Progressive Personalization: Designing a Better Experience

Nadia Udalova, Anne Pascual, Andrew Carey, Pavitra Tandon at UXDX EMEA. Video: https://youtu.be/qXZv3IXJTXk

Readable transcript: edited from the recording's captions for readability (fillers and false starts removed, punctuation and section headings added). Wording is the speaker's own. Timestamps are positions in the video. Names marked [?] could not be verified against the audio.

Showing the value of personalization

[00:00:00] Pavitra: Now it's time for an exciting discussion into the vast and questionable spectrum of personalization. Who's joined us: Anne is the SVP of product design at Zalando, Nadia is head of product design at Moonpig, and Andrew is head of design at Realife Tech. Today we'll be talking everything about personalization. So let's start with this. In your organizations, how do you show the organization the value that personalization brings, and what are the core metrics that you would measure primarily? Anybody who wants to take the floor can start.

[00:00:45] Andrew: Sure, I'll start. At Realife Tech, the main ways that we show the value of personalization are all around engagement, number one, and conversion. Conversion is: did the personalized content have the desired effect on the user? Did they go on and do the thing that we wanted them to do, which is maybe take a particular action, buy something, check something out? And engagement is part of that story as well, because the more people engage with personalized content, the more we can understand that personalization is actually working, as opposed to the A/B tests that we do for people who aren't getting the same type of personalization.

[00:01:28] Pavitra: Okay. Anne, would you like to join in?

[00:01:31] Anne: Yeah, happy to. At Zalando we serve over 45 million active customers in Europe, as the largest fashion platform, and fashion is something very personal. So explaining to the organization, through the research but also looking at customers' behavior, how personal fashion is, is a very important aspect of our work. But we don't need to do a lot of convincing, because in the end, creating a unique version of Zalando for every single customer is our goal: being the starting point for fashion, and also deepening customer relationships.

[00:02:10] In order to do so, you really need to understand the customer, tailor the experience to their needs that change over time, and then measure accordingly whether that has a positive impact for the customer. Which, similar to what Andrew said, is around not only acquiring customers, seeing them come once and purchase once, but rather understanding the behaviors and patterns over time. So frequency of visit, repurchase rate, but also time spent in our app is something that we observe, to make sure it's really the most relevant experience we put in front of them.

[00:02:48] Pavitra: Interesting. Nadia, would you like to tune in?

[00:02:54] Nadia: Sure, just to reflect what Andrew and Anne have said, very similar. But for us at Moonpig, personalization is a really big part of the product strategy. We have lots of data about the core job to be done of our customers, which is about giving the right card and the right gift for a specific occasion. For those who have seen my presentation, we know that this is not the easiest mission for our customers to be on. Therefore at Moonpig, for us it's all about, in the first place, the customer experience: understanding what is the relevant content, what are the relevant products we can offer to our customers. And obviously there is a business side to that, which is conversion, and how often customers are coming back to us, as business metrics.

Validating personalization with customers and stakeholders

[00:03:49] Pavitra: Interesting. My next question would pertain to that. While we talk about stakeholders from a business perspective, your stakeholders are also your customers, right? So when you actually approach them and talk to them, what would you discuss with them to bring out the points, so that you understand these are the personalization features that you need to incorporate and build in?

[00:04:20] Nadia: I can start. As I said, at Moonpig there is very good stakeholder alignment about the importance of personalization. We have done very extensive research within our company, looking at some of those big, successful e-commerce organizations that are really well ahead with the personalization of their content for customers, like Amazon, like YouTube. And there is lots of data that was proven. For example, around eighty percent of customers are more likely to switch to a company with a personalized offering for the experience, and about seventy percent of customers say that they would be switching to those brands who offer a better personalized experience. So there is a lot of data talking about the importance of that. And we do see, from some of the internal testing that we are doing at Moonpig, that whenever we are offering that personalized experience, this definitely has an effect on the frequency of customers coming back, but also in terms of their experience with us.

[00:05:37] Pavitra: Okay. So when you talk about the customer's experience with you, the feedback that you get, how would you share it back with the customers? Or would you do user testing to actually test it out and say, this is something we are coming up with, what do you think of it?

