Skip to content
Delivr.ai
Back to Resources
Article4 min read

What Data Neutrality Actually Means

Neutrality, firewalls, behavioural remedies and structural independence are four different things, and vendors use the words interchangeably. What each one guarantees, how long it lasts, and how to tell which one you are being offered.

By Delivr.ai

data neutralityidentity resolutionvendor selectiongovernanceprivacy

“Neutral” has become the most load-bearing word in identity resolution and one of the least precise. In the space of a single vendor call it can mean a sentence in a marketing deck, an internal access control, a regulatory obligation, or a fact about corporate structure. Those are four different promises with four different lifespans.

This page is the definition, written so you can tell which one you are being offered.

The definition

Data neutrality, in identity resolution, is the property that the party resolving your data has no commercial interest in what it learns from doing so.

Not that it promises not to act on what it learns. That the acting would gain it nothing, because it has no competing business to apply it to.

The distinction is the whole subject. Where there is no media business, neutrality is a property of the structure and there is nothing to enforce. Where there is one, neutrality is a commitment being kept — which can be kept well, and often is, but is a different kind of assurance and should be evaluated as one.

What an identity platform necessarily learns about you

It helps to be concrete about the information at stake, because the abstract version makes this sound paranoid.

Resolving your web traffic tells the platform which companies and people you consider worth identifying. Building a segment tells it your targeting thesis before you spend against it. Onboarding a customer file hands it your best customers by name. Running measurement through it tells it what worked and what did not.

None of this is leakage. It is the service functioning correctly — you cannot resolve identity without handing over the thing to be resolved. Which is why the question is never “is my data secure”. The answer to that is almost always yes. The question is what the party holding it does for a living.

Three kinds of neutrality, and only one of them lasts

Rank the available guarantees by what they survive.

A contractual promise. The owner states that the platform operates neutrally. This is worth something: it is enforceable, and breaking it is a commercial and reputational event. It is also unilateral in practice. It holds while the owner wants it to, you have to detect a breach to act on it, and detection is exactly the hard part.

An internal firewall. Access controls, separate teams, separate systems, audit logs. These are real engineering, not theatre, and a serious vendor will have them. They are also controls administered by the company with the commercial interest in the outcome, which is why regulators do not treat a firewall as a sufficient answer on its own.

A behavioural remedy. A regulator orders neutrality and attaches machinery: an independent auditor, a reporting cadence, mediation, anti-retaliation protection. This is meaningfully stronger than either of the above. It is also time-limited by design, and the thing it polices continues to exist underneath it for as long as the ownership does. The Publicis–LiveRamp order runs five years.

Structural independence. The owner has no media business to favour. There is no firewall because there is nothing to separate, and no audit because there is no conflict to review. This is the only one of the four that is not a promise — it is a fact about the corporate structure, and it stops being true only if the structure changes.

Three kinds of data neutrality, weakest firstThree tiers. A contractual promise, where the owner states that the platform stays neutral, holds while the owner wants it to. A behavioural remedy, where a regulator orders neutrality with an auditor and a fixed term, holds for the term of the order. Structural independence, where the owner has no media business to favour, holds because there is no conflict to police.A contractual promiseThe owner states that the platform stays neutral.Holds while the owner wants it to.EXPIRESA behavioural remedyA regulator orders neutrality, with an auditor and a fixed term.Holds for the term of the order.EXPIRESStructural independenceThe owner has no media business to favour.Holds because there is no conflict to police.NO EXPIRY
An FTC consent order is the middle rung: a five-year obligation with an independent auditor, reviewing compliance every six months for the first two years and annually for the remaining three.

How to tell which one you have

Four questions, in order of how much they reveal.

  1. What is the ultimate parent, and what else does it sell? Not the entity on the contract. If the parent sells media, agency services or demand-side technology, you are somewhere in the first three tiers by definition.
  2. Is the neutrality commitment in the contract or in the marketing? Ask for the clause. A vendor that describes itself as neutral and has nothing to point at is offering you the marketing tier.
  3. What happens on change of control? Most identity contracts handle assignment and say nothing about who the assignee is. A termination right triggered by acquisition by a media owner is cheap to ask for before you need it and impossible to get afterwards.
  4. What do I keep if I leave? If the identifiers are the vendor’s, neutrality is academic — you cannot act on a concern you cannot afford to act on. If the anchor is a hashed email you hold, you can.

Those last two are the ones that convert a governance worry into a decision you can actually make. The longer form is in the ownership diligence checklist.

Where this leaves the market

After six years of consolidation, most of the identity platforms a marketer can buy sit in the first three tiers. Publicis owns LiveRamp, Epsilon and Lotame; Omnicom owns Acxiom; WPP owns InfoSum; dentsu owns Merkury; UID2 is administered by a demand-side platform. The ownership map sets out which is which.

That is not a scandal. It is a structural fact that was easy to ignore while the neutral option existed by default, and is hard to ignore now that it does not.

Delivr.ai is in the fourth tier, and the claim is deliberately small: no agency holding company owns us, we do not buy media against you, and we do not compete for your media budget. There is nothing to audit in that, which is the point of it.

See Delivr.ai in action

Book a call and see how Delivr.ai turns your anonymous traffic into actionable pipeline.