Episode 133: Perfecting UX Personalization with Michelle Parsons of Product School
Episode 133: Perfecting UX Personalization with Michelle Parsons of Product School
In this episode of Product Thinking, Michelle Parsons, Product Executive in Residence at Product School, joins Melissa Perri to share what makes a stellar consumer product, the challenges of implementing user personalization, and Michelle's experiences at big consumer names like Spotify and Netflix. They also delve into the role of personalization in the dating app Hinge and its mission to create meaningful connections.
Michelle is a product leader with expertise in search, discovery, and personalization within top-tier consumer tech companies. She is also a Mentor at First Round Capital and a Member Board of Directors at Sky's the Limit. Before her current role, Michelle held the title of Chief Product Officer at Hinge. Her dynamic career includes roles as the Product Innovation Leader for Global Kids & Family at Netflix and a pivotal stint at Spotify, where she focused on Product, Personalization, and the Recommendations Platform.
If you enjoyed this episode, make sure to subscribe, rate, and review it on Apple Podcasts, Spotify, and Google Podcasts; instructions on how to do this are here.
You’ll hear them talk about:
[00:50] - According to Michelle, the essential components of a stellar consumer product go down to three main elements. First is user-centricity, as the foremost priority should be recognizing and addressing the genuine needs of the user. The second one is brand-product harmony, where a brand's offline marketing matches the actual product. Ideally, what's portrayed offline should seamlessly resonate with the online experience, ensuring a consistent and consumer-driven focus. And third, personalization or tailoring the product experience to individual preferences.
[06:59] - When diving into product personalization, it's essential to center on the user. Understand the challenges they're trying to overcome and where personalization can tailor their experience, reduce barriers, and enhance satisfaction. Every company's approach differs, but people have shared traits within our demographics. Using these similarities helps to create insightful user models.
[09:54] - Every successful product is built on a foundation of strong partnerships. Think, are you optimizing for quantity, like many likes, or quality, such as genuine matches? Take Netflix: is it about discovering new shows or revisiting familiar ones? Spotify showcases the complexity of this choice. While the homepage promotes familiar tunes, features like Discover Weekly introduce new music based on listening habits. Meanwhile, Year in Review offers an overview of your year's favorites, and Songs You Missed reveals overlooked tracks popular among similar listeners. So clear outcomes and hypotheses guide product development and data insights.
[25:37] - Dating apps like Hinge aim to cultivate genuine relationships. Users seek meaningful connections, hoping to find lasting companionship. If the app successfully guides them to this goal, they'll naturally promote it, adding the platform to their love story. Another example is Spotify and its Discover Weekly playlist, which can spark widespread interest. While platforms like Spotify encourage ongoing engagement, dating apps have a clear endpoint – finding love. However, negative or unsafe experiences lead to user burnout and deter potential users due to shared negative feedback within social circles. In contrast, Hinge stands out by seamlessly merging its branding with its product, ensuring a trustworthy and exciting user journey.
[32:38] - Reducing friction always delivers better outcomes, right? Most of the time. However, the desired outcome shapes the approach. For instance, on dating apps, a single photo prioritizes physical appearance over personality, which doesn’t foster meaningful connections. Therefore, apps like Hinge must constantly ask: What is the user's desired outcome, and how can the product deliver?
Episode Resources:
Melissa Perri- 00:00:37:
Hello and welcome to another episode of The Product Thinking Podcast. Today we're joined by Michelle Parsons, who is the Product Executive in Residence at Product School and a Mentor in First Round Capital. So before Michelle was at Product School, she was the Chief Product Officer of Hinge. She was also the Product Innovation Leader for Global Kids and Family at Netflix, and she worked for Spotify, where she led their product personalization recommendations platform. Welcome, Michelle.
Michelle parsons - 00:01:06:
Hello, thank you so much for having me, Melissa.
Melissa Perri - 00:01:08:
So you've worked at really big consumer names out there. You've got Spotify, you've got Netflix, KAYAK, Hinge. All of these companies are wildly successful. What do you think makes a great consumer product?
Michelle parsons - 00:01:21:
This is a good question. The three things I typically tend to focus on are first is orientation towards what the user actually needs. So a keen focus on a user problem and really a consumer focused outcomes based metric. So we are really focused in on the consumer needs, what they want, and then how do we solve them with our product. I think secondly is really about the integration of both brand and product and how those two things support each other both on and off product. I think oftentimes you can kind of get into these situations where you see amazing marketing or amazing billboards or ads and you come into the product and they feel like two different products. The best parts in my opinion are ones in which the offline representation matches the online. You can really tell that there's a very deep consumer driven then focus internally. And then I think third is really about personalization. How do you actually create a product that is built for that individual that knows and reflects that person is not just kind of built for every single individual out there.
