gianlucacarrera - Gianluca Carrera

Understanding Customer Churn and How to Fight It

Customer churn is and will always be a critical metric for product success — sometimes it can even be a nightmare! Let’s try to understand its complexities, and to provide some actionable insights for product managers to effectively mitigate it. What is Customer Churn? Simply put, customer churn is the rate at which customers stop doing business with us. It is really that simple, but it is also much more than a metric: it reflects the health of our product and services. High churn rates usually point to significant problems with user experience, product-market fit, or customer satisfaction (include pricing in the list). It’s fundamental to understand the root causes of churn: is it technical issues, lack of features, poor customer support, weak customer experience, laborious user journeys, pricing, or something else? Data are once again key in our search for an answer: analyzing churn through data-driven insights can reveal patterns and provide a roadmap for improvement. Customer Feedback is key Understanding customer feedback is key to addressing churn. Collect and analyze feedback frequenatly and constantly to extract invaluable insights into customer needs and pain points. It is not just quantitative data like, for example, usage statistics, but also, and as importantly, qualitative feedback through surveys, interviews, user testing sessions. Never underestimate the importance of qualitative feedback: as product leaders, we must support a culture where every piece of feedback is seen and interpreted as a great opportunity for improvement. Product Development Reducing churn is intimately and deeply linked with strategic product development. Prioritise features and improvements that deliver against customer expectations and needs — don’t waste time on anything else! Mind you: it’s not just adding more features (this could sometimes be a waste of resources); it’s also, and sometimes even more importantly, about enhanching user experience and satisfaction through product enhancements, minor or major as they might be. Rigid product roadmaps are always the enemy, and in this case even more so: roadmaps should be flexible enough to allow for rapid changes based on customer feedback, market trends, competitive analysis or strategic considerations. Customer Experience Improvement A great, rewarding and delightful customer experience is fundamental to fight churn. But mid you: it goes well beyond the product itself and it extends to all the touchpoints a customer has with the company (it’s not just the product, remember). It’s imperative to create seamless, engaging, delightful and value-driven user journeys. Continuous updates, great customer service, fast and efficient onboarding can positively impact churn. As product people our mission goes well beyond managing products, and it extends to creating and developing long-lasting, value driven and delightful customer relationships. Reducing churn is not a finite, one-off project: it is a continuous, never-ending process of questioning, researching, learning, adapting, and improving that requires a holistic approach involving product development, user experience, customer feedback, user research, and stretches to finance too (balanced usage of resources available and proper prioritization of initiatives). It’s an art supported and informed by science, data and experience. This is what makes product management so interesting and rewarding.

Triggering Platform Network Effects with Data

Let’s take a very broad definition of a platform: an infrastructure designed to facilitate interactions between producers and consumers. So Amazon is a platform, where producers make goods available to shoppers for purchase. But so are Facebook, Instagram and Youtube, where producers of content make it available for users consumption. You could swap producers for sellers, and consumers for shoppers. eBay is then a platform, where owners of goods put them on sales for shoppers to buy. And you could swap producers for developers, and consumers for users. Facebook again comes to mind, where developers can produce apps that can be used by users. But also Apple App Store comes to mind, where developers create apps that get installed by users on their mobile phones for later use. And let’s take at a pretty broad definition of network effect: the increase of value to users of a product or services as its adoption grows. In simple words, the more users use your products, the more valuable it becomes to other users. Do no confuse network effect with virality — they are not the same thing. They can come together (lucky days!), but don’t necessarily do. Simple example of a network effect: the phone. A single phone in the whole world is no use to anybody, but two can be used by two for a 1 to 1 call. Three can be used by three people for three different connections, while four can be used by four people for six different connections, and so on and so forth. Network effect can be created by features. For example, the ‘message’ feature in Facebook facilitate a network effect because it allows you to get in touch with somebody, and the more people you are connected to, the higher the value of such feature — like the phone call feature. So there is an incentive to users to connect to multiple users, which increases the value of the feature, and the value delivered by the platform to all its users. It is clearly better to be able to message ten people, than just one. But data also can create network effect, although in a more subtle, yet very powerful, way. Let’s look at Netflix, for example. It’s not really a platform in its fullest sense: content is bought by Netflix and made available to its subscribers for consumptions – but it works well for our example. Of course, the more the subscribers, the more capital Netflix has to buy more content. But that is more of a scale effect, than a network effect. How can you increase the value of consuming content on Netflix by way of just simply consuming content? You could do that with a suggestion algorithm. As hundreds if not thousands of titles are available on Netflix, picking the one that meets our tastes could be quite laborious — the proverbial needle in a haystack. But as we watch more content, we are giving away information on our preferences, and such preferences can be used to ‘profile’ the user, and therefore identify its preferences. The more people watch content, the more accurate the profiles become, and the more value to the user we can create by identifying close profiles and suggest titles that have not been watched by that particular user, yet have been (and rated favorably) by other users with similar profiles. It’s called collaborative filtering. The more people watch content, the more relevant are the content suggestions offered to us, and the more enjoyable is the experience, as we spend less searching, and more watching. And this clearly is a network effect, produced by data. There are additional benefits of using data in a platform business, which we will talk about in a future post.

