The Rise of First-Party Data: How Is It Rewriting Retail's Business Model?

Konois (Lun Hsieh) was recently chatting with a few retail business owners and noticed a common phenomenon: everyone is anxious about rising advertising costs, yet few truly understand why ads are getting more expensive, or what that price increase actually represents. In fact, this isn't simply a matter of supply and demand—the real reasons are far more complex, representing a massive reshuffling of the entire digital ecosystem.

Over the past decade or more, the logic of digital advertising was simple: use third-party cookies to track users, build user profiles, and then sell those profiles to advertisers. This system allowed the various ad networks, ad tech vendors, and media buying agencies under the Google and Meta ecosystems to rake in huge profits, and it also kept advertising costs relatively cheap—after all, the big players could obtain a complete picture of a user's online journey through relatively simple means, whether through DMP data management platforms or various smart ad-serving platforms, without having to worry much about how the data was acquired. But now, this system, this entire ecosystem, is collapsing. The essence of rising ad prices is that the amount of exploitable user data has shrunk. When data becomes scarcer, advertisers and brands have to spend more money to reach the users who can still be tracked. But that's just the surface phenomenon—the real change is that the underlying logic of the entire retail and advertising ecosystem is being rewritten.

Even before 2020, Apple had already set Safari to block third-party tracking by default, and in 2021 it officially rolled out the ATT (App Tracking Transparency) policy, forcing mobile apps to obtain explicit consent before tracking users. The result? That year, only 25% to 40% of users agreed to be tracked. Google originally planned to phase out third-party cookies by the end of 2024, but in September 2025, it announced it was pausing that plan, opting instead to roll out a new user privacy control system. In other words, Google hasn't actually abolished cookies—it has instead given users a choice —but this has effectively made usable tracking data even scarcer, and is essentially the web world's version of the ATT policy.

However, for the digital advertising ecosystem, another shift in user habits is quietly becoming mainstream. Ten years ago, if you wanted to buy something, you'd open Amazon, eBay, or Walmart's website and search for what you wanted—the e-commerce platform was your "gateway." But now? AI agents are becoming the new gateway. Users open ChatGPT, Claude, or other AI tools and directly ask, "Recommend a good coffee machine for me." The AI will give a direct recommendation based on the user's preferences, budget, and use case. Users don't even need to open an e-commerce platform to get purchase advice. What does this mean for traditional e-commerce platforms? It means traffic is being intercepted by AI systems: users no longer rely on search engines or the search functions of e-commerce sites, but instead rely on AI recommendations. This is a fundamental shift.

In the past, SEO's goal was clear: get your website to appear on the first page of search results. The related advertising goal was equally clear: get ranked near the top of the charts, or simply hijack ordinary keywords to rank higher. But as AI becomes the primary gateway to information, traditional SEO strategies are starting to fail. AI doesn't recommend products based on keyword matching, but based on contextual understanding and users' deeper needs. No matter how well-optimized a website is, if it lacks sufficient first-party data to back it up, an AI system won't be able to accurately judge whether that website can truly meet the user's needs. In other words, SEO is shifting from "keyword optimization" to "data reputation optimization"—your website's value in the eyes of AI no longer depends on how many backlinks or how dense your keywords are, but on whether you can provide real, verifiable data on user satisfaction.

In the age of AI recommendations, the most valuable asset isn't ad inventory, nor is it traffic—it's real user behavior data. Imagine a scenario: a user buys a coffee machine on Amazon, leaves a review, clicks on related products, and even adds coffee beans to the cart without checking out. All of this data gets recorded by Amazon. The next time this user opens ChatGPT and asks, "Which coffee bean brand should I buy?"—if ChatGPT can access Amazon's first-party data (of course, within the bounds permitted by privacy agreements), it can provide a much more precise recommendation.

This is why Amazon's advertising business is projected to grow from $4.9 billion in 2023 to $6.8 billion in 2025, making it the world's third-largest advertising platform. Amazon isn't just selling ad inventory—it's selling precision. The first-party data it controls gives it a level of accuracy that no other advertising platform can match.

Let's break this logic down into a simple framework: first-party data determines the precision of AI recommendations, precision determines user conversion rates, conversion rates determine advertising ROI, and ROI ultimately determines the price advertisers are willing to pay. Every link in this chain reinforces the others. Retailers that hold more first-party data can provide AI systems with more training signals, making the AI's recommendations more precise. The more precise the recommendations, the higher the user conversion rate. The higher the conversion rate, the more willing advertisers are to place ads on that platform, and the higher the price they're willing to pay.

Conversely, if a retailer lacks sufficient first-party data, it can't train a precise AI recommendation system. Without precise recommendations, user conversion rates are low. With low conversion rates, advertisers are unwilling to place ads, or are only willing to pay very low prices. Such retailers will eventually be marginalized. This is why Konois says: if you missed the trend of rising ad prices, that's fine, but you absolutely must get your data management and retail business model transformation sorted out. Because rising ad prices are just the surface—the real change is the intensifying competition over data.

Let's look at how global retail giants are handling this. Amazon treated first-party data as a core asset from the very beginning. It records not only users' purchase history, but also browsing behavior, search keywords, cart contents, and even return records. This data is used to train recommendation algorithms and is also sold to advertisers as precise audience targeting. The result is that Amazon's advertising business has become its fastest-growing business, with an annual growth rate of 23%.