Melissa Perri - 00:02:26:
When you're talking about personalization, I think that's a really hot topic in consumer products these days. I see a lot of people kind of throw personalization at a problem. They're like, oh, we should just personalize it and it'll make the product better. What do you think it takes to nail personalization? Like how should you really think about that as a product manager?
Michelle parsons - 00:02:45:
Yes, I think there's a couple of things that are really important. And maybe I’ll start with an anecdote about kind of like how I even got to kind of where I was in my own kind of mental model of products and building product and approaching products. You know, it was early in my career at KAYAK and it was a different time back then. This was like shortly after the price acquisition of KAYAK. So the team is still growing and scaling. It's a really competitive market, the travel space and there's very little brand loyalty. And so a lot of KAYAK really dependent on inbound search traffic. So someone, you know, maybe typing into a google or a search bar, you know, cheap flights to Barcelona or hotels in New York City. And as a product manager for the hotels business, we were really trying to focus on creating again, and coming back to the earlier points, that really delightful user experience. We had a really strong brand. I think KAYAK did a really fantastic job of articulating its brand message out into the world. And then we paid that back off and product I think really cohesively. But one day I walked into the office and I feel like any product person's always in the front line of defense with their engineering partners. And we had a significant drop in our search, our inbound traffic. And we were just like, okay, you know, thinking caps on, we got to get moving. And we spent several hours just trying to basically debug because we were like certain that it was a bug somewhere in our product, you know, either from a deploy or something of that nature. And what we found out actually, after some of that investigation was that we in fact had no bug, but it was just that google basically decided that they were going to launch and deploy their travel competitor product that really was basically a reskin of KAYAK, but it sat at the very top of their search results page. And for me, that was just a moment where you like, you lose, you know, 10, 20% of your traffic and that has significant impacts on your revenue when you're basically handcuffed into that modality of acquisition. So that was kind of like my own aha moment of, all right, well, we have a fantastic product.
I really believe in our brand and our messaging, the consumer problem that we're solving here. But what can we do to really ensure that our product is defensible and that users choose to come to kayak.com versus inputting a search result into a search engine? And I started thinking a lot about personalization. Well, how do you make it really easy? You can save them time, reduce that friction. And we started playing around with a little bit of lightweight personalization, mostly heuristics. We didn't really have the capacity at the time to be doing anything pretty large like Spotify or the Netflix of the world were doing at the time. But even with our light experimentations, what we were doing was saying, all right, if somebody is searching for a hotel in Orlando during Monday through Thursday, maybe we're going to prioritize business hotels. If someone's searching with the kids on Friday through Sunday or Friday through Monday, maybe we'll prioritize family-friendly or resort type hotels with free breakfast. And maybe the amenities that we showcase on the hotels might differentiate based off of those various search inputs. And that proved to be very successful, but they were experiments that we really couldn't scale at the time. And I just like started to become just extremely interested in learning more about how I could do that at Scale. And so that's really when I moved over to Spotify and so I think really bringing it back to your questions, like what do you need to consider and what do you need to focus on when trying to build out personalization when thinking through these problems? I think it really comes back against the user. What problems are they trying to solve and where will personalization actually add benefit to their experience and make it an experience that's meant just for them to reduce the friction that it takes to access or complete their outcome and then bring them the most delight. You know, when you come to an experience and you're like, my discover weekly, wow, this is just for me. I've discovered a new artist. You tell all your friends about it, right? It's these moments that really bring you joy and it's beyond, it meets your parts more than just a utility, more than just a transaction really.
Melissa Perri - 00:06:35:
So with personalization too, it sounds like it really matters that you personalize the right thing. So how do you figure out what you should personalize? And I guess what products are good candidates for personalization? Like should you personalize everything or is there like a system to actually hone in on what's going to make this more valuable?
Michelle parsons - 00:06:55:
I think that's every company is going to be slightly different. Ultimately, we're all people who are trying to access products. And so our demographics, who I am, where I live, when I was born, all of these things, like what my interests are, what I tend to gravitate towards, all of these different aspects of who I am are kind of like your fingerprint almost, right? That's very unique to me. And of course, there'll be other people who have similar interests and similar backgrounds and similar demographics that we can use to kind of create these bigger models, understandings, representations of people. But I think it really comes down to trying to understand when you're, let's say maybe a dating app, for example, like what's the end goal there? The end goal is to connect to individuals who have mutual interests, mutual compatibility, who will ultimately be able to be successful in a relationship. That's the end goal for the user. And so then we're trying to understand things like mutual interest compatibility. And those are the hard things that we really don't, that aren't as like simple data points, because a lot of those things are just like, are a bit softer and they're really hard to articulate and put in words. But what you can personalize, right, is, all right, if you tend to have information on a profile, what things do you tend to spend a lot of time viewing? Do you spend a lot of time viewing photos or text, or maybe like the kind of the resonate blocks? And might I be able to differentiate your experience versus somebody else's experience by optimizing those particular pieces or modules on a page, you know, in a Netflix capacity? Am I somebody who likes romance films versus horror? And so my home screen might look a little bit different. I think good candidates for personalization are really discovery inputs. So places where you're going to go search results pages, visual images, metadata, things of that nature, that's going to help me make a decision on what to pick or to reach my outcome a lot faster. And again, with more delight.
