In the eleventh episode of AdTech | AlikeAudience, Bosco Lam, Co-founder of AlikeAudience, discusses the new post-pandemic digital consumer landscape with business and technology journalist Duncan Craig.
Tune in to get intriguing insights on:
- • Challenges and changes in the AdTech industry and digital advertising media industry in 2023
- • The emergence of new consumer touchpoints
- • Changes in consumer behavior post-pandemic
- • The importance of audience behavior in driving media planning and data strategy
- • Challenges and risks of data privacy, including the rush for consent and first-party data in a post-cookie environment
- • The importance of data collaboration as a way to overcome potential blind spots in data strategy
Hello from AdTech | AlikeAudience
Duncan Craig: Hello, and welcome to the first AdTech with AlikeAudience podcast of 2023. This podcast is brought to you by AlikeAudience, the premier audience-targeting company with high-performing mobile audience segments. Every month we spotlight leading executives, CEOs, and marketers from industry-leading companies.
My name is Duncan Craig. I’ve been a business and technology journalist for a decade and worked in AdTech content and comms since 2013. We’re aiming to speak to as many interesting people in the AdTech and digital marketing and advertising industry across half the world.
Guest for this Episode: Bosco Lam
Duncan Craig: Today, we’re talking to the co-founder of AlikeAudience, Bosco Lam. Bosco is an addressability working member of the IAB Tech Lab in the US, has expertise in behavioral economics and consumer data. And he’s passionate about empowering marketers to reach their target customers through connecting data and media, and developing privacy-safe audience targeting solutions, all very topical issues in the marketplace right now. Bosco, welcome.
Bosco Lam: Thank you, Duncan. Thank you for having me here today.
Impact of consumer behavior changes and the emergence of new touchpoints
Duncan Craig: Great. I think there’s a lot of turbulence in the market right now. And you really do have to start the 2023 discussion with the macro view. And because AlikeAudience operates in three markets, the US, Southeast Asia, and Australia, I know you’re gonna have a pretty considered viewpoint.
There are a lot of changes in consumer behavior post-pandemic, and we’re experiencing challenging economic conditions, you could say. So Bosco, big topics. Given that landscape, what’s your view on the landscape and on the emergence of all these new consumer touchpoints and attention measurements? Because all these things are drivers for media planning and data strategy, right?
Bosco Lam: Yeah, indeed, I would say 2023 is a challenging year, you know, people have walked out from the pandemic, America and Europe were the first, and Asia is catching up this year. And a few other big topics, you know, the strong dollar is making Americans import good bargains from overseas, with an all-time high trade deficit. A lot of companies are adopting hybrid work arrangements, which allows a lot of flexibility. I would personally prefer a weekday getaway instead of a weekend getaway, which I love too.
There are also some permanent changes post pandemic, you know, people are so used to shopping online, and they are highly aware of health and wellness. So I will say 2023 is the year for retail media networks, particularly for international products. This tells me that pharmaceutical and medical retailers are huge. So, all in all, what I was thinking is happening this year.
Adaptation of the AdTech industry
Duncan Craig: Thank you. I think there’s a lot of change going on, as we all know, wherever you are in the world, especially in the AdTech industry or the digital advertising media industry, How quickly do you think the industry is adopting, you know, to handle these changes and to be resilient in the face of these changes?
Bosco Lam: Well, thank you, Duncan. That is actually a good question. When we look at resiliency, we always think about short, medium, to long term, right? Before we jump into that, I will actually start with our thesis behind AlikeAudience. Everything begins with consumers. Or I use the word audience interchangeably, how they spend attention and money. These behaviors are the fundamental drivers that affect media plans and data strategies, including data collection and collaboration, targeting, analytics, and measurements.
So going back to this year, I think, in the short term, 2023 is a challenging year. Not because of the economy, but also how consumers are spending the money, like their attention and time span is so fragmented. Say, for example, more short-term content and videos, a hybrid work environment like you and I are encountering right now, that creates a lot of mixing moments between work and life.
You just cannot target a business person in the daily 9 to 5 routines. And I think in the long term, we have to look at the tightening of privacy policy. This is an ongoing conversation. And also, there are uncertainties around the big tech arrangements, you know, there are cookies, and antitrust lawsuits.
It is not only in 2023, but it lasts into 2024 or, you know, a few years, but that is coming. So I think, in the long run, data owners, no matter if you are the brands or publishers, you have to first revisit internally how your data strategy says they’re applicable for retaining existing audiences, and to acquire future audiences. In the meantime, we also have to have external partners with data owners that you may collaborate with and find innovative ways to outpace your competitors to reach your audiences. And I don’t think one year will be enough. It is more of a long-term strategy.