Walmart launched the Walmart Connect platform, integrating its massive offline physical store data with its online e-commerce data. It has even installed smart shelf screens in physical stores that can display personalized ads based on a user's online purchase history. This kind of omnichannel data integration allows Walmart to offer services that traditional ad platforms simply cannot. Carrefour launched the Carrefour Links platform, focusing on leveraging its massive member data. It provides advertisers with precise audience insight reports to help them better understand consumer behavior. What do these three companies have in common? They all treat first-party data as a core competitive advantage, not a byproduct. They are all shifting from "selling products" to "selling data services."

While helping several retailers with digital transformation, Aotter has seen some successful cases, and has also run into some pitfalls along the way. Let me share a few practical insights. Many retailers still treat their e-commerce platform as the final destination, hoping users will purchase directly on their website. But in the AI era, your e-commerce platform is just a data collection point, not the end of the sales journey. A user might see your recommendation through an AI assistant, then see your ad on social media, and finally complete the purchase on someone else's platform. But as long as that user's behavioral data gets recorded by you, you can use that data to improve your recommendation system and attract more similar users. In other words, you need to shift from the mindset of "I want to sell things on my own website" to "I want to collect user data and then influence user decisions across every touchpoint."

This is not a simple task. You need to integrate all your data sources. A user's behavior on your website, their behavior in your app, their purchase records in physical stores, the information in your membership system—all of this data needs to be integrated into a unified user profile, which requires robust technical infrastructure and strong data governance capabilities. Second, you need to ensure data privacy and security. In today's world of increasingly strict privacy regulations like GDPR and CCPA, you must ensure users' data is properly protected. This isn't just a legal requirement—it's also the foundation of user trust. If users don't trust you, they won't authorize you to use their data. Third, you need to build a data management system capable of continuous learning and improvement. The value of first-party data doesn't lie in how much data you currently have, but in how quickly you can learn from that data and apply those learnings to your business. This requires a technical architecture that can support rapid iteration.

The traditional advertising model is: I have an ad slot, I price it based on factors like placement, time slot, and audience, and then I sell it to advertisers. The new model is: I have a user base, I can predict their purchase intent with 95% accuracy, and I price based on conversion rate. This is a fundamental shift. The former is "supply"-based pricing, the latter is "performance"-based pricing. The former has limited profit margins, while the latter can achieve profit margins of 70% to 90%. To achieve this shift, you need to be able to demonstrate your precision to advertisers. This means you need to build a complete attribution model that can clearly track the causal relationship between ad placement and user conversion. This requires a combination of data science capability and ad tech capability.

There are several reasons we emphasize that now is a critical moment. First, the pace of AI adoption has exceeded everyone's expectations—ChatGPT reached 1 million users in 2 months, and Claude is also growing rapidly. This means the shift in user habits is happening much faster than we imagined. If you're not yet preparing for the AI era, you'll soon be left behind. Second, competition over first-party data is intensifying. Major platforms (Amazon, Walmart, etc.) have already begun fully leveraging their first-party data. If small and mid-sized retailers don't start building their own first-party data infrastructure now, they'll find themselves completely unable to compete in the future. Third, the trend of rising ad prices will continue. As third-party cookies are gradually phased out and privacy protections strengthen, usable user data will become increasingly scarce. This means advertising costs will keep rising. If you don't change your business model now, your advertising costs will climb higher year after year, eventually eating into your profits.

But conversely, if you start building first-party data infrastructure now, you can gain a competitive advantage in this new era. You'll be able to provide more precise recommendations, attract more users, and achieve a higher advertising ROI. This way, even as advertising costs rise, your overall profit won't decline—it may even increase.

What Konois wants to say in closing is this: the future of retail doesn't belong to whoever has the most physical stores, nor to whoever has the cheapest products—it belongs to whoever can best leverage first-party data, and that includes not just online e-commerce but also offline physical stores. In an era where AI recommendations are becoming mainstream, e-commerce gateways are losing their dominance, SEO structures are changing, and user touchpoint pathways are being reshaped, first-party data is a retailer's lifeline. If you master first-party data, you can train a more precise AI recommendation system. With precise recommendations, you can attract more users and achieve higher conversion rates. With high conversion rates, you can demand higher prices from advertisers and earn higher profits—the old-era ROAS metric is gradually becoming less of a true indicator, after all, third-party tracking attribution is just an estimate, and human motivation is easily over-optimistically estimated under the standards of the old ecosystem.

If you missed the trend of rising ad prices and didn't adjust your advertising strategy in time, that's okay—but you absolutely must get your data management and retail business model transformation right. Because that is the fundamental factor determining whether you can survive in the AI era. Only when the roots run deep enough will the fruit be truly beautiful. First-party data is your root.

Sources:

1. Apple ATT Policy Impact - Singular
2. Google Cookie Deprecation Timeline - Cookiebot
3. Amazon Advertising Revenue Growth - Statista
4. Retail Media Network Overview - AnyMind Group
5. RMN Deep Analysis - Effect Studio
6. Global Retail Media Market - WARC