[00:05:55] Nadia: If that question is to me: we're actually doing these studies at multiple levels, both qualitative as well as quantitative. At first, and actually maybe even before that, general discovery, I would say. We do multiple rounds of research in order to make sure we are going in the right direction and we are avoiding being creepy. I think we're going to discuss this further as well. We start from a general landscape, general domain analysis. Then whenever we want to validate that we would want to go one direction or another, we check in with our customers, asking what they think about a particular direction that we want to undertake. And then when we have even more details on top of that, we would go to user testing or A/B testing, to make sure we have multiple layers of validation that we are going in the right direction, that we are iterating, that we are growing our functionalities and our features as expected.

[00:06:53] Pavitra: Sure. Anne, would you be following the same process within your company as well, or something similar?

[00:07:01] Anne: Yeah, we have several mechanisms that we apply to balance the different needs and objectives. In terms of our group strategy, we define customer promises: what is it that we want to deliver to our customers when it comes to flawless assortment, or inspiration and advice, tailored convenience and fulfillment options. Those are all high-level goals that we agree to. Then we break them further down by applying our product development process and working backwards from the customer. So writing a PR/FAQ that has been informed by user research, looking at data, commercial data, customer data, and then working towards a solution that can be looked at in terms of desirability and remarkability.

[00:07:50] Then gathering feedback, similar to the user research mentioned before, where we have a UX carousel, so stakeholders can book time with actual customers to see how they react to certain ideas and flows. But then we also do testing at scale. And lastly, we gather ongoing customer feedback, also in situ in our app, again with qualitative and quantitative sense-making, which is really, really important to close the feedback loop. I do think that with all the wealth of data, it's important to keep it simple, and make sure we look very concretely at what value we create for customers, what they are telling us in return, and how we continuously improve the customer experience according to their needs.

[00:08:39] Pavitra: I love how this is all very customer-centric, and you manage to balance the business with what the customer wants. Andrew, what would your process be at Realife?

[00:08:53] Andrew: Well, most of our clients are not the end users. We build products for our clients, and then they have their own end users. So we have to consider both our clients as stakeholders, and then ultimately the user as well. We have a number of user advocates in the design team and the product team, but managing actual clients and those types of stakeholders is a bit of a different beast. Sometimes stakeholders tell you what they think they want, and our job is trying to lay out everything that's coming towards us, both from clients and from users, and what our own goals are, and try to find a path through it that's going to solve the problems that we're being asked to solve. Sometimes the solutions that we come up with are not what someone's asked for, but a way to address what's been asked of us in a way that takes it further, and solves even a different problem that people may not even realize they're asking about.

[00:09:56] It's a tricky process. We're a small company, and we do a lot of user testing when we can, a lot of stakeholder interviews, a lot of alignment between all the different departments of the business. Usually, when we're trying to solve a big problem, the best way to do that is just to not have any barriers between any of these different groups, so that there's a free share of information, there's honesty and buy-in all along the way in the process, so that we get to the solution that ultimately will be the best one.

[00:10:26] Pavitra: That inter-team communication seems like a dream. There's usually a bit of friction that happens between teams, and some teams get very possessive, I'd say, with regards to the amount of data that they share with the rest. I like the fact that the way you guys work is very open and very honest and very, very collaborative and communicative.

Where personalization becomes creepy

[00:10:55] Pavitra: I've asked you this before, so I'll put it a bit more eloquently this time around. Where would you draw the line between personalization, because there's a lot of tracking et cetera, and being creepy? What would the main distinction criteria be? Nadia, would you like to start?

[00:11:23] Nadia: Yeah, for sure. I think there is definitely a fine line between those. We are trying to make sure that we are always asking customers for their preferences around that, and trying to understand what would be going over the line and what would be a good amount of personalization. To give you an example of what we've been discussing as a creepy experience: auto-rendering content with personal data that was given to us by our customers previously, a few months, years or weeks ago. We had been considering at some point, maybe we will make it easier and use some of the data to help people preview the cards, or show how the pictures could look on a certain product. But then the conclusion was, okay, this probably will be a step too far. So that would be a good example of that creepy experience. We don't want to overuse information that customers have been giving to us, or that maybe we've been collecting.

