User identification is an integral step to display relevant ads. But how do browsers and apps identify us? Each system has different approaches.
Identification is a procedure by which a user is recognized in an information system. To perform the identification procedure, the subject must first be assigned with an appropriate identifier (the subject has to be registered in the information system).
Types of user identification
Cookies
The main way to identify a user on the web is cookies. This is a small text file that the server sends to the user's browser. Cookies allow us to avoid having to enter a password every time we access a site with authorization. And thanks to cookies, advertising platforms target us.
Cookies are used to identify the user: they track the state of the session (site visits, pages viewed, presence time), record the user's preferences, store personal settings and keep statistics about visits, transferring this data to the server.
Starting with IOS 12.3 Safari restricts the use of 3rd party cookies. Cookies can now only be targeted for 1 day after the first contact. Recently, the Chrome browser, starting from version 76, blocks the use of 3rd party cookies.
Read more about how the world will look like after cookies in our article.
The main mobile identifiers are IDFA and GAID (often referred to as "IFA") for iOS and Android. Unlike cookies, IDFA or GAID are much more stable, as users rarely change their ID, which is changed in the phone settings.
However, there are restrictions on data transfer. Starting with iOS 14, in order to gain access to the IDFA, an application must explicitly request the user's permission to do so. Many of us have already opted in to share data with app developers.
Another way to identify a device is fingerprinting (device fingerprint). This is a user definition based on device configuration information. This method is already used by all major platforms to increase attribution accuracy.
It is most commonly used to attribute app installs or to get bundles of IFAs or cookies. Another option for user identification is to use a pixel on a website.
A pixel is a JS code that is installed on a website or application or delivered along with an ad and tags the user. Allows matching with SSPs and data providers (i.e. controlling the number of contacts with the same user during the advertising campaign).
The pixel is needed for:
- Forming a remarketing audience.
- Accounting for targeted actions.
- End-to-end analytics.
- Building a look-alike audience.
The biggest issue with the pixel is that if the user has not visited the site, you will not be able to identify him without other means of identification.
Segmentation
One we identified the user in one of the above ways and want to show ads to them. To do this, we need to set up the necessary goals in the analytics system in advance (this is important) or, after collecting the data, carry out segmentation.
What are the criteria for segmentation?
Socio-demographic data
Everything is simple here: by gender, age, place of residence, income and education, we can select a group of people that is necessary for us. This is the simplest and oldest segmentation method.
Traffic source
Also a fairly simple type of segmentation, as it allows you to separate users according to the way they got to the site. Source data is collected automatically in analytics systems.
Audience from a website or web page
The collection of the audience will be based on visits to the specified site addresses or pages. Such segmentation is done in order to launch retargeting.
Advertising campaign
With this segmentation, you can set up second-party retargeting, as there is an emphasis on users who have already interacted with ads in some way: they saw, clicked or watched a video.
This is also a kind of retargeting, but more narrowed, since not all users have seen ads.
User Lists
This type of segmentation is a good option if there are device IDFAs, phone numbers or emails, identifiers from the SSP which you want to target.
You can show ads to those users that you have collected in past advertising campaigns, in your applications, offline points of sale or using forms on the site.
Events
Allows you to collect an audience for the collected event on the site. First, it is necessary to create an event in advance, which can be anything: clicking on a button, submitting a form, watching a video, buying a product. This should be pre-configured because events are not automatically recognized by analytics systems.
Look-alike audience
This audience is created based on the one that already exists. The audience is collected on the basis of similar interests and socio-demographic data.
Mixed audience
This type of segmentation allows you to mix different segmentation criteria, creating new combinations of data that are necessary to achieve advertising goals.
Learn more about targeting options in our article.
User identification systems are different and they continue to evolve and change to protect personal data. At the moment, cookies and IFA are popular and reliable ways to "know" the user.
