What Is Zero-Party Data?

Marcus signed up for a specialty coffee subscription and told the brand exactly what he wanted: single-origin Ethiopian beans, medium roast, ground for a V60 pour-over, fruity over earthy, delivered every three weeks. His first delivery was a dark-roast Colombian blend. The welcome email recommended an espresso machine.

Most brands collect zero-party data fine. Most brands fail to use it. According to SAP’s 2026 Global Engagement Index, 54% of brands can’t access their customer data in real time, and 66% still rely on third-party data for personalization even as that signal gets weaker every quarter.

Forrester coined the term “zero-party data” in 2020 to describe data that a customer intentionally and proactively shares with a brand. The definition is the easy part. The harder question is what happens after the customer tells you what they want, and why most brands still get that part wrong.

Zero-party data is information a customer gives you on purpose. Marcus telling a coffee brand he wants medium-roast Ethiopian beans for a pour-over is zero-party data. So is a skincare customer selecting “sensitive, combination” during onboarding, or a fashion shopper telling a brand they only wear petite sizing.

It’s declared, specific, and comes directly from the person it describes. That’s what makes it valuable, and that’s what makes ignoring it so expensive.

First-party data, by contrast, is observed. You watch what someone does on your site, track which emails they open, log their purchase history. It’s useful, but it’s inferred. A customer who browses hiking boots might be shopping for themselves, or buying a birthday gift. You’re reading behavior and making your best guess.

Zero-party data removes the guessing. When Marcus tells you he wants fruity Ethiopian lots, you don’t have to infer that from five browsing sessions and a purchase pattern. He just told you.

Each data type carries a different level of accuracy, consent, and shelf life. Understanding how customer data works across these categories is the foundation for any personalization strategy. Zero-party data is the most accurate signal you have access to, but it’s also the most perishable. Preferences drift. What Marcus told you last quarter may not reflect what he wants after he spent a holiday in Italy and came back obsessed with espresso.

A head of CRM at a mid-market fashion retailer built her entire personalization strategy on third-party cookies and lookalike audiences. For three years, it worked well enough. Retargeting kept the funnel full, and nobody questioned where the data came from because the ROAS numbers looked fine on a slide.

That strategy is eroding. Third-party cookies are disappearing across browsers, privacy regulations keep tightening, and the audiences built on borrowed data get less reliable every quarter. The ROAS numbers don’t look as fine anymore, and now somebody is questioning where the data comes from.

The shift to zero-party data is a signal quality move as much as a privacy compliance one. What a customer tells you directly is more specific, more current, and more trustworthy than what a cookie infers from a browsing session three weeks ago. “I want medium-roast Ethiopian coffee for a pour-over” carries more activation potential than “visited coffee category page twice.”

The SAP 2026 Global Engagement Index puts numbers behind the gap. Two-thirds of brands still rely on third-party data for personalization. More than half can’t access customer data in real time. And 60% of all customer data sits unused as dark data, collected but never activated.

Those three numbers describe the same problem from different angles. Brands are collecting more data than ever, relying on the wrong sources, and can’t move fast enough to act on the right ones. Zero-party data addresses the quality issue directly. It gives you the most accurate input available. But without the infrastructure to activate it in real time, it joins the 60% that sits in a database doing nothing.

A preference center is the obvious starting point, and it works. But the highest-value zero-party data gets collected in context, at moments when sharing information feels natural and the value exchange is obvious.

During onboarding, a skincare brand asks about skin type, concerns, and routine. That’s three data points the customer volunteers because they expect better recommendations in return. The coffee subscription that asked Marcus about roast, origin, and brew method did the same thing. Ninety seconds of questions, and the brand had a foundation that would take months of purchase history to approximate.

Post-purchase is another high-yield moment. A cycling retailer asks what kind of riding the customer bought the helmet for. Road? Gravel? Commuting? That one question turns a transaction into a preference profile that shapes every future recommendation.

Loyalty program enrollment works for the same reason. When someone signs up, they’re already signaling interest in a deeper relationship. Asking about communication preferences, product interests, and frequency tolerance costs ten seconds and gives the brand a data foundation behavioral observation alone would never build.

The common thread: customers share data because they expect a better experience in return. Every preference center, quiz, and onboarding flow is a promise. “Tell us what you want, and we’ll use it.” If you collect and don’t act, you’ve spent the trust, and rebuilding customer loyalty after that breach is far harder than earning it in the first place. Marcus told the coffee brand exactly what he wanted, and they sent him a dark-roast Colombian blend. He cancelled before the second delivery.