Melissa Perri - 00:08:45:
So when you're leading a team, you've been a Chief Product Officer over a personalized app and you've been product leader over many of these personalized apps. What's your system for like working with engineers and for working with the product managers and the engineering team to make sure they nail the personalization algorithms, right? I think AI is a huge topic these days. You're going to read a lot of data about humans to be able to make these choices. Like what's the right way to approach that so that you make sure you're getting the best output?
Michelle parsons - 00:09:13:
I think the first and foremost, right, this is always a partnership, always a partnership between tech and product. First on the product side of the house, it's like what problem are we trying to solve? Because you can really optimize for a number of things. Let's take a matching algorithm. Are you optimizing for quantity, like the number of likes? Do I want to send a dozen likes or am I optimizing for quality, which is matches, likelihood to actually match, right? So like that really differentiates how you might approach building something on Netflix. Am I optimizing for discovery or am I optimizing for familiarity, right? So there are different outcomes that I think ultimately will drive the features that you end up prioritizing when considering how you're going to approach building out that algorithm. I think Spotify is a great example because there's a variety of algorithms that power all aspects of the product, right? Everything from like your homepage is just going to be, hey, these are your typical artists or playlists that you go to. There's going to be front and center at the top of the page, but there's also a discovery component here, right? So it's taking a lot of your previous interests, the thing that you've listened to, whether those be things like artists or genre or even like resonance and sound, right? So like the types of music that you're listening to, it might be country or might be pop, but maybe the similarity or the thread there is the beat of the music and that's what you're actually going to personalize for. But there's also playlists on there that are really focused in on just genre. So like your discover giving mix, for example, which is, hey, this is like your pop mix or this is your country mix. But then we have things that are focused in on pure discovery, like your discover weekly, which is, let's take all the music that you've heard before and find you gems, hidden gems that you've never discovered before ever. And then you have things like year in review, which is looking at a macro data set of everything you've ever looked at to draw insights and then, you know, what are the ones I love the most, which was the ones that got away. The music that you didn't listen to, that people with very similar tastes to you listened to that you missed. And so you can really take any kind of angle or approach into these problems by again, starting with your core alchemy or hypothesis. And then I think that dictates what you prioritize in terms of how you're going to approach building and what you're going to look at from a data perspective.
Melissa Perri - 00:11:31:
That's really interesting. I like the ones that got away. I got to look at that playlist. I don't think i've even actually seen that playlist, but I love that type of stuff on Spotify. I think they do such a good job with the personalization on there too. When you were working there, how did you test it and make sure that you were on the right track, you were personalizing for the right things? And did you have any like key aha moments or things you didn't expect?
Michelle parsons - 00:11:53:
So I think the Spotify example, I get kind of bringing it back, but there's a lot of different algorithms are powering a variety of data sets, powering the experience. I actually think there's a really interesting anecdote from early on. So when I first joined Spotify there, you can remember the app back in the day, we had a heart button, we had a thumbs up button, we had a plus sign button. So there are so many ways to give positive signal to the algorithm that you like something. And that was never actually being taken into account in any of the algorithms. So it was really funny because you think that a lot of what we see taking talent was listen, right? So that was like basically it was your soft signal, right? Not implicit signal almost. I listened to this thing, I repeat this thing, so therefore I like this thing. But the explicit signal of adding to a playlist or thumbs up being our hearting was not being taken into account. And there was actually this aha moment that we had, which is like, wait, why are we taking explicit signal into account to integrate into our algorithms? Well, one, it was a little bit costly, but the other thing that we realized is that there were so many ways to give explicit signal and we didn't know what it meant. And so you have thumbs and you have hearts and you have pluses and it's like, wait, well, what does a user actually think that this signal is doing and how do we interpret that? And so actually the first thing that we had to do, we actually got a whole lot of PMs and engineers into our room because they were owned by different parts of the business, like those individual symbols or input items. And we needed to actually consolidate and say, okay, well, how are you going to progress and move forward with one action type? And is that the right move? Should we have more than one action type to signal love versus add the playlist versus X, Y, Z? So we ended up, you actually notice now in the app, it's just the plus side. And so you add it to a playlist or you can add to a favorites playlist now actually, is that was kind of a manifestation of those conversations. But now we also can take that strong signal into place and layer it at the top of the implicit signals like repeats and listens and skips and all these different types of signals that we're getting from users. So I think that's a interesting moments where it's not as obvious, but you see something like, of course you should be incorporating explicit hearts, but then you like pull back the layers, a couple of layers deep, and then you get into like, ah, here are some other problems and that we would basically have to fix before we get there.