[00:12:44] I think a good way to deal with this could be progressive personalization. We've been thinking about onboarding: progressively asking customers, with every visit, what kind of offers they might be interested in, or what features they would like to try with us. So collecting their interests piece by piece, and not trying to put all of the personalization, all of the cards that we have, directly at them at the same time, scaring customers away.

[00:13:21] Pavitra: Anne, do you have any comments on this one?

[00:13:27] Anne: Yeah, creating an experience of trust is really, really important to us. So putting the customer in control of their personal data is absolutely a priority and an important part of everything we do. And I would agree with the previous comment made around engaging the customer to play an active role, so thinking about how it can actually be fun to provide some explicit signals around preferences.

[00:13:56] One example is our brand homes, the places where individual fashion brands can share their latest campaigns, collections, et cetera. They are something that customers can decide to follow, and from then on they get notified whenever this brand has something new to offer them. By doing so, the experience gets better and better for the customer, but they also see that they're getting rewarded for telling us what they like, so that from then on, no matter where in the customer experience and along the entire customer journey, they see more relevant products, more relevant stories and campaigns offered to them. So that's been a very positive element that we see customers picking up very nicely.

[00:14:41] Pavitra: That's fascinating, because Nadia, you talk about offers, and Anne, you talk about the brand home. That is such a nice and indirect way of asking for the customer's preferences, instead of giving them a survey or a questionnaire as soon as they open your website or the app, to ask them, "Look, what do you want, what do you want?" That's cool. Andrew, would you have any thoughts on this, personalization versus being creepy?

[00:15:10] Andrew: I do. Just to add to what my fellow panelists have said: really, what it comes down to for us is having a clear, concise understanding of the contract between the end user and the technology. We want to use the end user's data to be able to personalize the experience and provide them something better, something that's going to elevate their normal experience. But we want them to understand exactly what we're doing with that data, why we're doing it, and what value it actually gives them, so they can buy into this process and know that they're giving us something and then we're giving them something back, and we're not just trying to slice and dice their data behind the scenes and do creepy stuff.

[00:15:55] There's also a lot that you can infer from someone's behavioral data inside an app or piece of software, but sometimes it's also just good to ask the user what they want and what kind of things they're interested in, and we've just been talking about some clever ways to do that. I love all that stuff as well, because it just breaks down this barrier, and it doesn't seem like the user's just constantly giving, giving, giving without getting.

[00:16:24] Pavitra: It's like a two-way street. Yeah, two-way love.

Drawbacks of personalization

[00:16:29] Pavitra: Okay, we have an audience question for you guys. What would be the drawbacks of personalization, besides all of the business first?

[00:16:41] Andrew: The drawbacks. Well, some people just want something to do a single thing, and they like the way it is, and they don't want it to get too clever. There are lots of users who maybe, and I have this experience myself, I've been a long-time software user, and sometimes when a company has changed something, and I think it's for the better, users don't want to change. They just want the same thing that they always had, and they're happy with that. Our job, basically, is to stage that into their lives a bit more, and, like I say, give them this context for why it's good, and try to counteract some of that stuff.

[00:17:19] Also, I'm sure we can all look at all the tracking data that we get from the different products that we run and see that there are certain people who just don't opt into these types of data points, like push messaging or location, and they don't want these things to be tracked. So it's really on us to have a fallback experience for those people as well.

[00:17:43] Pavitra: Interesting. Nadia, I see you unmuted. Would you like to add to that?

[00:17:47] Nadia: Yeah, I think the biggest one is the one that we just previously discussed: managing comfort levels. Andrew, you just reflected on that a little bit too. We have to make sure that we are not scaring people away with the sudden data that we make look like personalization. We have seen quite a lot of examples of that on the internet these days, and that was also partly mentioned in my presentation as well.

