This article was originally published in the First Round Review.

Not terribly long ago, headlines predicted that Facebook would die because it couldn’t figure out mobile. Of course, we all know the ending to that story. The company invested a ton in making the switch and survived. But it wasn’t without serious struggle. Ultimately, it required much more than a shift in technology — the whole culture of the company had to change. Everyone had to understand what users truly wanted from mobile and the steep costs of screwing up.

If she’s learned one thing in her 12 years building products for mobile at companies like Trulia and Nokia, it’s that people treat their phones as an extension of themselves and their capabilities. Mobile is deeply personal. On one hand, phones can leverage identity data and context like location to deliver remarkable, textured experiences. On the other, it’s painfully easy to violate trust.

In this exclusive interview, Moatti speaks from a wealth of experience — including running cross-platform notifications at Facebook. She provides the most critical lessons for creating mobile products that not only capture people’s fascination, but hold it long enough to become a part of their daily lives.

MOBILE’S HOLY TRINITY

“When you’re trying to become your best self, you think about your body, mind, and spirit — there’s a similar trinity when it comes to mobile development,” says Moatti. “You want to look good — mobile products have to look good. You want to have a meaningful life — mobile products need to be meaningful and personalized for users. You want to keep growing and learning — mobile products have to change and grow with the people who use them too.”

People bring their devices with them everywhere, and no matter where we are or who we are, we expect them to work flawlessly. In short, we expect magic from mobile. This is a big part of what made the shift from web so challenging for Facebook. As a free web service, the cost of a mistake for the company had traditionally been rather low. So low that “move fast and break things” was a motto that made sense for a long time. Whenever something went wrong, a few engineers would team up to fix it — no harm, no foul. But this isn’t how mobile operates.

When a mobile application oversteps or breaks or isn’t intuitive, the cost is very high.

“From a tech perspective, a lot of problems take a long time and more people to fix,” she says. “Native apps are not nearly as flexible and fixable as web sites. From a user perspective, because people are more emotional about their phones, you can break trust, you can turn people off, and they don’t come back as easily — or ever. If the cost of an error on web is 1X, the cost of a similar error on mobile is 10X.”

When Facebook made the move to mobile, it had to ditch its “break a few eggs to make an omelette” mentality, a big change in the company’s core values. But it was worth it in the end. Now well over 90% of Facebook’s traffic comes through handheld devices. What follows are the key lessons they learned along the way, and that Moatti has seen repeat again and again at companies striving to do the same.

RETHINK YOUR FUNNEL

Digital discovery is critical and terrifying. There’s a lot of pressure to maximize the number of people who can find your product — i.e. if you don’t nail this out of the gate, you’re already sunk. It comes down to 3 steps:

• Set a clear goal: Whether you want to get downloads, impressions, or sign-ups, pick just one metric goal to track and optimize toward.

• Place that goal at the bottom of your funnel: Work backward to determine what has to happen each step of the way to get users to that destination.

• Optimize the conversion rate of that funnel: Lots has been written about the topic by growth hackers and lean startup consultants, but that doesn’t mean there’s no new ground to break.

Sounds good, but Moatti says that this very systematic approach eventually yields diminishing returns. You may be able to improve your goal by 20%, which is already an amazing accomplishment, but on mobile, you need more if you’re really going to be successful, because the ecosystem is so complex.

“There are two tools you definitely should use to take back control and better navigate the discovery process: shortcuts and hooks,” says Moatti. “Shortcuts reduce the time and effort it takes for users to get to your goal. Hooks get as many people into your funnel as possible.”

Shortcuts

Effectively, shortcuts make your funnel less deep, by cutting out some steps. Take Amazon for instance. When you buy a book, it immediately shows you other books people who bought that book purchased. You don’t need to go back and search for other books by the same author, or other books on the same topic. The list is right there in front of you. That’s an example of a shortcut.

Hooks

Hooks broaden the top of your funnel. Zillow is the perfect case study here. On average, people buy or sell a home every five years. Given that rate, it wouldn’t make sense to only market to people who have identified as home buyers or sellers. That audience would be a tiny moving target. Instead, Zillow redefined its addressable market with its “Zestimates,” an estimated value of a home, which is a general curiosity for most people — who doesn’t want to know the value of the home they live in, or the one their boss lives in? It was the hook that got millions to engage with Zillow repeatedly.

You need to psychologize your audience to devise the proper hook. Moatti advises starting with this question: How can I make my service interesting and fun for everyone? What data can I expose that would be addictive for people to know and share? Basically, what do they not know yet, that ideally is of popular interest and regularly changes? For Zillow, that was housing prices. Find and test your hook to maximize your app’s discovery.

DEMONSTRATE YOUR VALUE, IMMEDIATELY

Your onboarding process should be defined by one theme: Showing users exactly how valuable your product will be for them. What’s important for them to know to recognize this value? Put it front and center.