Melissa Perri - 00:14:07:
So I have a question that i've wondered for a very long time. Why is there no like thumbs down button on Spotify to be like, the song sucks, you know, like you recommended it to me and it's like, no, I don't like it.
Michelle parsons - 00:14:17:
So there used to be thumbs up and thumbs down back in the day. And the reason why there is no thumbs down is because it is hard to understand, do you not like this song or do you not like it in this context? Right, because that is actually really meaningful when you're thinking about, are you in a playlist that's been treated for you and just a vibe with the other songs and mesh well, or do you actually hate the song and you never want to hear it again? And what's the reason why you don't like it? So sometimes on the surface, again, even with the thumbs up of our heart, right? These internal decisions that you're making as a user and then what we have to be able to understand, you know, that hypothesis that we have to be able to extract from the decision that are being made are really difficult, which is why things like input signals are really powerful here because if I'm sitting there repeating a song over and over, if I'm playing that song every single day, well, that's some very strong signal to the recommendation system that you love that thing. But if I happen to heart it and I never come back to it ever again, well, what is that really telling me? I liked it in that moment for some other playlist I was making. Those are the considerations we have to keep in mind. And I think across the board when you're thinking about both implicit signals, explicit signals, how do you leverage those types of data points to inform the hypotheses that are driving the decisions for the features that you'll be making long-term?
Melissa Perri - 00:15:32:
Interesting. And it does sound like there's a lot of user behavior behind trying to figure out why do people give the certain signals that they do? Like, why do they do a heart? Why do they do a plus? Do they add it? Like, did you do a bunch of user research to get into there and figure out how do we think about ranking the strength of the signals? Like, what did that look like?
Michelle parsons - 00:15:52:
Always, I think there is a room for both data, hard data, and then your qualitative kind of conversations with your users. I think that helps give color to decisions that you can not necessarily extract by just looking at numbers. And so we went to kind of understand our users, like how were they using these different tools and these different moments and why? And so you got some of these things, right? So I thumbed this thing down because I didn't want to hear it in this playlist. And so those were some of the insights that drove us actually to like, we'll just remove thumbs down. You can skip a song completely actually in a playlist. And actually, if you go into Spotify, you can actually say, skip a song in this playlist or skip a song everywhere. I just on everywhere. And so that's kind of the way that we actually were able to kind of get around that particular insight, which was we want to get some optionality here as well. So we're not essentially also is like discounting this particular artist or song across the experience for an individual who might want it somewhere else.
Melissa Perri - 00:16:47:
That makes a lot of sense. So you gave them the options and then you could still use your data. You didn't have to go out and like study every single person's context or anything like that.
Michelle Parsons – 00:16:54:
Correct.
Melissa Perri – 00:16:55:
But to say, okay, now we kind of know the meaning behind this.
Michelle parsons - 00:16:57:
Of course, these things will change over time. And as your products and features evolve, you're constantly going back and iterating and retesting. But yeah, definitely.
Melissa Perri - 00:17:05:
So when you're starting a new company too, and your most recent one before you're joining product school was Hinge, you came in at a very interesting time, right at the pandemic, right? Right. So we're all stuck at home. We can't really go on dates. What a weird time. How did you kind of come in there bringing your experience with personalization, thinking about another market that needs to be extremely personalized, which is dating. How did you come in, get the lay of the land and kind of understand where those algorithms and personalization was at when you entered Hinge?
Michelle parsons - 00:17:35:
I think, you know, Hinge is like, i think more than just the algorithms because a lot of where you start your journey in dating is your profile. It's your onboarding and your profile. It's your representation of yourself in a digital format that is used to inform decisions that individuals are making. And in kind of the dating context, it's a yes or no. And those Yeses or No’s will inform your algorithm and inform kind of who you see and vice versa who sees you. So I think the interesting thing about Hinge is it's, again, another highly competitive market. Dating, there's so many dating apps out there. And dating apps really depend on liquidity of their marketplace. So where the people are is where people go over time. And you really want to ensure that you're solving the core problem, which is you start coming to your product to do what? To get out on a date, to find a relationship, whatever that thing may be. And so we actually started there really more than anything, which is are we actually solving our users' needs or not? And I think one of the things that differentiates Hinge, I mean, there are a number of things. And I think one of the main things was it is a focus on user outcomes above all else. And so when I got there, it was really about how do we ensure that every across the entire company is seeing the same song? And our north star metric there was great dates per user. It's not revenue, it's not MAO, it's not session or time spent.
