How does IPv6 impact digital advertising?

A few years ago, while working for a previous employer, I undertook an assignment to understand how IPv6 could impact our programmatic advertising business.

We ran into several IPv6-related issues that affected monetization on our supply-side platform, and we needed to figure out what was going on. Did IPv6 pose a problem for our platform and represent a looming existential threat to all digital advertising? I needed to know.

Thoughts of IPv6 doom and gloom flitted through my mind.

"We are in the middle of an upgrade to a foundational component of the Internet! The value we use to power core advertising features is going extinct. The horror!"

I was off to work conducting product discovery. I began peppering our systems team and external partners alike with questions. What I learned helped form the base knowledge I'll share in this article.

Even though I learned that IPv6 did cause issues, my initial IPv6 learning journey unceremoniously concluded with no product enhancements — because we saw barely any IPv6 traffic. We mostly received IPv4 addresses, the "old" IP address version that IPv6 will eventually replace.

We eventually set up an IPv6 report that spat out the percentage of incoming ad requests to the SSP that contained an IPv6 address and tracked the growing adoption over time. The report would let us know when the problem was worth solving.

Given the crushing weight of our backlog and client demands, we decided there were bigger fish to fry. We would kick the can down the road and save this problem for wiser future iterations of ourselves.

But now it's a few years later, and IPv6 didn't go away, so what's up? Should publishers, advertisers, and ad tech platforms worry about IPv6?

What is IPv6?

The first obvious difference between IPv4 and IPv6 is the format. IPv4 contains four "octets," or four unique values separated by decimals:

192.168.0.1

We call them "octets" because they represent 8 bits of information, totaling up to 32 bits.

In contrast, IPv6 contains up to eight unique values totaling 128 bits of information using hexadecimal format (separated by colons instead of decimals):

2001:0ac8:82a2:0000:0000:0000:0060:7239

This Quora answer illustrates things nicely, but you don't need to understand octets or bits to comprehend the rest of this article.

So why are IP addresses changing? Because devices need an IP address to connect to the Internet, and we are running out of IPv4 addresses. There can only ever be about 4.3 billion unique IPv4 addresses, and our collective need has surpassed this capacity.

When IPv4 came on the scene in 1983, nobody could have imagined the exponential growth and adoption of Internet-connected devices — but in a world where most humans own a device that requires an IP address, we needed an upgrade.

IPv6 has us covered for a while because it can allow up to 340 undecillion addresses or 340 trillion trillion trillion addresses.

The world is slowly moving towards IPv6 support, and the protocol is gaining traction. Google tracks the percentage of users that access their services over IPv6 and makes that data publicly available. As of this writing, 37.69% of traffic Google sees is IPv6 traffic.

You can check if your device supports IPv6 by visiting test-ipv6.com. I have run tests that yielded positive and negative results, so what is happening?

Some user hardware (router, modem, phone, laptop, connected TV) that can support IPv6 can also support IPv4 — this is called "dual stack" support, and it eases the glacial transition to IPv6. A user may emit only an IPv4 on their device one day but then have an IPv6 on the same device the next day.

The same user could also have an IPv4 on one of their devices but an IPv6 on a different device — we will explore below how both these scenarios create challenges for advertising.

Why do ad platforms need IP addresses?

DSPs, SSPs, DMPs, and other ad tech vendors use IP addresses to power various core digital advertising features. These ad platforms receive IP address values in ad requests, bid requests, or any calls to their servers for tracking purposes.

Below, I have detailed some of the most notable features that use IP addresses.

Geotargeting

Ad platforms work with specialized geolocation vendors like MaxMind or Digital Element to map IP addresses to an actual physical location. I wrote a separate article about geotargeting to explain how this process works if you want to learn more.

Typically ad platforms can make real-time API calls to a vendor or self-host a lookup table. Due to the low-latency demands of running real-time programmatic auctions, some ad tech platforms opt to host the database since it is the more performant option.

The good news is that geolocation vendors can still resolve IPv6 addresses to physical locations, so there is not much to worry about here — but platforms may still need to update their integration to support both IPv4 and IPv6 lookups.