For instance, new Facebook users (yes, there are still many), are not immediately thrown into a newsfeed or asked to create a profile. The one thing they’re asked to do is connect with their friends who are already on the site. From the first moment, it’s clear that people you know are already using the product — which is pretty compelling.

“Everything you see encourages you to connect with friends, until you reach the magic threshold of 20 connections,” says Moatti. “That onboarding threshold is basically the line at which the product gets cool and relevant enough for people to see the value and stick with it. If you have 20+ friends, you get a great, full experience.”

You should know what your product’s onboarding threshold is. At what point, and after what actions, does your product become sticky? Why?

Pandora is another example. Given all the music services out there, its challenge is to demonstrate its unique value proposition quickly. So its onboarding flow has to pull double duty — both teaching people how to use the app and showing off what it does best. Accordingly, Pandora’s onboarding threshold asks users to create their own station so they can experience the power of its recommendation engine.

DON’T SHOW TOO MUCH, DON’T ASK TOO MUCH

One of the biggest mistakes Moatti sees is apps trying to cram too much into their onboarding flows. “I get it, you’re passionate about the service you’re creating, so you want to tell people all about it. But no, you really have to restrain yourself.”

It’s vital that you get new users to focus on one task at a time.

Helping your users focus on what to do next is value you’re providing. Think of it that way.

Next to your app straight out crashing (the ultimate kiss of death), an overly complicated welcome process is the biggest turn off.

“People don’t want to hear about your service, they want to see why it’s useful for them,” she says. “When I used to run the rental division for Trulia, we made sure the app showed rental listings immediately — as soon as someone opened the app. They were immediately reassured that they’d find what they were looking for and that we wouldn’t waste their time.”

Before asking them for any personal information, the app would very organically allow users to set criteria around the type of rentals they were looking for — whether they wanted a two bedroom, or parking, or something for less than $3,000 etc.

“Instead of asking people directly to sign up, demonstrate to them why they should,” Moatti says.

Once a user had specified their criteria, the app would ask them if they wanted to save their search so they could pick up where they left off later, and if they wanted to receive notifications of new listings that matched their preferences. Only if they said yes did the app prompt them to enter their email. And at that point, it seemed like a good exchange for something concrete.

“Crossing that big threshold where users give you their contact information becomes a lot easier when they perceive they’ve already gotten value from your service,” Moatti says.

This is where the initial onboarding flow ended for many users. But when they came back, triggered by a notification of new listings, for instance, the app led them to engage even more deeply — again by focusing on the next logical, single task.

“When users came back to look at a rental, we’d ask them if they wanted to apply. And if they said yes, we immediately showed them other rentals that also met their criteria,” she says.

When Trulia first started doing this, they noticed that many users would go on to apply for multiple related rentals they were shown. This made sense — renters are usually hoping to get selected instead of being overly selective themselves. So Moatti and her team took the app one step further. “We decided not to ask them if they were interested in other rentals, we just went ahead and applied for them on their behalf.”

This was a pretty risky move, but after the change was made, they saw a spike in their net promoter score. “When you think about it, renting is a pretty painful experience. No one wants to apply 25 times for 25 different places they know they’re likely not to hear back from,” she says. “When we applied for them, we increased the likelihood they’d hear back from someone, and they ended up being much more satisfied with the Trulia app.”

You should be looking for and testing out such shortcuts all the time, according to Moatti.

By sending notifications and automatically surfacing only the listings that met user criteria, Trulia drastically cut down the amount of searching users would need to do to find what they wanted. “We really evaluated all the actions we wanted users to take and got serious about cutting out as much as we could,” she says. “As a result, they knew we’d only show them content that interested them, and we’d make life easier in surprising ways.”

KNOW WHEN TO ASK PERMISSION

One of the top questions Moatti fields from app makers is when they should be asking permission to send push notifications, access a phone’s camera or contacts, and more. There’s a lot of data pointing to the right way to do this.

“Strive to ask permission either as early or as late as possible,” she says, citing WhatsApp as a prime example. “The beauty of WhatsApp is that you can message with your friends instantly. So it’s critical for it to send push notifications whenever someone messages you. That’s why it asks for push permission as early as possible, even before signup. If you say no, the app actually says something like, ‘Sorry, you can’t use WhatsApp because push is so critical to the experience.’”

Shims, Jigs and Other Woodworking Concepts to Conquer Technical Debt

At the same time, WhatsApp waits as late as possible — until you want to send a picture — to ask for access to your camera roll. “If they asked upfront when I first started using the app, I might be confused, it might give me pause, I might be overwhelmed. If I want to send a picture, then I have more understanding and high intent to agree with an ask, so that’s the perfect time to request permission.”

Moatti recommends making a comprehensive list of all the permissions you might possibly need for your app to deliver a full experience. Think about all your capabilities based on platform. Android, iOS, watches, iPads, etc. all have different abilities you can bake into your product. You want to make sure you’re being thoughtful about all of them.