Of course, all those things are things that we measured. But the core hypothesis or thesis is that if you spend time focusing on your user outcome, which is what they're coming to you, the job to be done, their core job to be done, which is finding a relationship, what are all the things that have to be true then to get you there? And ultimately, that will lead to things like MAO and retention and time spent and revenue and all the other things that the business cares about. When you start with the business outcomes in mind, you pretty much make a lot of short-term decisions so I have to get there. So we started there, we did a lot of foundational research to really understand were people actually being successful and what were their attitudes and behaviors towards the product. And one of the critical things that we found during that study was that only less than 15% of our users at the time were getting to a positive outcome, which was getting out of the date. And so, wow, that is not great, right? And so we had to understand why. And the other interesting thing that we found was that over 60% of our users, when you surveyed them, were feeling burnt out from the experience as a whole, which prompted them to do what? Have negative retention, turn out, right? But the other core thing you saw is like, they were coming back because there was no alternatives. And so these things to me were really big, aha, like, okay, we've got to solve this problem. People are not being successful. They have like really negative attitudes and behaviors towards the experience. And they're coming back not because they love the product, but because there's no other option.
So how do we fix that? And for us then, it started to prompt us to ask these questions such as, well, what's happening over the course of the user journey that's preventing people from getting positive outcomes? And it oriented us all the way back to the profile. And the single thesis really, which is that, how do we get people to more accurately express who they are on dating apps, so that when people are assessing them for compatibility, for mutual interest, that they're having a better and easier time assessing that. So when they give a signal, whether it's positive or negative, so a heart or an X, that signal is stronger for our algorithm, right? So all these things feed back into each other ultimately, and give us a better opportunity or better chances of actually meeting our users' needs. So we took a couple of different swings there, which were really focusing in on some of the things that you can actually get in the product today that you can get out in the real world. Like, you know, we're doing here, looking at each other on video, you're seeing my interactions, you're seeing my facial expressions, my movements, maybe my humor and my voice and things of that nature, and you never got that on dating apps, right? And so we were trying to figure out, well, how do we re-engineer that essentially, in a context where you have a flat 2D device that's not really set up to satisfy that use case very well. Of course, you can add videos, of course you can add your voice, but then the other hurdle is, how do you people to actually do that in a world where that's not the norm? Because you're thinking about the emotional perspective and the vulnerability perspective of an individual who's already having to be judged on their photos.
Now you're asking me to do extra things? Well, that's going to be hard. So a lot of what we did early on was focus on those core problems that then allowed for us to actually really take some significant leaps. And because we were basing everything on those user outcomes, and I think on a foundation of, hey, we want our users to be more authentic. We want to live up to our mission, which is help get users out on great dates and become the dating app meant to be deleted, get people out into relationships, what things need to be true, and then how do we normalize and make some of these behaviors that are normal safe? And so those are the things that we ended up doing. We did a variety of different tests around boosts and it's an algorithm changes to ensure that those people that were adding things like voice prompts or video were actually having that paid off. So that you created this then, it's almost like this normalization flow within your kind of card stack rank. I saw people doing it, so then I was more likely to do it, and then it's pretty as positive flywheel and as positive momentum for the product.
Melissa Perri - 00:22:52:
I think it's really interesting that tagline for Hinge is the app that's designed to be deleted, like you just said. And is that scary to the company, right? Like when you're leading a team around that, where you're basically telling guys on your team, hey, hey, hey, product team, by the way, we don't want people to use this forever, like get off of it. Most, I think, products are out there going, please stay forever and give us money. How do you think about orienting your strategy around that?
Michelle parsons - 00:23:18:
Look, my personal opinion here is that the goal of dating apps really across the board is to get somebody into a relationship. That's the people are coming and spending their time there, they're not, of course, there's always the emotional kind of toll of like, I got to have a breakup, and so I go on there for validation. And those are very human emotions and human things. But I think at the end of the day, people come to the apps because they're looking for partnership, they're looking for companionship. And my perspective and my philosophy is that if we do a really great job of getting somebody to that end goal, what's going to happen is that they're going to go out and talk about the product in a very positive capacity. And not only that, but if we do an excellent job, and let's say that that's their life partner, well, now we're in their story forever. And so 10 years down the line, we're always part of that narrative, we're always part of that origin story, that kind of that meet you, right? That moment where these individuals met and it was kind of the beginning of the rest of their lives. And so then so long as the product keeps up with the data and the users and new needs and new technologies, then you always have free marketing, essentially.