Every marketer who’s worked with preference data has seen this: a customer signs up for a pet food subscription and says they have a senior Labrador with joint issues. Then they spend the next three weeks browsing puppy food. A loyalty program member selects “vegan” as a dietary preference, then adds a leather jacket to their cart. Someone says they want weekly emails, then stops opening them after the second one.

Stated preferences and observed behavior don’t always agree. And when they conflict, most brands don’t have a system for deciding which signal to trust.

This is where zero-party data stops being a silver bullet and starts being one input among several. The customer browsing puppy food might have adopted a second dog. Or they might be buying a gift. Either way, the preference data alone gives you half the picture, and half a picture is how you end up sending senior-dog joint supplements to someone who just brought home an eight-week-old puppy.

Treating zero-party data as the only signal leads to a segmentation model built on six-month-old survey answers and a preference center nobody updates. Combining it with behavioral data, purchase history, and engagement patterns builds a view of the customer that’s both declared and demonstrated. What they say they want, cross-referenced with what they do.

Every competitor article about zero-party data stops at collection and maybe segmentation. None of them ask the follow-up question that matters: what happens when you connect what a customer told you with what your business data says is possible?

Marcus told the coffee brand he wants single-origin Ethiopian beans. The brand’s supply chain data shows their Yirgacheffe lot is three weeks from selling out, with no restock confirmed. Without that connection, they keep recommending Yirgacheffe until it’s gone, then send Marcus a “sorry, out of stock” email that reads like a breakup text. With it, they email Marcus before his next delivery: “Your Yirgacheffe is running low. Based on your taste profile, we think you’d love this Kenyan Nyeri AA. Want to try it, or skip this month?” Marcus feels informed, not surprised. The brand keeps the subscription alive.

That’s the difference between personalization based on marketing data alone and personalization connected to operational signals. You can recommend based on what the customer told you and also confirm you can deliver it tomorrow, at a price point that works for both sides.

SAP Engagement Cloud connects customer preferences with operational data from across the business. When engagement data sits alongside order history, service interactions, and supply chain signals, the personalization layer has enough context to make promises the business can keep. Enriched customer profiles draw on first-party behavioral data, declared preferences, and business signals to build a complete picture of what each customer wants and what you can deliver. AI marketing takes this further by acting on that combined data in real time.

A CRM manager looking at a dashboard sees Marcus’s declared preferences and last week’s browsing behavior. She can build a segment and trigger a campaign. That’s useful, but it’s also a snapshot of two data sources reviewed by one person on a Tuesday afternoon.

AI can hold declared preferences, observed behavior, and operational signals simultaneously and act on all three in real time. Marcus has told you he likes fruity single-origin coffee. He’s been browsing cold brew equipment this week. Your inventory data shows a new cold-brew-optimized Ethiopian lot just landed. AI-powered segmentation surfaces that match and triggers a recommendation before the next scheduled campaign goes out.

SAP Engagement Cloud uses behavioral predictions, including purchase likelihood, lifecycle stage, and channel engagement propensity, to determine when and how to act on the data a customer has shared. Send time optimization delivers the message at the moment a customer is most likely to engage. Product affinity analysis connects declared preferences to purchase prediction data so recommendations reflect both what the customer said and what the data suggests they’ll buy.

AI doesn’t replace the marketer’s judgment. It acts on the full data picture faster than any manual process, which means the preferences customers share get used before they go stale. SAP Engagement Cloud’s personalization engine is built to do exactly this across every channel.

Marcus cancelled his coffee subscription after one delivery. The brand had his preferences, his brew method, his origin preference, and his frequency. They sent him a dark-roast Colombian blend anyway. At that point, they’d told him everything he needed to know about how much they valued his input.

Three things separate activation from warehousing.

First, audit your current collection touchpoints. Most brands have more zero-party data than they realize, spread across preference centers, onboarding flows, loyalty programs, and customer service interactions. The data exists. It’s the connections between those touchpoints and your activation layer that are usually missing.

Second, connect preference data to your behavioral and operational data layer. A preference on its own is a starting point. A preference cross-referenced with purchase history, browsing patterns, and inventory availability is an actionable insight.

Third, close the loop. Show customers that sharing their preferences changed their experience. The next email references what they told you. The product recommendations shift. The communication frequency adjusts. When customers see the value exchange working, they share more. When they don’t, they stop opening the emails.

SAP Engagement Cloud connects the data customers share with the business signals that make personalization deliverable, from declared preferences through behavioral patterns through operational data. That’s the path from collecting zero-party data to activating it at every touchpoint, and it’s the difference between a preference center that collects dust and one that drives revenue.