Potential risks and blind spots while working with First-party data
Duncan Craig: Fascinating. Wow, there’s a lot on your plate this year, then Bosco. I think you touched on one of the topics there, which is data privacy, there’s a mad rush around the world to get a 360 view of a consumer, or consent issues. And also first-party data, which is the hot topic in a post-cookie environment along with contextual and other targeting mechanisms. Do you think publishers are making the right strategic moves? And what are the risks and the potential blind spots in this area?
Bosco Lam: Yeah, I think this is a day-to-day topic that we speak about with our clients, our partners, agencies, and obviously with publishers as well. They’re having a huge concern with cookie deprecation. It seems like once there was the announcement (from Google to phase out cookies), everyone seems to have stayed away from external data, and focused on their own first-party data sets.
Well, actually, I like the Apple analogy. Duncan, you own an apple, and I own an apple. We both come with our own first-party apples, right? You show me your apple, which becomes my second-party apple. I’m biased if I always explained that my first-party apple is better than my second-party apple. I may have a rotten apple. But Duncan, you come with a fresh apple with a best-before date, and you trace all the way back to the product origin. So a lot of apples here, but I just want to demonstrate focusing only on first-party data will have blind spots.
Don’t get me wrong. Leveling up the first-party data strategy is still fundamental, but data collaboration will bring you a more comprehensive audience view. And it’s only possible when two or more data owners like you and I, be it a brand or publisher, have equally leveled up. So like having fresh data and having legit user consents to share such kinds of datasets. Otherwise, you trade a rotten apple for a fresh one. You will lose a friend like you, you lose trust or lose a good data partner.
Duncan Craig: Interesting analogy. Thanks for that, Bosco. Well, I’m assuming there are still a lot of bad apples out there to weed out of the ecosystem, whether it be via an open web programmatic exchange, media buying, or some other mechanisms. And we know that data clean rooms are having their moment because everyone wants to have the fresh apple, correct?
Bosco Lam: Yes, indeed.
Privacy concerns around customer data collection for brand insights
Duncan Craig: We’ll have to watch that space. I think the other issue for marketers, agencies, brands, and everyone in the ecosystem is defining data collection. I guess we’re, you know, 10-plus years on from the advent of automated digital advertising. The question that remains in my mind is, you know, are we here to collect everything for the sake of collection? We saw the rise and fall of DMPs. For example, you know, do I need to know all about Bosco Lam, to what granularity of exposing that customer data or gaining the customer data on Bosco is required in order for brands to get good insights on their individual? What are your thoughts about this topic?
Bosco Lam: Yeah, this is actually scary. If I’m the consumer Bosco Lam, and everyone is talking about me. Yeah. Oh, well, they are both practical questions. And also, I would say data ethics, like how granular we should track or collect data and how we give back the consent, and also the rights, right? You know, data privacy rights back to the consumers.
So imagine, in the last decade, the industry mentioned a lot about a holistic view of customers. You want to know all the touchpoints so you can attribute your dollar from a marketing standpoint, right? And the complete customer journey would help you plan out your media strategy, your data strategy, and even your sales, like how you drive sales and how you retain a customer.
So data collection has been focusing on understanding users both offline and online. Or even there, we identify user profiles with behaviors before and after they register. So everything has to be down to one-to-one granularity. But today, the challenge is that we have a tightening privacy policy, which is actually a good thing to protect both the company because you know where you should play and how you should respect the customers or the consumers. So having these more standardized and transparent guidelines, we, both as a company and a consumer, think there will be a lot more trust, such that we know a brand, how to use our data, and also for brand, how they are bound by what kind of data that they can use.
Apple’s Attention Tracking Transparency: Need for Data Minimisation
Bosco Lam: And within this boundary, I know, like even Facebook, they suffer from Apple’s latest ATT policy. I would say one-to-one marketing or advertising was still alive. But it is not an easy task going forward. The industry has to accept that it is getting more challenging if we don’t innovate and still insist on one-to-one advertising. So imagine that 9% of your customers, by default, opt-out from direct marketing, even though you have that customer identity, online and offline purchase records. What action can you add to such data? You just can’t.
So to answer your question, collecting the most granular consumer data would raise a lot of risk and costs while the return diminishes. I’ll say in the future, the art is to find the optimal standpoint where the collected data is good enough but actionable with positive returns, obviously, respecting your customers.
Duncan Craig: Especially since Facebook lost 10 Billion US dollars in revenue from Apple’s Attention Tracking Transparency push last year, and there are some people in the industry now realizing that in order to communicate messages to consumers, they might have to do with less data now. That’s what you seem to be saying, right?
