Device ID

A device ID is a unique, anonymized string of numbers and letters that identifies a single mobile device: a smartphone, a tablet, or a wearable such as a smart watch.

What is a device ID?

1:16

A device ID is a unique identifier linked to a particular mobile device. It’s anonymous, because it doesn’t contain any personally identifiable information (PII) about the user, such as their name, email, address, or credit card number.

In the past, any app installed on the device could retrieve the device ID, allowing marketers and developers to measure campaign and in-app activities without access to users’ personal information.

However, that changed with the introduction of Apple’s App Tracking Transparency (ATT) framework, which now requires app owners to get permission to access an Apple user’s device ID. This shift towards a privacy-first world brings challenges for marketers looking to measure and optimize their campaigns. With less data available on individual users or devices, aggregated data is becoming increasingly important — we’ll explore this in more detail later.

Different types of device IDs

There are two main types of device IDs: Apple uses the ID for Advertisers (IDFA) and Android uses the Google Advertiser ID (GAID). They essentially work in the same way to connect a user’s actions with an ad campaign, an install, or in-app activities. The key difference today is that for Apple devices, marketers and app owners can only access the IDFA if the user has opted in.

The Apple IDFA comprises eight characters, a dash, and then three sets of four characters. All letters are upper case. Here’s an example:

The GAID follows the same format but uses lower-case letters, as follows:

What is a device ID used for?

The device ID is an important tool for attributing marketing activities and mapping the user journey — from ad engagement, through install, to in-app events. Here’s how it’s used:

Marketers can use a technique called deterministic attribution to track a user’s actions across multiple channels and interactions. Matching the device ID to each interaction provides a detailed and highly accurate picture of user behavior.

The device ID allows marketers to personalize the user experience by offering relevant ads and services based on user behavior and preferences.
It makes it possible to segment users by device type, usage patterns, and more for sharper campaign targeting.
It gives app owners a better understanding of how much users are engaging with their app by collecting in-app event data. This allows them to pinpoint when, where, and why users engage the way they do: whether they drop off or churn at a certain point, or continue down the funnel and become loyal, revenue-generating users.

How does it work?

The device ID can be retrieved by any installed app, once it’s launched for the first time. (As mentioned above, this is only possible on Apple devices if the user has consented to tracking.)

After the initial launch, the device ID can be used for attribution purposes by measuring the install and tying it to previous activity.

Let’s take attributing an app install as an example.

A user clicks on an ad for an app, and gets taken to the relevant app store to download it.
Once the app is installed and launched for the first time, a mechanism known as the attribution SDK (software development kit) is triggered and records an install. It then goes back through to search for a matching click or view ID on its database.
If the SDK finds a match within the attribution window, the ad is attributed with getting the user to install the app.

How to find your device ID

Finding your device ID is straightforward, whether you have an Android or an Apple device.

On Android, open your phone keypad and enter *#*#8255#*#*. As soon as you type the last digit the GTalk service monitor device will pop up where you’ll be able to see your device ID.

For an Apple device, you can download “My device IDFA by Appsflyer” from the app store to discover the device ID.

The impact of privacy changes

Since the release of iOS 14, with its increased focus on user privacy, Apple has required apps to ask users to opt in to sharing their IDFA.

Previously, Apple users had to opt out of tracking by enabling LAT (Limited Ad Tracking), which caused their IDFA to display as a string of zeros. The current ATT, on the other hand, means users can choose to opt in.

The proportion of users who do opt in to tracking varies by app type, but 2022 figures show an average global opt-in rate of 46%. That leaves a significant proportion of devices that can’t be tracked for marketing purposes. But it’s not game over for marketers: new tech solutions make it possible to measure campaign success without compromising user privacy.

The future of measurement with (or without) device IDs

With device-level data now in shorter supply, marketers are turning to aggregated data to inform their iOS campaigns. With this approach, data is consolidated across groups of users, revealing broader trends rather than individual actions.

As marketers adapt to the privacy-first landscape, various alternative methods have emerged to help attribute marketing activities.

SKAdNetwork

This is Apple’s privacy-centric framework that aims to measure app installs and campaign performance without compromising user anonymity.

While SKAdNetwork offers a solution based on aggregated campaign data, it comes with various limitations and complexities that marketers need to navigate. For example, postbacks (reports on activity) are delayed and there’s a limited number of them, focusing on the early stages of the user journey. This makes it harder to measure lifetime value or re-engagement activity.

Working with a mobile measurement partner (MMP) can help you get more from SKAdNetwork: they’ll handle the data securely and provide in-depth analysis and insights.

Machine learning and predictive analytics

Machine learning algorithms can help us understand trends in user behavior, so that we can predict how valuable these users might be over time, and whether a campaign is likely to be successful.

Probabilistic modeling can make predictions based on aggregated data, revealing behavioral trends while protecting user privacy. Combining these techniques with machine learning is a powerful and accurate way to understand user behavior and predict campaign success early on.

Incrementality

Incrementality testing uses a control and test mechanism to show the true value of your marketing efforts. In particular, it can tell you what portion of business was the result of a campaign, and what would have occurred organically.

This additional layer of intelligence can give you greater confidence in the success of your campaign.

Web-to-app flows

Web-to-app flows take a user from a webpage to a corresponding app.

In the context of iOS 14.5, web-to-app flows help marketers connect the dots without the use of the IDFA. This is because the journey includes ad networks and owned media, so first-party data can be used to optimize the experience.

Key takeaways

While device IDs used to be central to measurement and optimization in the mobile ecosystem, the shift to a privacy-first landscape has pushed app marketers to explore new solutions.

A device ID is a unique identifier linked to a particular mobile device. There are two types: IDFA for Apple and GAID for Google.
Device IDs have historically been one of the most accurate and reliable ways to match a user to an action, and as such were a key method of measurement for marketers and attribution.
Device IDs help marketers to map the user journey and understand a user’s preferences. They also enable detailed segmentation, personalization, and behavioral analysis.
Under Apple’s ATT privacy framework, app owners have to request access to a user’s IDFA on iOS devices.
With device-level data more limited, marketers are turning to aggregated data to measure, attribute, and optimize campaigns anonymously. Apple’s SKAdNetwork helps with attribution, but comes with some challenges. Additional solutions include machine learning, incrementality, and web-to-app flows.