White Paper · State of AdTech 2026

The IP Data Layer

Published Sep 28, 2026

Advertising technology has spent a decade losing its signals. Cookies survive in name and fail in practice. Device identifiers sit behind prompts that most users decline. The bid request itself remains self-declared.

Through all of it, one signal has remained constant with every request, on every platform, in every market, because the internet cannot deliver a packet without it: the IP address.

Download the White Paper Preview the Findings

Built on the IPinfo AdTech Benchmark Analysis, 2026 — live programmatic supply, measured at production scale.

IPinfo Research

White Paper · 2026

The IP Data LayerFour opportunities to build better advertising data, with a signal you already get

ipinfo.io

100%
of filtered supply samples had resproxy traffic

~100 km
bound per millisecond of latency

5% of IPs
carry half of all web requests

The problem

An Ecosystem of Partial Views

A DSP sees bid requests but only for the auctions it enters. An exchange loses sight of the impression after the auction clears. Publishers see their own traffic and rarely anything upstream of it. Every participant operates on a partial view — each accurate, yet still incomplete.

Where IP-level decisions occur

Bid request
Auction
Impression
Measurement

Request-level attribution

Publishers see their own traffic and rarely anything upstream of it.

Markers show the points where IP-level context changes a decision. No single participant sees the whole path from request to impression. Click a stage to see more.

A preview of what's inside

Four Opportunities, One Signal You Already Get

Traffic quality, location, connection context, and place are each drawn from the IP address already present in every bid request. Every figure is sourced and numbered in the full white paper.

Opportunity 01 · Traffic quality

Traffic Quality You Can Measure

Residential proxies reach the bid stream on genuine consumer IPs that geolocate correctly and pass every standard check. With resproxy detection, weigh how recently an IP was seen as a proxy and how often, so you filter persistent proxy exits pre-bid while keeping real users. Clean supply can then be priced at a premium, with quality proven to buyers in data instead of claims.

50%+of resproxy traffic in the most exposed sample had already passed standard checks^[4]

[3] IPinfo AdTech Benchmark Analysis, 2026 — unfiltered feed
[4] IPinfo AdTech Benchmark Analysis, 2026 — post-verification supply samples

Share of bid requests flagged as residential proxy traffic

Standard IP categories alone<2%

With residential proxy detection~20x

Directional, on billions of unfiltered US bid requests. A flag denotes exposure of the address, not a verdict on any single request. [3]

Opportunity 02 · Location

Location You Can Verify

Declared bidstream location is often inherited from whichever database most vendors agree on, and consensus quietly erodes accuracy for everyone. Check it against measured evidence instead, such as network latency and observed infrastructure, and mismatches surface before the bid. Geo-targeted campaigns then reach the places they paid for, and supply with verified location can command the premium it has earned.

~100 kmthe bound each millisecond of round-trip time places on an address^[6]

[6] Method and definitions annex

How measurement locates an IP

The bid request declares Las Vegas. Probes ping the address from many cities; each millisecond of round-trip time bounds it to ~100 km, and the circles can only intersect where the IP physically is.

Opportunity 03 · Connection context

Context Without Identity

One IP address can mean one device, one household, or thousands of users behind a carrier gateway. Treat them all the same and frequency caps misfire, reach is miscounted, and heavy shared addresses look like fraud. Device-count and connection-type context turn that assumption into a variable, so you can pace, cap, and value each IP by how many people it actually represents — without identifying any of them.

5%of client IP addresses carry over half of the world’s web requests^[5]

[5] Method and definitions annex

One address can mean

One device
One household
Thousands behind a carrier gateway

Every question that matters is a question about the connection, never about the person using it.

Opportunity 04 · Place

Places You Can Target

Venue-level context associates a connection with the hotel, airport, campus, or stadium it serves, at building-level precision across web, in-app, and CTV. Without it, a traveler in an airport lounge and a coffee shop customer look like any other connection, and the moment they are in goes unpriced. With it, buyers can target audiences by the places they occupy, and sellers can package venue inventory as a premium, privacy-safe segment.

No GPSno device identifier, no individual profile

From position to place

Geolocation: Las Vegas, NV
Places: convention center

Hotels · Airports · Stadiums · Campuses

Get the full report

Download The IP Data Layer

The 16-page white paper carries the full benchmark analysis, the four opportunities in depth, and the role-by-role matrix of how each player uses the IP data layer.

The residential proxy visibility gap What standard traffic-quality checks see, and what clears them
Four seats, four responses How an exchange, an ad network, a media-analytics firm, and a verification platform act on the same flag
Why consensus erodes location data And what measurement-based geolocation replaces it with
Method and definitions GIVT, SIVT, and how residential proxy exposure is measured here