Bosco Lam: Yeah, I think there is a common word in the industry, data minimization. I think that it’s actually a good idea, right? KISS, keep it simple, stupid. If you can do more with less, why not? We just do not want to bear the risk or responsibilities of over-exposing customers’ data, you know, not only from the cost standpoint but also from the risk standpoint. And so I will say KISS, keep it simple, stupid. If we can do more with less data, why not?
Managing fragmentation challenges while addressing privacy laws and consent issues
Duncan Craig: Interesting, I guess it comes to the next challenge, which is fragmentation. Whether it’s media fragmentation, channel fragmentation, audience or data fragmentation, we know that everyone is splitting off into the preferred channels. So we’ve got a diverse range of digital touchpoints. Connecting all these dots is challenging, let alone measurement and attribution of the customer journey. And we’ve got these tightened privacy laws, which obviously speak to the issue of consent. How are you managing this challenge or these challenges, Bosco?
Bosco Lam: Yeah, in AlikeAudience, we always encourage our team, our engineers, even we externally share with our clients to rethink, we are not in a perfect world. Imagine one-to-one targeting as playing the darts, throwing your darts; you always want to hit the bull, when you have low bias and low variance. So in technical terms, bias-variance trade-off is a legacy problem to minimize in machine learning, which is what we think is the art in the future of leveraging this technology rather than collecting all the data sets and just go out and throw the darts.
So having high bias and low variance is when you target, say, for example Gen Z, but your campaign ended up reaching Millennials only. The other extreme is low bias with high variance, which is when you know you should target Gen Z, and your campaign has actually reached out to Gen Z, plus some noises of millennials or misclassified baby boomers.
So back in our example, you know, 90% of our customers opt-out with noises in data, we have to identify missing attributes. We have data that is available on smartphones but not other touchpoints, you have mentioned like CTV, and tablets. We have to cleanse them, or we have to collaborate with other transit data owners to enhance our first-party data. This is what we always mentioned to level up your first-party game. The art is how we can leverage the remaining 10% up in data as the seed data such that we can project to the remaining 90% with a constraint of minimizing the bias and variance. This is where machine learning comes into the game of marketing. Well, I think we should have another stand alone podcast, just to deep dive into this session.
Things to consider while projecting Seed-pool data
Duncan Craig: Yeah, I think you’re right, Bosco. Because, you know, it’s fascinating about that 10% of opted in-seed data pool. Can you just explain briefly now how you take that data and project to the remaining 90%? Can you do that?
Bosco Lam: Yeah. First of all, 10% is a hypothetical number as some even have it less than a single digit opt-in data. I just put it in a hypothetical way.
Duncan Craig: I know that could be 1 to 5%, right? Yeah. Okay, please continue.
Bosco Lam: Yeah, in fact, it is challenging working in a very small sample set of data sets in order to project into a larger set or population of data. And first of all, to work on the seed data, we have to understand, by default, if there are any biases already existing in the datasets.
We have a lot of data ethics on that; are we discriminating against certain races, you know, genders or other equality issues. We have to do a lot of filtering on having the data that is a good representation of my existing customers. So we just cannot use that 10%. We have less data after we apply these filters.
And obviously, after that, we have to look into the core attributes that would be helpful in our modeling. Say, for example, if I run a fashion e-commerce, what are the key attributes that induce purchase? There are questions around having the right size of the product and having enough stock, their user journey, and looking at whether they are coming from social media? Are they coming from CTV? So we look at these attributes so that we project to the larger population. Should we invest our media in particular channels, or short-form videos, or the time and date when people are browsing fashion products before they go to bed? So these are some interesting examples that we can derive from having machine learning models.
Digital advertising in 2023
Duncan Craig: Thank you, Bosco. Very interesting, I guess, for the last question for today. What are your other thoughts in your mind for 2023? There are so many issues we could talk about. What issues do you think will define digital advertising this year globally?
Bosco Lam: Oh, wow. There’s a lot to do. Obviously, it is only February. I think time flies already with 10 months left. Well, I would say the market will be as challenging as always, but we’re not alone. AlikeAudience always encourages collaboration. Not only from a data point of view, but also to build trust and transparency. Our ecosystem has to be an open garden, right Duncan?
Duncan Craig: Yeah, right. You’re a strong believer in the open web.
Bosco Lam: Yep. Oh, as always.
Duncan Craig: Great. Well, Bosco, thanks for your time today, and I wish you a very good year. We hope to tap your insight again, whether it be through another podcast or blog at some point. Thanks for being here.
Bosco Lam: Thank you, Duncan.
Subscribe and Stay Tuned!
Duncan Craig: To our audience, thanks so much for listening. Thank you so much for listening. To find the show notes, transcript, and more information about AlikeAudience segment offerings, log on to the website alikeaudience.com. And if you enjoyed this episode, don’t forget to hit subscribe and leave us a review. We’ll catch you in the next session. Bye.