[00:18:14] But another personalization pitfall could be, well, at Moonpig, the type of personalization we are dealing with. We are also trying not to focus too much on what the same customer was doing last year, for example ordering a gift or a card for a birthday, let's say of his mom or her mom. Because people's behaviors change from time to time, what we notice as well is that if they chose a present or a gift, or wrote this type of message last year, it doesn't necessarily mean that it's going to be exactly repeated this year as well. So it's looking into historical data, but also bringing other data into the attention, to make sure we are still being relevant, and we are also looking at what other people are doing around the same missions or same occasions. So it's not easy, personalization. It's being smart about this, and not necessarily repeating from year to year what you have done before.

[00:19:29] Andrew: It reminds me of something my mom used to tell me, which is, "Andrew, no one ever always or nevers." People aren't always one thing or another; there are a range of different things. We're human and we change all the time, many times in a day, and so trying to be too prescriptive with something might seem a bit overbearing, or might turn some people off.

[00:19:50] Pavitra: Your mom has very wise words, Andrew.

[00:19:54] Andrew: She does, and I'm only realizing that in my adult life.

[00:20:01] Pavitra: And Anne, what do you think about the drawbacks of personalization?

[00:20:07] Anne: In the fashion context it's exactly that: people change over time, their interests, fashion needs change. Style is an ever-moving target. I think ideally personalization doesn't just take a snapshot of a person once, never reconsider it, and keep showing the same things to the customer, but also opens up ideas and imagination about what else could be exciting to wear and to look like. So that's where I think, on one hand, listening to the customer not just once but continuously, and creating this deep understanding that really adapts over time and becomes more and more relevant, and also knowing what is out there that could be interesting for the customer, is important.

[00:21:01] At the same time, a drawback could be if the data quality is not great, when it comes, for instance, to the size information of garments. You could be wrong if you don't triangulate size advice and size recommendations accordingly, based on past purchase behavior and on feedback that customers have given us. That's really where we focus a lot, on fixing some of the fundamental problems when buying fashion, which is super personal and very individual. Hence that's where large and good data sets on products, fashion items et cetera are helpful, so that we can give actual personalized size and fit recommendations.

Narrowing down, and the end of cookies

[00:21:45] Pavitra: That's interesting. All three of you have made the point that people change, and as Andrew's mom said, several times a day. So how would you narrow down? Because if they're constantly changing, then the trend keeps changing at a very, very fast pace. So how do you narrow down what personalization feature you'd like to put forth to the customer? Because even if you put out offers, as an example, you obviously cannot put hundreds of offers out there. You need to narrow down specific offers to target a specific audience.

[00:22:23] Andrew: For us, the way that we manage that is through a system that we call context. It's a sort of governance for the personalization, and it allows us to quickly figure out, based on what the user is currently doing, or what area they're in, or the thing that they're coming to visit in the live event space, what they're trying to do at this moment. That allows us to narrow down the types of things that we can show them. So understanding this context we think is really crucial, because without it you're just sending them something because they meet certain criteria, but context allows us to frame that criteria in a really specific way.

[00:23:05] Pavitra: Interesting. Okay, you've all mentioned that personalization is part of your product strategy. So now, with the change in the rules and regulations, how would your strategy with personalization change with the absence of cookies? Because you won't actually be able to track as much as you would like.

[00:23:33] Andrew: We operate mostly within mobile apps, so the cookie situation is a little bit different there. As I touched on before, a lot of what we're trying to do is be really explicit to the user, in plain language, about what we want to do with the data that they're letting us look at, and let them buy into that contract at face value, so that we're not trying to do too many clever things behind the scenes. And all of the partners that we work with, we try to make sure that they have the same sort of transparency and clarity, so that when we're presenting stuff to the user that may rely on partner data, this consistency and transparency comes through.

[00:24:13] Pavitra: That's interesting. Nadia, after our previous conversations I actually went and downloaded Moonpig, and I have to say I really, really do like it. And because I use the app, I wonder, even on the website, how would your strategy change based on the absence of cookies?

[00:24:32] Nadia: I find this a little bit of a tough question to answer, to be honest. I think it goes a little bit outside of my specialty, so I would probably hold off on comments, just to make sure that I don't say anything that I'm not allowed to mention, or that I'm not really aware of.