Data Monetization: What’s Really Holding You Back?

I finally found time to expand a bit on a previous post about what’s holding up data monetization. Many of you reached out with interesting questions and observations, especially around the three key barriers I identified: resources, culture, and commitment. After years leading data initiatives, I’ve seen these barriers play out repeatedly. Let me share some concrete examples and practical insights from the trenches, starting with what seems to be the most misunderstood barrier: resources. RESOURCES: The Missing Pieces Resources encompasses far more than budget – it’s about the right blend of capabilities. While organizations often focus on technical resources like data engineers and data scientists, successful data monetization requires a much broader ecosystem of talent: Beyond human resources, successful data monetization demands significant investment in: The challenge isn’t just acquiring these resources – it’s orchestrating them effectively. At dunnhumby, we learned that success came from building balanced teams where technical capabilities complemented business acumen and industry expertise. CULTURE: The Silent Killer Culture is perhaps the most insidious barrier to data monetization. In my experience at dunnhumby, Yahoo! and other organizations, I’ve seen how cultural resistance can undermine even the most sophisticated data initiatives. Data monetization cannot be relegated to a specialized team – it needs to permeate the entire organization. Each function takes hundreds of decisions daily that could benefit from data-driven insights. From supply chain optimization to customer service, from product development to marketing, data should inform every significant business decision. The shift from gut-driven to evidence-based decision-making represents a profound cultural change. Consider these fundamental shifts required: The most successful organizations build what I call a “data-first reflex” – where reaching for data becomes as natural as reaching for your phone. But this doesn’t happen overnight. It requires: COMMITMENT: The Foundation Trust me on this one: if you’re not ready for a multi-year journey, you’re setting yourself up for failure. I learned this lesson building Walmart Luminate at dunnhumby, and I’ve seen it play out countless times since: half-hearted commitment leads to half-baked results. You can’t treat data initiatives like typical IT projects. This isn’t about installing new software or updating systems. It’s about fundamentally changing how your organization creates value. At PubMatic, our commitment to data-driven advertising meant completely rethinking our approach to publisher relationships and product development. It wasn’t always comfortable, but it was necessary. Here’s what real commitment looks like: The most dangerous approach? Claiming data is a priority while treating it as a cost center. I’ve seen too many promising initiatives die this way. You need to back your commitment with real investment – in people, technology, and organizational change. But let me be clear: commitment doesn’t mean blind faith. Every initiative needs clear success metrics and regular evaluation. At dunnhumby, we tracked everything from data quality scores to client adoption rates. The goal isn’t just to invest in data – it’s to create measurable value from that investment. TRUST & COMPLIANCE: The Foundation of Customer Data Value I’ve learned firsthand at dunnhumby, working with retailers like Walmart and Tesco, that trust isn’t just a compliance checkbox – it’s a fundamental business asset. When dealing with customer data, trust becomes the bedrock of sustainable monetization. Consider this: at dunnhumby, we tracked billions of transactions representing millions of customers’ shopping behaviors. This massive data asset was only valuable because customers trusted retailers with their data, and retailers trusted us to handle it responsibly. One breach of this trust would have undermined years of value creation. Building trust requires: The most successful organizations don’t view compliance as a constraint but as a catalyst for building better data products. They understand that customer trust, once lost, is nearly impossible to regain. Moving Forward The path to successful data monetization isn’t mysterious – it’s methodical. Start by honestly assessing where you stand on each barrier: Successful data transformations rarely happen by accident, they are the outcome of great planning and hard work and commitment. They require deliberate action on each of these fronts. What’s your experience? #DataMonetization #Innovation #DigitalTransformation #Leadership #Data #Privacy

ChatGPT vs DeepSeek – what it means

Some of you have asked about my take on #DeepSeek’s recent announcement, and it’s sparked some interesting thoughts about where the AI industry might be heading… If DeepSeek’s claims about #R1 training costs hold true – suggesting they achieved similar capabilities to GPT #o1 at apparently 1/10-1/20th of the cost – we might be approaching a fascinating inflection point in AI economics. Let’s explore what this could mean: If this plays out, the commoditization of AI models wouldn’t just be an opportunity – it would be an invitation to reimagine every industry, every workflow, every customer interaction. Just as cloud computing democratized infrastructure and led to an explosion of innovation, cheaper AI could unlock possibilities we haven’t even imagined. The game might be shifting from “who has the best model?” to “who can create the most value with these models?” – and that would be a game anyone could play. What’s your take? How would your business strategy change if AI capability becomes truly democratized?