Fraud Detection

An IP address is one piece of telemetry fraud detection vendors use to determine the veracity of any ad request. Standard techniques include looking for many duplicate requests from a single IP address or comparing IP addresses from a request to lists of known data centers generating bot traffic.

Legitimate vendors should support IPv6 in any fraud detection techniques. It would be a tough sell for these vendors to say they don't use IP addresses to help monitor fraud for a sizable chunk of traffic.

Frequency Capping

Ad platforms sometimes use an IP address to power cross-platform frequency capping (controlling how often an individual user should see an advertisement). Using an IPv4 address for frequency capping controls how often an entire household can see a given advertisement since every device in a house has the same IPv4 address.

Ad servers, SSPs, and DSPs use device IDs to control frequency capping at the device level — but can also use things like publisher-provided identifiers (PPIDs) to power user or account-level frequency capping. But if other identifiers are not present in an ad request, an IP address always provides a reliable fallback identifier across all platforms.

One potential issue for advertisers with IPv6 is that two devices on the same network can have two different IP addresses. However, the prefixes of those addresses should remain the same and can still allow for cross-platform frequency capping at the household (or network) level.

So if your FireTV had an IPv6 address:

2001:db8:abcd:0022:0000:0000:0000:0000

Then a laptop on the same network could have the following:

2001:db8:abcd:0022:ffff:ffff:ffff:ffff

Attribution

Ad platforms can and do use IP addresses as a cross-platform identifier to tie together ad interactions and conversion events.

For example, an IPv4 address is handy when attributing Connected TV ad views with a desired outcome on another device since each device in a home will emit the same public IP address to the outside world.

So if you watched an ad for Geico on your Apple TV and subsequently signed up for car insurance on your laptop, then Geico (or its ad tech partners) can attribute that purchase to the CTV ad you viewed.

As we learned in the previous section, this practice could remain viable for IPv6 by observing the prefix portion of an IPv6 address.

But things could go sideways for an advertiser if an older IPv4-only CTV device initiates an ad request containing an IPv4, and your laptop connects to Geico's website via IPv6. They now have two disconnected IPs that complicate tying these things together.

Linking together ad interactions without IP addresses requires complex and sometimes questionable cross-device graphs to connect disparate data sources. Universal identifiers present a more effective and sensible way to accomplish the same goal but are still nascent.

IP addresses are virtually ubiquitous and offer a straightforward way to measure advertising outcomes across multiple devices. But just because you can use an IP address to measure attribution doesn't mean you should — there are ethical and privacy considerations to weigh that we'll touch on below.

Audience Targeting

In addition to device IDs and browser cookies, advertisers and data providers curate data sets for ad targeting using IP addresses. Building vast data sets based on IP addresses as the identifier of choice offers cross-platform and scale advantages.

Data sets based on cookie IDs only apply to web environments, and device IDs only work on in-app environments on operating systems that provide accessible identifiers for advertising. Both types of data sets do not work across platforms, but they offer device-level specificity.

There is a perception that IPv6 can be a device-level identifier, which is somewhat true because every device has a unique IPv6. However, manufacturers can implement techniques to protect user privacy by assigning temporary IP addresses that often change to prevent tracking. I anticipate this becoming standard practice due to growing user privacy demands.

But advertisers can still use IP addresses for household targeting, and IPs are available for all devices, inherently offering massive scale for targeting purposes. Employing data sets based on IP addresses allows advertisers to target audiences across all devices and platforms.

I take that back — targeting audiences using IP addresses doesn't work on all devices. Users behind VPNs can hide their actual IP address from prying eyes, and big tech players like Apple and Microsoft have introduced VPN-like services to cloak their users' IP addresses.

These companies rightfully recognize that users have little control over their IP address and look to curtail its usage for advertising. If you want to go deeper on this subject, I wrote an article about the future of IP address as an advertising identifier and the existential threats the practice faces.

Some DSPs implement a platform-wide policy to not use IP addresses for audience-targeting purposes due to privacy concerns or disbelief that the practice will remain viable as solutions to hide it altogether grow in popularity.

