The myth of the modern identity graph by AtData


Over the past decade, marketers have become experts in identity resolution. We’ve mapped behaviors across devices, stitched together customer records, and built sprawling graphs that promise a 360-degree view of the customer.
But much of that picture is built on signals that are quietly eroding. Cookies are on their last legs. Mobile advertising IDs are fading. IP addresses, once useful for geotargeting and household resolution, are now muddied by VPNs and privacy restrictions. Even phone numbers, often used for logins and targeting, are frequently recycled or falsified.
And yet, we keep building.
We keep refining the graph, convinced that if we connect enough data points, we’ll arrive at the truth of who someone is.
However, there’s a difference between a person and the shadow they leave behind. When the signals aren’t stable, when they’re recycled, expired, or easily faked, the story we’re building starts to fall apart. What we’re left with isn’t identity. It’s an outline.
What we lost while stitching
Let’s look at the signals most modern identity graphs rely on:
- Third-party cookies: Third-party cookies are dead. As of early 2025, more than 85% of global web traffic is cookie-restricted.
- MAIDs (Mobile Advertising IDs): Apple’s App Tracking Transparency framework dropped opt-in tracking to just 25% of users. Android is following suit.
- Phone numbers: Carriers recycle millions of numbers annually, but they don’t start fresh. A study from Princeton University found that 66% of recycled numbers were still linked to accounts from previous users, and 9% continued receiving sensitive messages.
- IP addresses: A single household might share one IP across multiple users and devices. Add VPNs and dynamic IP assignment, and the signal becomes increasingly fuzzy.
These aren’t just small cracks, they’re foundational.
The problem is, these identifiers we’ve been leaning on are either going away or were never designed to anchor long-term identity in the first place.
The illusion of a real person
Marketers are targeting personas that appear complete but aren’t tied to anything real. That leads to mistargeted campaigns, flawed attribution, and skewed engagement metrics.
It also creates blind spots.
AtData has worked with clients who thought they had clean, verified audiences, only to discover, after layering in email address intelligence, that 15% or more of their contacts were invalid, temporary, or fraudulent. These weren’t outliers. They were being treated as customers, rewarded with discounts and offers, and counted in performance reports.
Why did these accounts slip through?
Because traditional identity graphs weren’t built to question them. They didn’t look at how long an email had existed, whether it had ever shown signs of real activity, or if it matched known fraud patterns. They just connected what was there and assumed it meant something.
The inbox as identity: What we’ve been missing
In a fractured identity ecosystem, email remains one of the few signals with real staying power. It’s used to log into apps, recover passwords, receive purchase confirmations, and connect across channels. A primary email address often spans years of digital behavior.
But the real value of email isn’t in the address itself, but in what it reveals over time. An inbox carries a history: how long it’s been active, how consistently it’s used, how it behaves across contexts.
When you pay attention, you’re no longer just verifying a contact, you’re evaluating whether the identity behind it holds up. And in a landscape full of disposable signals, that kind of continuity is rare.
Email also stands up to the tests that other identifiers fail:
- Longevity: A genuine inbox lasts far longer than a device ID or session cookie.
- Cross-platform utility: It shows up in CRMs, ESPs, ecommerce platforms, and fraud detection tools.
- Behavioral context: You can track how often it’s used, whether it’s tied to risky domains, and how it fits into the broader customer lifecycle.
MIT research even shows that people revisit the same email threads over months or years, highlighting email’s role as a persistent thread in online behavior.
A better signal yields better outcomes
The marketing upside is clear. According to the DMA, email delivers an average ROI of $36 for every $1 spent, outperforming all other digital channels.
That kind of return doesn’t come from volume alone. It comes from accuracy. From knowing the messages you’re sending are landing in active, trusted inboxes, and reaching people who are actually there.
AtData’s analysis across retail, financial services, and lead generation clients shows that email-based identity, when validated with activity and quality signals, can reduce fraud exposure by 10% to 25%, cut bounce rates significantly, and drive stronger re-engagement performance.
And unlike MAIDs, cookies, or third-party IDs that fade with every platform change or policy update, email-based intelligence is rooted in persistence. It’s consent-based and directly tied to user behavior.
With activity, reputation, and engagement signals layered in, it becomes a foundation marketers can confidently build on, even as the rest of the identity ecosystem shifts beneath them.
Identity that holds up
This isn’t about abandoning the tools we’ve built. It’s about asking them harder questions.
If all we do is connect signals without confirming the person behind them, we aren’t resolving identity; we’re just reinforcing assumptions.
Understanding your audience starts with knowing whether there’s actually someone there. That means grounding identity not in what appears connected, but in what can be confirmed. Not in the shadow of past behavior, but in signals that hold weight.
It’s easy to build around the idea of a person. But it’s much harder to build around a real one.
Ready to ground your identity strategy in something real?
Discover how AtData’s email address intelligence helps you move beyond stitched-together profiles and build marketing programs rooted in truth.