And so it's the same way for Spotify. I discovered this amazing new artist on my discover weekly playlist. And now we're like, well, what's that discover weekly playlist? I want to discover new music. And so it's these moments I think that, yes, some apps like Spotify want to keep you on forever and that's great, they're providing a utility there, right? So on a dating app capacity, there is a finite end. You're not going to date forever, hopefully. But let's say that you have a bad experience on the app and we don't get you to your end goal and you're just constantly, let's go back to that burnout number. Now I'm burnt out. Now I'm telling all of my friends how much this app is horrible. What if I connect you to people that you really don't have compatibility with and you go out on a date and it's just like a waste of your time or it's a bad experience, an unsafe experience? Well, you're going to for sure talk about that. And what's going to happen is you're going to end up negatively influencing a lot of people in your sphere. Because dating is not a solo sport. Dating is always a sport you do with all your friends and your family. And so that's where the influence comes in. And I actually think that's where Hinge really nails it because their branding and their product are so congruent that when you experience the app both on and offline, you're kind of experiencing again, back to my first point, you want those things to be cohesive, your experience on product.
Melissa Perri - 00:25:41:
I think that makes a lot of sense. And one thing that you're kind of honing in on that I want to expand upon for our listeners out there is like Hinge was kind of in decline before the pandemic. Right. I was on Hinge way back in the day when it first came out and it was novel. It was new. Everybody was on it. It was very exciting in New York City. And then Tinder kind of took over as the default dating app everywhere. And it kind of migrated out of this like hookup dating app into more of a relationship hookup dating app. And then during the pandemic, it's like everybody got off of Tinder almost completely, it feels like, and went back to Hinge. And that was wild for me. Like, I reentered the dating scene after the pandemic and I was like, whoa, everybody's back on this app that I used like 10 years ago. What happened? Right. Like, how did Hinge kind of turn it around?
Michelle parsons - 00:26:28:
I think a couple of things, again, like my perspective is that one pandemic really forced, I think, just humanity and people to kind of come to terms with themselves and kind of reckon with their own emotional kind of behavioral health. Right. You also couldn't see people because it was unsafe in some capacities. And so when you're thinking about what matters to you, you're going to invest time in solid, formidable relationships. So whether that be with your friends, your family and so a random fleeing out on the night like might not be as high in your priority list like just going to grab a drink or something. And so I think people started to like reevaluate what they wanted as a result of I think mental health becoming something that was just more prevalent in just like colloquial conversation and relationships being something that people had to reevaluate. Wow. Do I spend time with you or with you? And what's it going to add value to me when I'm in a state where I'm working from home and I'm going from my bedroom to my kitchen? You can only go out for, you know, X, Y, Z contained amount of time. I think that was one thing that we reevaluated our relationship with ourselves and other people. I also think that one of the core things for me is that focus on user outcomes. Hinge did it right. They were really focused on intentionality as the course and it drives the business still. It's really about the app design to be deleted for intentioned dating. That really has been true throughout the ethos of the product because it started with friends of friends. You're less likely to go somebody who you kind of had a third degree connection to or a second degree connection to. So you're more likely to say, hey, not really interested in dating anymore. And so that kind of like I think has carried through and the product decisions and the principally how we approach the product there when I was there was really oriented around that. It was about, all right, if our end goal is to get users out on great dates, what things must be true? It's not a quantity game. So we limited likes. You can only send eight likes per day unless you want it to kind of break the limit and then you can upgrade. Right. And so the limits that were imposed were really meant to help further that notion of intentionality. So when I'm taking my time, there was no swiping. It's a scroll mechanism. You like a piece of content. So now I'm actually evaluating you as a person a bit more fully than making a snap second decision on a swipe app. The onboarding three times as long. And so you actually have to put more effort. You can't just put one photo on Colin Baines and get into the app and use it. And so all these decisions were built around, again, that end outcome. And then I think folded back into the product in a way that helped bring cohesion, again, the brand messaging, the marketing, the actual product, and then focused in on that user outcome.
Melissa Perri - 00:29:07:
I think it's really interesting because what you're getting into right now is kind of different than how we think about user experience. Like what I'm hearing you say is that adding more friction actually produced better user outcomes. So like making it longer on board people, you know, filling out more information, not being able to just like easily click away, like being thoughtful about it actually sounds like it made it better for the users in the long run. And that's not typically how we think about usability. Right. We're like, oh, limit the clicks, limit the amount of time you spend doing something. How did you draw that comparison back to we need friction? What was that signal?
