If Turning Your Router Off Creates a New Household, Your Identity Layer Is Broken

Cars have regos. Businesses have D-U-N-S. Households have AtlasID.

By Malcolm Treanor, Co-Founder & COO, WINR

Most identity systems are built on things that do not last. Email addresses change or we add another one. Devices get replaced, lost or recycled to family members. IP addresses churn. Yet we keep pretending these are stable foundations for analytics, targeting and automation. Anyone who has actually tried to operate at scale or across multiple geographies knows they are not.

In fact, a significant share of the multi-billion-dollar AdTech ecosystem is built on one of the weakest data points available: the IP address. A signal designed for network routing has somehow been promoted into a proxy for people, households and intent. It was never meant to carry that weight, and everyone in the room knows it.

If your idea of identity starts and ends with an email, a MAID or an IP address, you are building on sand. Cars work in the physical world because they have registration numbers. Businesses operate globally because they have a D-U-N-S Number. Those identifiers exist for one reason: they give systems something stable to point at.

The same argument has already been made for businesses in the context of AI and data trust. Dun & Bradstreet’s Chief Data and Analytics Officer describes it plainly in this piece.

Households have lacked an equivalent. That gap is now becoming impossible to ignore.

What Actually Breaks When Identity Keeps Changing

When models fail in production, it is rarely because the maths is wrong. It is because the underlying data cannot agree on who it is talking about.

Most enterprise datasets still treat identity as something to be re-solved every time a new use case appears. One version for analytics. Another for marketing. Another for fraud. None of them quite line up, and everyone quietly works around it.

When large parts of the ecosystem are leaning on IP addresses that rotate, collapse multiple households into one, or fragment a single household into many, the outcome is predictable. Automation breaks. Measurement drifts. Results look fine in isolation but cannot be trusted over time.

If your systems, agencies and media partners all resolve identity differently, you are not operating a system. You are managing a set of disconnected interpretations.

Most people already understand this intuitively.

_{On the left: how a household actually exists. On the right: how most data stacks try to describe it.}

This is most visible inside Agency Data & Technology Platforms, where teams are expected to reconcile brand first-party data, media platform reporting and clean room outputs that all define households differently. The work rarely creates insight. It exists to paper over incompatible identity assumptions and keep reporting moving.

Household Identity: The Missing Reference Layer

Without a persistent household ID, teams end up using models to solve problems they should not have to solve in the first place.

They want to forecast demand, control frequency, suppress households that have already converted, reduce fraud, or understand whether yesterday’s activity relates to the same home as last month’s. Instead, the model spends its time re-guessing identity from emails, devices and IP addresses that have already changed.

That is not intelligence. It is repeated guesswork, and it shows the moment you look past the dashboards and PowerPoint.

This is not a marketing enhancement. It is plumbing. And when the plumbing is wrong, everything downstream leaks and often stinks.

Without it, teams spend more time explaining results than trusting them.

“It also means your most expensive data engineers stop doing work that should not exist.”

It also means your most expensive data engineers stop doing work that should not exist. When AtlasID is taken as read, they build models instead of fixing identity, rebuilding household logic and explaining for the third time why the numbers do not match.

This is usually where the benefits start to show up first.

Data and engineering teams feel it immediately, because senior time stops being consumed by reconciliation work that adds no value. Brands feel it next, because suppression, frequency and attribution start behaving predictably rather than drifting month to month. Agencies feel it because they finally have a neutral reference that is not owned by a single platform.

None of this shows up as a big bang transformation. It shows up as fewer workarounds, fewer re-runs, fewer meetings and fewer explanations. That is usually the first sign the foundations are finally right.

Why Households Matter More Than Individuals

The industry still talks about individuals because that is how digital identity has been sold. It is also why so many systems struggle.

People move. They change emails. They share devices. They reuse infrastructure. Tablets get handed down to kids. TVs are shared by everyone in the home. IP addresses rotate, merge entire apartment blocks into a single signal, or split the same household apart overnight.

Very little of this behaviour is individual in practice. It is household-driven. Household identity provides a stable, real-world anchor that survives device churn, platform changes and the simple fact that turning your router off and on should not create a new household. This is exactly what AtlasID delivers, built on WINR Atlas.

Consider a simple automotive scenario. A car is purchased in the wife’s name, using her email address and home address. The household’s streaming subscription sits in the husband’s name. To most systems, these look like unrelated people.

AtlasID resolves them to the same household. That allows the automotive brand to suppress short-term CTV advertising to avoid waste, while retaining the household for future replacement or second-vehicle acquisition. The linkage happens at household level, without relying on IP addresses, device graphs or brittle stitching.

Billing, logistics, purchasing decisions, risk and media consumption all operate at household level far more often than vendors like to admit. Households persist even as individual attributes churn.

If you anchor identity at the wrong level, everything built on top of it becomes fragile.

Introducing AtlasID

AtlasID is a persistent household identity reference.

It exists to give organisations a single, stable way to recognise the same household across fragmented data, channels and time. Once assigned, AtlasID can be reused rather than rebuilt for every new use case, which is where most identity stacks quietly bleed time and money. AtlasID is anchored to the household as it exists in the physical world, with the postal address acting as the primary organising signal, using WINR Atlas as the underlying address intelligence layer.

If your idea of identity is an email address, a MAID or an IP address, you are building on sand. AtlasID starts with where people live, not how platforms track them.

Put simply:

• Cars have registration numbers.
• Businesses have D-U-N-S Numbers.
• Households now have AtlasID.

Why This Matters Now

Once models are tied to real money and real decisions, identity problems stop hiding.

They show up as budgets that do not reconcile, suppression rules that quietly fail, forecasts that drift quarter to quarter and senior people spending time in rooms arguing about whose numbers are wrong.

If turning your router off can change what your systems think a household is, that outcome is guaranteed. You can tune models, buy more expensive tools and run more expensive processes, but you will keep circling the same problems because the reference point itself is unstable.

Fix that, and a lot of downstream problems disappear. Ignore it, and you will keep paying smart people to explain results instead of acting on them.

Perhaps think about what it means if your definition of a household cannot survive a router reboot.

Frequently Asked Questions

What exactly is AtlasID?

AtlasID is a persistent household identifier. It gives organisations a stable way to recognise the same household over time, even as emails, devices and IP addresses change.

How is this different from a CDP household?

Most CDP households are stitched probabilistically and rebuilt repeatedly as new data arrives. AtlasID is a reference layer. It defines the household once and allows systems to reuse it rather than constantly re-infer it.

What exactly is AtlasIDWhy anchor household identity to the postal address??

Because the address is the most stable real-world signal households have. People move far less often than they change devices, emails or IP addresses, and the address is where billing, ownership and consumption ultimately converge.

Is AtlasID deterministic or probabilistic?

AtlasID is a persistent reference built from validated address intelligence, with deterministic structure and controlled evolution over time.

Does this replace identity resolution or CDPs?

No. AtlasID sits underneath them. It gives identity resolution, analytics and activation systems a shared reference so they stop disagreeing with each other.

Is this only relevant for advertising?

No. The same household instability shows up in fraud, risk, forecasting, attribution and analytics. Advertising simply exposes the problem fastest.

Why hasn’t this been fixed already?

Because it has been easier to kick the can down the road than to agree on a shared reference. IPs, devices and stitched households allow every platform to claim it is “mostly right” while pushing reconciliation downstream. A persistent household ID removes that ambiguity, which forces harder conversations about ownership, accountability and change.