[00:24:48] Pavitra: No problem. Anne, would you have anything? Because I use the Zalando app all the time. I'm an avid Zalando user, so I would like to hear if you have any thoughts around this.

[00:25:04] Anne: Can you repeat the question? Is this still around consent?

[00:25:07] Pavitra: No, it's the absence of cookies.

[00:25:11] Anne: Again, I would say asking the customer for consent, and explaining very clearly the value that they get in exchange, is super important. It's the beginning of a trustworthy relationship to have with a brand like Zalando. So that's again where it's not just asking for consent once, but then reminding the customer again and again that we are able to show you these products, et cetera. So it has to come through in everything the customer sees, and they should be reminded of it in a positive, non-intrusive way.

Unique brand experiences and recommendation rabbit holes

[00:25:44] Pavitra: That's interesting. There's a question that's come through around this. With all of this personalization, would you say that it is still possible to create unique experiences that a user would associate with a specific brand?

[00:26:04] Hmm, this is an interesting question. Anne, Nadia, do you want to go?

[00:26:09] Andrew: It is possible, and I think it's happening now. Look at Google, Amazon, Spotify and Starbucks. All of these huge brands leverage the data that we put into their systems to be able to give us personalized experiences, and we're really hooked into that at this moment. Look at things like Google Maps and contextual awareness, what's near you. All of these things are so personalized, to the point now where the personalization is almost invisible, and it's just part of the core experience. You don't even think of it as personalization, because that's just what it does. So I think it's totally possible, and it's happening everywhere.

[00:26:53] Nadia: But there was a part of the question where I'm really interested in what the person who asked this really means. I think it was something about personalization related to a brand. Again, there are a lot of ways of doing personalization and segmentation out there, segmentation in the first place, and I think many organizations are leveraging similar approaches to that. So this is what confused me, the last part of the question. I would like to understand what that part is, to make sure that we answer this question. But I think Andrew covered it pretty well.

[00:27:33] Pavitra: Cool. We have one last question. How would you safeguard against unwanted product recommendation rabbit holes? The person says, "I sometimes miss the randomness and unexpectedness that is often lost with recommendations." Like the YouTube rabbit hole: you go from one into the other into the other. So how would you safeguard against unwanted rabbit holes with personalization?

[00:28:06] Nadia: This is a great question, and I think it's up to us, the designers, the user experience creators, to make sure that, again, as Anne was referring to earlier, we give the rules into the hands of the customers. I am aware that YouTube actually has quite a few settings where you can take certain content out of the recommendations, or block certain types of content. So I think this is a great example of how you can give that control back to the users, and I think we as product designers, as user experience designers, have to make sure that we give our customers the possibility to do so.

[00:28:45] Andrew: Yeah, I agree. I think there's a point where the algorithms will do what they do, but if you get stuck in a really bad cycle of "this is not anything I want to see, how do I make this stop?", there is nothing wrong with actually allowing the user to tell you, "Please make this stop." If it's going to compromise the experience by continuing, you've got to give them a way to bail out.

[00:29:10] Anne: Yeah. One way we also look at this is: what are the things we want to be known for, that we believe in, and that we think are worth calling out to every customer? As an example, we offer mainly fashion, but also beauty, and we know that many customers still need to become aware of that. So communicating additional offers like our beauty experience more proactively is something we would do. Similarly with sustainability: making customers aware of the sustainable assortment, but also our pre-owned offering, and then highlighting through product marketing the trade-in functionality, so that every customer can declutter their wardrobe if there are things there that they no longer wear. So in a way, personalization happens on the product level and experience level, but it also happens in the sense that we want to make customers aware of things that we think are exciting and could be potentially interesting to them, although they may not have thought about it in the first place.

[00:30:16] Pavitra: Fantastic. Thank you, you guys. We've come to the end of this discussion; we are almost out of time. There are more questions coming in, but we'll leave them for another time. Thank you, Anne, Nadia and Andrew, for sharing such insightful information. I'm sure everybody who's been here has a takeaway, and hopefully this has inspired some to start the personalization conversations within their organizations as well. So thank you again, and I'll now pass the baton back to John.