Michelle parsons - 00:29:45:
I think, again, like sometimes we take these like one size all frameworks or approaches and say like, ok, well, the less friction, the better. Well, it depends on what your outcome that you're trying to drive is. And if your outcome is getting used out on great dates and getting them off the app ultimately, then a little bit of friction is important because how are you going to get into a date or a great date or relationship if you have one photo? Well, that's impossible. Cause now I'm judging you based on your looks. It has nothing to do with who you are as a person. And we all know that good relationships are built off multi-dimensional aspects of who a human is and their compatibility with another person. And so philosophically, I tend to approach these problems first and foremost with, I will like, you know, ring the bell over and over again. It's like, what is the outcome that the user is coming to us for? And then how are we best positioned to solve that particular problem based off of the part that we're developing here? What are their problems? What are the opportunities that we have? They're the pain points that exist. And how do we get them there? I'll give you an example, kind of like going back to Netflix. So Netflix itself was really about discovery. I mean, you think about it, you go to Netflix, I'm going to watch Netflix tonight. We go on, you find something, you discover something, create amazing, a new show, a new movie, fantastic. You know, when I was running kids, it was very different, but the entire product was built. In a high friction mode to get them into discovery. Kids don't want discovery. If you listen to Dharta, they know and have kids, they want to repeat the same thing over and over and over and over again, times one billion. And so, and that's developmentally a part of development because repetition helps children learn, but also it's comfort because when you're a kid, you want to feel safe.
So there's a behavioral piece there. And so what we were doing on the product was essentially after a kid had watched our algorithms basically prioritize discovery moments. So once you watch something to its end, it was actually removed from your homepage. And it really only existed in this row called watch it again, which would like flow up and down the page on your Netflix home screen. And it was really, really, really negatively impactful to kids and their parents. When you think about the two user types, one, a kid's dexterity being able to navigate on spell. They can't go to search, like where is the search button? That's an menu that pops-out over here. And so like you but paradigms don't like ring true with a four-year-old, they don't get it. They can't really type that well, they can't spell. And so it's frustrating. And then what's even more frustrating is like, kid can't find the thing that they want and starts to scream and their parents trying to cook dinner or is trying to finish up some work. And the parent has to rush over. And now we've created a bunch of negative experiences. So on that road or in that particular experience, we want to reduce friction as much as possible. How do we then help kids access their favorites instantaneously and delightfully and as the empowered navigator? So they're not having to go ask their parent. But on the business end of the house, that wasn't our goal.
From a business perspective, we really needed kids to discover new content because that's how our business would continue to grow and we would continue to get eyeballs onto the new content that we were investing in. So we had to use, in that particular state, we had to basically use the familiar, how do we get kids into the familiar as fast as possible? And then how do we sprinkle in moments of discovery or moments of impression so that something becomes familiar over time? The more time you spend watching and less searching, the better we are going to be at actually giving you moments where we can introduce new characters and new shows and new content to you. And so that was also a very successful tactic for us. But there's a different end of the spectrum, right? So one is, again, the business outcome wanted that new discovery, but we needed to then focus in on what the user outcomes and needs were, Hinge front, both the business outcomes and the user outcomes were aligned, so that was great. And so, you know, there, I think, again, the approach is you need to come back to what are the goals always in my perspective.
Melissa Perri - 00:33:30:
So as a product leader too, you kind of have to balance the user outcomes and the business outcomes, right? Like you got to make sure your company succeeds, otherwise we all know I have no jobs. But you got to really focus on those customer outcomes, as we heard you say, this whole podcast, which has been really great, to make sure that you're building a quality product. What is your advice for product leaders out there where maybe the business is really pushing on these business outcomes and your CEO and your leaders and your board is like, we got to hit these numbers, we got to hit these numbers and they're looking at those financial metrics and they're very set on that. And you're trying to come in with maybe some ideas that you know will relate to business outcomes in the future but they might be not common ideas or things that are not widely accepted by the people only looking at the business but they focus on the user outcomes. How do you balance those two things and make sure that you're solving for both?
Michelle parsons - 00:34:24:
I always try to connect the user outcomes back to the revenue and to attention. And I think you can do that with just like Metrics trees or KPI trees where you're basically starting with your node. If it's retention, what are the things that lead to retention and can trace that down all the way to all the components? That's one way to visualize, to showcase how investing in one area might actually lead to outcomes that are positive for the business. Now, sometimes there are just the realities like we must hit Q1 revenue numbers or else we have to lay off 20% of the business. Well, sometimes that's the reality. And I think that what I try to do is say, okay, how can we attack this problem while still harming the experience the least as possible? And maybe that means, hey, might show a couple more upsell screens. But the part's still usable. I'm not taking away a feature and then saying, now you had this for free, previously now you had to pay for it. I think we need to be creative when it comes to things like this. Are there people who have never seen a Paywall ever before? And that happens. Even at hinge, we realized there are people who had never seen a Paywall ever. So how do we help them understand the value or the additional value they can unlock through product marketing campaigns or through some pop-ups within the app, obviously in a very intentional way? But I think it comes down to having product principles in place first and foremost. What will you never sacrifice principally as a product team for user outcomes? And so one was like, we aren't going to, our end goal is to focus on intentionality. And so we're never going to do things that adversely impact that particular goal. We're not going to put up a pop-up for a pop-up sake. We're not going to take away features that exist. And so again, everything's for intentionality. You might add some chat limits. So like right now, on some apps, you can only talk to so many people at a certain time versus have 20, 30, 40 people on your match list that you're talking to at the same time. Well, maybe you'll actually apply a limit there. But that's actually still in line principally with intentionality. Because you can't talk to 40 people at one time. We actually saw from research that having 10 plus people in your match list actually degraded your experience. It made you feel more burnt out because you now had to balance so many different conversations and people, and you were spreading your attention so thin. So you might introduce a limit there. Okay, that's looking a very creative solution that does impact the product, does impact the experience, but it's done for the business. It can make the business money, and it can also impact the user's experience, even if they don't immediately understand it. And then how do you explain it? I think that's very important.
