbnb's experience!

Oh, calling me by my first name!

That's a good way to make Airbnb feel friendlier than other hotel booking apps.

Also, notice these…

…shortcuts to jump back to my previous searches.

With that said, none of these apply to my recent change of plans.

Searching is the main thing I need to do right now, but the search field contrast is so low, I almost missed it.

Also, don't you find it weird that—

—"Kuala Lumpur" is the default placeholder, yet it's 16 hours away from Lisbon by plane?

Location Personalization

Airbnb has the data required to know that I'm using their app outside of my home country (Canada).

In this kind of situation, showing users nearby cities instead of random ones across the globe would make the app feel more aware of their context. The "time to first successful search" would also likely improve.1

¹Conversion XL, Geo Personalization AB Test (2019)

Anyways, let's search for "Porto"! {tap}

Oh! I like that they already show me results without forcing me to fill in "Dates" or "Guests"!

Nudge. Don't Push.

Consider nudging users in the right direction before forcing steps.

Two things are required in the case of filters though:

Users need to understand the benefits quickly to have an internal motivation to act and
The default results must provide enough value.

Just keep in mind that users don't want more filters, they want better results¹.

¹Hopper Blog, Users Don’t Want Filters (2018)

Let's start with "Dates"!

{tap}

Also, this byline is a convincing reason for me to use the filters…

The date-picker experience was pretty standard, but…

Usage-Based Personalization

95% of my previous ~200+ Airbnb trip searches were for me and my girlfriend (2 Adults, 0 children, 0 infants), with clear booking patterns (e.g. a strong preference for a King-sized bed).

Airbnb could increase their conversion & save users precious time by using past search data patterns to suggest default filters for users' searches.

…this "Guests" filter makes me wonder the same thing every single time:…

Me + my girlfriend = 2.

But how does Airbnb make the difference between:
A) ❤️ when I travel with my girlfriend versus
B) 💼 a business trip with my cofounder? (i.e. 2 beds)

Wouldn't now be a good moment for the app to ask me to specify it?

Anyways… {tap}

Nice, this "Work Trip" button sounds relevant to me! {tap}

Status Quo Bias

When in doubt, users tend to not take action unless they understand exactly what will happen. That's when photos or illustrations can help clarify the context and outcome.

Side Note: Considering the exponential trend of "digital nomads", Airbnb definitely has an opportunity here.¹

Digital Nomad Google Trends

¹Google Trends, Digital Nomad (2019)

Wait… What's a "business traveller"?

Are we talking about:

A) digital nomads in frugal & trendy coworking spaces?

or B) corporate suits on expensive business trips?

Hm… I'd rather not mess up my filters. Skip! {tap}

Another nice nudge towards the rest of the filters. Well played!

Let's see the remaining options of this "Filter Fest".

{tap}

…"Hair dryer" and "Hangers" !?

My girlfriend & I never used those filter options once in the past 5 years.

Social Proof Pitfalls

Social proof states that "we are most likely accept the actions of others as correct".¹

But beware: when "popular filters" lack personalization, they tend to backfire and make users doubt the overall relevance of the other suggestions.

¹Social Proof, Growth Design Library (2019)

As I scrolled down, that's when I remembered something important…

"Instant Book" is here as promised.

(Putting it above the rest might have been more coherent with their previous nudges though.)

As for "Wifi" & "Kitchen", those are fair recommendations, but…

Airbnb has a filter for "number of beds", but none for "Bed Size".

That's annoying because large beds are rare in Portugal, and you see—

—I'm 6'3'' 225lbs.

I take a lot of space.

In fact, whenever my girlfriend and I try to sleep in tiny double beds, we just—

Plus, Airbnb does not save your preferences across different searches.

You have to re-enter the same ones everytime, and—

—considering the length of their list of filters, that requires a lot of effort.

Customization vs Personalization

Customization is when users deliberately choose between options. It's a high cognitive load task.

Personalization is when you anticipate users' needs and automatically show them a relevant experience. ¹

e.g. Google Maps successfully leverage past user behavior to give a "Match Score" for restaurants.² Airbnb has a great opportunity to do the same for places to stay.

¹ConversionXL, Content Personalization
²Mashable, Google Maps Match Score

I don't feel like sorting through this.

Let's go straight to my most important search criteria:…

{tap}

…the location!

{tap}

And that's where it gets weird for Airbnb because you see—

Hm… that default suggestion seems isolated.

I like to be able to walk to a nearby restaurant or coworking space without having to…

…spend hours in commute every week.

—the fastest way I found to make sure a listing is well-located is to…

…search through another app called Momondo.

It's time consuming, but at least—

—they have this amazing feature called "Hotspots".

Based on your preferences, it shows you exactly…

{tap}

…where you should stay!

Feature vs Jobs To Be Done

Airbnb has a map feature for its listing, but that shouldn't be confused with one of its underlying Jobs To Be Done¹:

"When I choose a place to stay, I want to be near what's important to me (e.g. sight-seeing, restaurant) so that I can save time and enjoy my trip more

Failing to address this means their users have to use other apps, which is risky when those alternative apps offer competing hotel booking services.

¹Jobs To Be Done, Lean Methods

Okay now that I know which neighborhoods to aim for…

I want a place that's as close as possible to a coworking space (& ideally a gym).

And yup—it's gets even weirder because I now have to… {tap}

…use yet another app:
Google Maps.

Plus, I have to spend even more time to find relevant spots on the map.

—oh!… 27€/night? Seems like a solid deal!

And you know what?…

Hick's Law

The time it takes to make a decision increases with the number and complexity of choices¹.

The contrast between the endless choices in Airbnb and the serendipitous discovery of this well-rated & cheap option in Google Maps made it seem like a better option. That's part of the reason why personalization is so powerful.

P.S. we explain decision-making in detail in our Product Psychology course.

¹Interaction Design, Hick's Law (2019)