Data Monetization: Three Routes to Value Creation

Data monetization remains a hot topic in boardrooms, yet many organizations still struggle to extract significant value from their data assets. Despite the enthusiasm around data monetization and AI, why do so many companies still fail to capture meaningful value from their data? Drawing from Barbara Wixom’s seminal work at MIT CISR, particularly her IWS (Improve, Wrap, Sell) framework for data monetization, let’s explore three proven routes to create value from data, enriched with real-world examples from my experience working with some of the world’s largest organizations. Barbara also leads an outstanding executive course on data monetization at MIT – I had the privilege to attend it recently and would strongly recommend it to anyone interested in this space. Improve: Using Data to Enhance Internal Operations ‘Improve’ represents the foundation of data monetization, and research shows it still accounts for roughly 50% of all value extracted from data. Think about it: how much value is trapped in your organization’s data, waiting to be unlocked through better internal decision-making? The ‘Improve’ route includes everything from optimizing supply chains to enhancing customer service, from refining product development to streamlining operations. While less headline-grabbing than external monetization, it often provides the quickest returns. During my time at dunnhumby, we worked with a major retailer to leverage their customer transaction data to optimize their private label portfolio. By analyzing purchase patterns across customer segments, we identified significant gaps in their offering, leading to the development of new product lines that generated significant incremental revenue and customer satisfaction. Have you fully explored how your internal data could drive operational improvements? Wrap: Enhancing Products and Services with Data The ‘Wrap’ route is about augmenting existing products or services with data-driven insights. What if your core offering could deliver significantly more value through data enhancement? Think of it as a data-powered enhancement layer that transforms a basic product into a premium, insight-driven solution. This approach leverages existing customer relationships while creating new value streams. Stop for a second and think to this twice: leverage existing customer relationships for improved product and value creation, leading to increased revenues. No new customers needed, no new products needed. At Yahoo/Overture, we pioneered this approach in digital advertising. By wrapping our ad platform with rich audience data and performance analytics, we transformed a simple ad placement service into a sophisticated performance marketing solution. This ‘data wrap’ allowed advertisers to optimize campaigns in real-time, increasing both value delivered and revenue per customer by an order of magnitude. In fact, today’s digital advertising industry would be unrecognizable without these data wraps. What started as simple demographic targeting has evolved into sophisticated, real-time optimization powered by rich data layers. The entire programmatic advertising ecosystem is built on wrapping inventory with data to enhance its value. Sports broadcasting offers another compelling example. What started as simple live video feeds has evolved into rich, data-enhanced experiences. Real-time player statistics, motion tracking, win probability calculations, and augmented reality overlays have transformed significantly how we consume (and enjoy) sports. Think about it: would you watch a Premier League match today without expected goals statistics, player heat maps, or real-time performance data? These data wraps have become so integral to the viewing experience that they’re no longer just enhancements – they’re expectations. And they have contributed to a significant increase in rights value, and customer satisfaction. When was the last time you evaluated how data could enhance your core products? Sell: Monetizing Data as a Product This most ambitious route involves monetizing data or data-derived products as standalone offerings. Could your organization’s data be valuable to others in your ecosystem? This requires a fundamental shift: from seeing data as a byproduct to viewing it as a product in its own right. Success demands not just high-quality data, but also sophisticated product management and deep understanding of customer needs. And much more (check Barbara’s work for details). It’s why many organizations aspire to this model, but few execute it successfully. Leading the development of Walmart Luminate at dunnhumby exemplified this approach. We transformed Walmart’s rich customer data into a suite of insights products for CPG manufacturers. But perhaps most importantly, Luminate became the foundation for a new way of working between retailer and CPGs. The data and insights became the common currency for negotiations, enabling more fact-based discussions and collaborative planning with significant benefits for both sides, and substantial value creation for customers. What would it take to transform your data into a product that others would pay for? The Path Forward Research shows successful data monetization rarely follows just one route. However, the dominance of ‘Improve’ (50% of value creation) underscores a crucial point: before looking at external monetization, organizations should focus on extracting value from data internally. It’s the low hanging fruit, requires less capabilities, smaller investments and has lower risk. The key is creating genuine value for users. The core of data monetization isn’t about selling data – it’s about transforming data into insights that drive better decisions and outcomes. What’s holding your organization back from extracting more value from its data? This post draws from Barbara Wixom’s research on data monetization at MIT CISR. Her work continues to provide valuable frameworks for understanding how organizations can create value from their data assets. If you know some examples of successful data monetization in your industry, please share to everybody’s benefit.

3 Data Don’ts That Will Cost You

Here are three critical mistakes I keep seeing companies make with their data: Don’t believe quantity makes up for quality Don’t skip the foundation work Don’t build without clear business goals Your expensive data lesson? I’d love to hear it. #Data #Analytics #DigitalTransformation #ProductStrategy