Turn your list into a matrix where you mark whether you want to ask that particular permission as early as possible or as late as possible or not at all. For example, you might want to access their location data as early as possible, but their iOS health data as late as possible. Cross out the permissions you don’t need. You end up with a game plan.

But don’t stop there. “You should always be A/B testing the right moments to ask for critical permissions,” says Moatti. “At Facebook, we experimented with asking to send push notifications before log in and then immediately after login. You’ll uncover opportunities to optimize.” On top of that software updates come with changes to permissions, so you’ll want to revisit this regularly.

BUILD IT BEAUTIFUL

Design plays a major role in whether a mobile strategy succeeds or not. People have much higher expectations for the look and feel of mobile products than they do for their web-based counterparts. This might be because devices themselves are sleek and high-design. It might be because Apple in particular has placed such a premium of beautiful experiences. Regardless, the bar has been raised, and mobile products that look dated, unpolished, or clunky are at a drastic disadvantage.

In technology, beauty isn’t so much about pretty pictures and more about efficiency: Nothing is wasted, and that’s beautiful.

“You might be building for a phone or a watch or glasses — it doesn’t really matter — as long as you’ve devised the most efficient experience for that platform. There should be no extra functions or buttons or text. You don’t have a lot of real estate and you can’t let people get distracted. You need to lead them along one clear path through your app.”

This explains why ads aren’t nearly as effective on mobile and why the business model needs to be different, she says. “You see subscription models working much better on mobile. If you want to run ads, they have to be sophisticated and organic. The cost of doing this wrong is way too high. People will see one ad and decide never to use your product again.”

Heed the ‘Thumb Test.’

“Imaging that your user is on their phone, waiting for the bus, surrounded by distractions and your app has to hold their attention,” says Moatti. “It’s very important that they be able to use your product using an average thumb. If they can’t, you fail the Thumb Test, and will lose users.”

She sees so many mobile products where buttons are either too small, or placed too close together, or links are confusing, or new unwanted windows pop up. “As soon as you reroute someone to a screen they didn’t want or expect, they’ll lift their head up and you’ve lost,” she says. “Someone should be able to complete all critical actions quickly without ever losing focus.”

MINIMIZE UNPREDICTABILITY

With mobile apps, surprises are very rarely a good thing.

When a user is suddenly presented with a screen, a choice, an action that doesn’t seem logical or intuitive to them, there’s a high chance they’ll bounce and never be seen again. You want to rigorously remove any experience that could feel this way.

Your navigation, in particular, can either make your app clear and pleasant to use, or be a huge source of confusion. “If you look at solid apps out there, their navigation is singular and clear,” says Moatti. “Facebook, for example, has the five primary options across the bottom of the app. There are a limited number of ways to go, and this constraint is comforting.”

Too many companies try to offer too many features, she says. As a result, they end up with multiple navigation elements. They might have the few choices in the bar at the bottom, but also a hamburger menu in the top corner brimming with even more choices. “My recommendation is to have a single navigation system in only one place in your app,” she says. “Pick the one that works best for you whether it’s a hamburger, or a bottom bar. It can be whatever fits your style as long as it restricts users to the core choices that make your product great.”

Gestures are another area ripe for unpredictability. Many app makers get too creative with the gestures they require, and it only makes user experience worse. “Right now, we’re at a point with mobile where you should never be coming up with new gestures for users,” Moatti says. “Still, I hear developers say all the time, ‘Instagram introduced the double tap! Tinder made swiping a thing!’ But we’ve gone too far.” Now there’s a standard lexicon of gestures you should draw from to make your app accessible.

Recently, she saw an app ask for a corkscrew gesture to zoom in and out. Even though it illustrated what to do, it was very unintuitive. “It ruined the entire thing for me, even though it was one small part of the app, it made me question the construction of the product as a whole.”

This extends to making sure gestures produce the expected response within your product. If you’re asking people to swipe, then that action should move elements one way or the other. If you’re asking people to tap on an image, that should probably take you to a screen with more details. Invest the time to think about what people expect from the actions you want them to take in your product. As often as possible, you want them to get exactly what they expect.

IT COMES DOWN TO CULTURE

There’s a reason Facebook had to dramatically alter its mindset to become a great mobile company. More important than having the talent to build something smart and high-quality, it needed this talent to have the time, space and inclination to think through these challenges. After all, Facebook was instrumental in defining the gestures, look and feel, and expectations people now have across the board for mobile experiences.

Each of the lessons above boils down to being mindful about users’ goals, how human attention naturally travels through an experience, how to present clear and logical choices, and how to make a service feel incredibly personal without overstepping.

Beyond that, each of these areas requires constant, methodical experimentation. None of this is possible without refactoring your culture to invest in and prioritize these behaviors — many of which are not immediately productive.

Connected devices have forever changed tech’s ‘move fast and break things mentality,’ Moatti says. When you’re designing a product to extend a person’s identity and capabilities in the world the way that mobile does, it’s better to move slow and deliver delight.