Melissa Perri - 00:36:58:
So if you were giving advice to a product leader about this, about getting clear on your intentionality, making sure that you're explaining this extremely well to the board and to the CEO and to your teams, what are some tips you would give to make sure that this is coming across in a good way?
Michelle parsons - 00:37:14:
I think boards and the executive team, not every single board's going to care about the user outcomes. They want to know how is the business doing? Are we going to hit our quarterly projections or not? And so I think this comes down to an exercise of storytelling, really. I love to talk about this like reverse pyramid approach, which is like, here's the state of the world and how do you get more granular and deep into kind of the tactics that you're going to use. And so again, if our goal is revenue, let's talk about that. Also talk about how the outcomes that we're trying to drive for the business are going to directly relate and impact that and showcase what tactics you'll take and how they have impacted previously. If you have data, if you have experimentation, so I'm a huge fan of experimentation and trying to identify your core hypothesis that you can test very rapidly to give you some insights so that you do have that data point to bring to your board, to bring to the executives, that showcase, I understand that it's going to be a little bit scary, but we have proof here, here and here. And here are the experience that we ran and here are the outcome that they drove. And so now maybe we invest a little more time, either downstream or upstream or a bit more resources. And here's what we might be able to expect, the projections. And so I think having those hard examples are really impactful. And then relaying that back up to your macro goals. Again, you need to take into account your stakeholders. Who's your audience? And then sell to them. At the end of the day, we're all salespeople. I actually believe that. And it's our job to really ensure that we're advocating for our users 100% of the time, while influencing and managing up and across so that we are able to basically satisfy both needs.
Melissa Perri - 00:38:50:
Definitely powerful words for leaders out there to remember. My last question for you, Michelle, is about, we've got all this generative AI out there. Sure, you've heard that like 80 times in the past couple hours. A lot of that helps with personalization, helps with other things. What are you excited about? For the future of data and personalization and how it can help consumer apps.
Michelle parsons - 00:39:13:
I think at this point now, we're going to be able to unlock just a lot more in companies that didn't have the capacity to do that before. So I think this is the beauty of Gen. AI and AI data in general is that it was limited sometimes to the company that could afford it, that could invest in it, that had the teams to build out and support these databases. And now we can unlock that for more and more and more companies so that they can personalize things across the board down to copy. I'm going to say that, and I’ll hit this one over the head again, is like, ChatGPT out of Gen. AI tools really unlopped the ability to very rapidly create a bunch of different ways to position things and then test it out into your product. And ultimately, how you are presenting your product is the first experience that people have. Your onboarding screen, what you see in those first, you know, sign up flows, when you check out flows, how you express something and explain something to somebody is really important. And I think over time, we'll be able to do a lot more and more companies will have greater access, democratized almost, to be able to leverage that across their entire user experience, whether it be through personalization algorithms, ranking, communication via copy. And so I think that's going to be really impactful.
Melissa Perri - 00:40:24:
Awesome. Well, I'm excited for that world. I'm excited to see what all this pans out to be. Well, thank you so much for being on the podcast today. If people want to learn more about you and see more of your work, where can they go?
Michelle parsons - 00:40:35:
Yeah, I'm just available on LinkedIn. So you can search for me, Michelle Parsons there. And I share product tips often and different things that I'm working on there.
Melissa Perri - 00:40:44:
Great. Well, thank you so much for listening to the product thinking podcast. And if you liked this episode, make sure you hit subscribe anywhere you're listening to this, whether it's apple or Spotify or any of the other podcasting apps out there. And if you could leave us a review, that would be amazing. Just go over to Spotify or Apple. Rank us, leave us a review. We'll be back next week with another Dear Melissa on Wednesday, so make sure that you go to dearmelissa.com and let me know what you're thinking about and what questions you have. We'll see you next time.