By Shaun Scott
In my first article, I used Severance to expose how marketing fragments one consumer into disconnected identities. In my last, I argued for putting that person back together. But unifying identity is only half the challenge. The harder question is who brands trust to do it, and why the biggest number in the pitch may be the one that matters least.
Everyone Leads with Size
If you’ve ever sat through a pitch for an identity solution or vendor, you know that they tend to open the same way. Two hundred plus million people, billions of connections, etc. Those numbers usually go unchallenged because they're the easiest thing in the deck to compare.
But size only tells you how many people are in an identity graph. As my colleague Michael Fuchs put it, the number of people in an identity graph doesn’t tell you about the quality of the identity underneath it. It doesn't tell you whether the digital signal you're paying to reach belongs to the person you think it does or whether you’re burning impressions on the wrong member of a household entirely. That decides whether your money works, and it rarely makes the pitch slide.
Meet Tina
Before we get into what questions to ask vendors, let me introduce you to Tina. Tina is 22. She games, and like more than half of 18- to 24-year-olds, she still lives at home. Her household's name and address are verified, but the transactions, utility bill, and credit file all belong to her mom.
So, when Tina's phone attaches to that address, a household-first graph resolves her to the bill payer. You target Tina. You reach her mom Tara, age 69. Every dashboard says the campaign delivered and nobody questions it. But when this happens at scale, the wasted spend adds up.
Questions You Need to Answer
While many identity conversations focus on scale, the better questions are about precision:
How is a digital signal tied to a person? Every major graph has a strong offline core of names, addresses and purchase histories. The hard part is connecting a cookie, device ID or email hash to that person. One of the largest data companies said publicly this summer that confidence in digital signals is low and requires a hybrid approach.
Person first, or household first? Many graphs anchor to a verified name and address, validated by transactions. That proves a real person lives at a real address. It doesn't prove which person in the house is behind the phone.
What gets rejected? Every vendor tells you what goes in. Ask what they throw out. A graph that keeps every assumption always looks bigger.
This Isn't Just a Media Problem
When identity is imprecise, every team pays for it, not just media.
Measurement and finance inherit the error. A 2026 industry simulation found imprecise identity distorted measured ROI by roughly 70%, enough to cancel a working campaign.
CRM and loyalty lose the person they're trying to keep, because the welcome series goes to whoever the household record says is there.
Creative hedges against everyone who might be watching, instead of speaking to Tina.
Owned channels and service send the wrong next message, like a retention offer after a complaint.
When the right person gets the right message, they respond and become someone you actually know. Every resolved person adds a real customer and personality to your owned data.
How Precision Works
Start with the scale, because it's real. Dentsu’s reference base covers 268 million individuals (US Adults 18+) across 130 million households, roughly two people for every household. But the household is the container, not the unit. The person is. What makes the difference is how digital signals get connected to the right individuals. We use our universal digital ID, dentsu.Tag, to support the verification of these incoming signals. Its run in production for more than a decade across 500+ brand sites, building, enhancing and connecting to a network of authenticated people tied to billions of digital signals. Those people signed in and identified themselves, so the connection between signal and person is proven rather than inferred. That verification does two things:
1. It verifies every digital signal before it enters our reference base. Each signal is tested against the network. If it can't be tied to an individual, it's discarded, not downweighted.
2. It resolves anonymous visitors to real people in real time. More than 60% of anonymous site visitors across our brands resolve to an individual in under 200 milliseconds, each match carrying a confidence score the brand can see. That powers personalization, targeting and insights at the person level, on the page Tina is viewing.
For your brand, that means the gap between “someone visited our site” and “we know who that person is” closes before the page finishes loading. You can personalize the interaction in real time, suppress wasted spend on the wrong person, and build a customer record that gets smarter with every interaction. For Tina, that means the brand knows it’s her and not her mom before the page finishes loading. The message, creative, and offer are personalized, and the measurement reflects what’s actually happening.
Your Next Three Moves
- Ask how many digital signals resolve to an individual, not just how many people are in the identity graph.
- Ask what the graph is checked against, and what fails the check.
- Measure precision across the org: CRM match quality, loyalty growth, and how often you message the wrong person.
Scale gets you everywhere, but precision makes sure you're talking to the right person.