Ad tech companies live under a widening lens of privacy, regulatory, and user scrutiny. They may figure that using IP addresses to track users at any level could prove untenable.

But you can still audience target households using IPv6 addresses, right? Sure, you can by using the prefixes of an IPv6.

Audience targeting is really only matching an identifier in an ad request to an identifier in a data set targeted by an advertiser. (Ok, it's more complicated than that, if you want to learn more, check out this DMPs Explained article).

But issues do arise specific to the IPv6 migration we are undergoing. For example, if a user's device supports both IPv4 and IPv6, then an advertiser's DMP and a publisher's SSP may collect one or the other from a user.

If an SSP receives the user's IPv4 in an ad request, but an advertiser collects and stores their IPv6 in a DMP, then there will never be a match when the SSP transmits an IP address to the advertiser's DSP when it's time to buy an ad.

Fragmented support across devices also causes issues. An older IPv4-only device would miss any targeting applied with an IPv6 and vice-versa. Also, IPv6 is still in the minority, so it may not facilitate the scale or effectiveness many advertisers desire.

So IPv6 can work for audience targeting, but some advertising or data platforms may not collect or support it because the proverbial juice is not worth the squeeze.

How should I prepare for IPv6?

Here is my most generic possible answer for publishers, advertisers, and ad platforms:

  1. Look at your reporting data to gauge your IPv6 exposure

  2. Talk to your ad tech partners and understand how they support IPv6

If your business relies on any of the previously mentioned features above, you should at least understand how your ad platforms deal with IPv6. You may find that they are way ahead of you, or you may be surprised to find that they fall on their face at the mere sight of an IPv6.

Publishers and advertisers should understand if their ad platforms support IPv6 for basic features like geotargeting and frequency capping. They should also broach the more complex subjects of attribution and audience targeting if applicable.

If a DSP or advertiser employs technology that can only use an IPv4 address to connect a CTV ad view to a conversion event on other devices, they might ignore bid requests containing IPv6. DSPs may only support IPv4 because they understand the varying adoption rate of IPv6, and the odds of two devices having a matching IPv6 is relatively low and not worth supporting.

So what should you do if your ad platform doesn't support these features with IPv6? Like any ad tech discussion, you must approach the situation with nuance before begging for IPv6 feature prioritization.

Is there a reason to support features based on a value that we all probably should have never used for advertising purposes in the first place? Users can't control their IP, so is it fair or ethical to make it a core part of your advertising strategy?

It makes sense to instead focus on technologies that will stand the test of time. Companies could instead invest in adopting universal identifiers based on consented first-party data like UID2 and PAIR or something out of Google's Privacy Sandbox like FLEDGE that adheres to privacy-by-design principles. Although, supporting IPv6 could bridge the gap before these concepts gain traction.

Publishers may want to assess the impact on addressability that IPv6 causes to decide if there is a looming issue. Ad servers or SSPs should be able to help publishers understand what percentage of their incoming ad requests carry an IPv6. Publishers can further break this data out by device OS and look for correlations to drop in bid rates by DSP.

If you want to be more exact and have access to bid request/response logs, you can examine the effect of an IPv6 in a bid request on metrics like bid rate and CPM. But it would be best to also account for whether a bid request contained other useable identifiers like a device ID or a universal ID since DSPs can use these values to power similar features.

Advertisers who leverage data sets based solely on IP addresses may want to rethink their strategy. These advertisers could see a continuing decline in scale and effectiveness as IPv6 adoption grows — if ad platforms or data providers do not plan on supporting audience targeting using the new format.

So should you worry about the IPv6 impact on advertising? It depends.

Pull reporting data and talk to your ad tech partners to assess your current level of exposure. From there, you can decide to advocate for future-proofing current IP address features to support the new format or shift focus to technologies that accomplish the same thing and may withstand increasingly stringent privacy expectations.

Subscribe to the Ad Tech Explained Newsletter

Ad Tech Explained provides explainers, insights, and analysis on the latest trends in advertising technology.

Photo by Taylor Vick on Unsplash

Reply

or to participate.