How to identify users and track their behavior even after cookies are regulated

From the perspective of privacy protection, tracking user behavior on the Internet is becoming regulated.

The use of cookies to serve digital ads is one of the restrictions, and Google is also developing a new advertising mechanism to replace cookies. Meanwhile, a new 'mechanism that identifies users without using cookies and tracks their behavior on the website' has been released.

How to track users for analytics in a privacy-first, cookie-less future

In the first place, a cookie is an identifier for a person who visits a website. Cookies allow websites to store user browsing information and input information on the user's PC and retrieve that information from the second and subsequent accesses. It allows you to identify individual users who visit your website, so you can track the behavior of a particular user, but if the cookie is denied, it becomes impossible to track. If cookies are not available, the second and subsequent visits to the website will be recognized as a different person from the first visit.

The European Data Protection Regulation GDPR stipulates that the consent of the user is required for the website to use cookies, and past surveys have shown that only about 20% of the total consent is given. In other words, websites are not assigned identifiers to most people, and many of the user's actions on the website are anonymous.

For websites, knowing 'what the user was doing before the conversion' is directly linked to sales. However, when cookies are not available, it is difficult to know what transitions and behaviors individual users are taking on the page. There are various analysis tools such as Google Analytics, but it is difficult to cross multiple analysis tools.

Meanwhile, Narrator , a data analysis platform provider, explains in a blog 'How to identify users even if cookies are abolished.' The figure below clearly shows the method. The flow is shown in which a person who clicks on a Facebook advertisement and visits an online shopping site purchases a product. At this time, a parameter unique to the URL of the Facebook advertisement is attached, and when the user purchases the product, the user is given an identifier called an email address, and the corresponding email address and the parameter are linked to all the user's actions on the website. It is a mechanism to grasp.

With the method shown by Narrator, it is possible to connect 'information of the user who reached the conversion' and 'behavior of the user before the conversion' and analyze what kind of behavior the user who reached the conversion was doing. When Narrator actually tried this method, client companies were able to bind 95% of their anonymous data to converted users, even after Apple tightened their privacy policy.

Specifically, the following three steps enable user identification.

Step 1: Track all page views and sync them with your 'data warehouse'
Using Google's data warehouse ' BigQuery ', Segmen's 'Analysis 2.0 Javascript library ', Snowplow's ' JavaScript Trackers ', etc., all pageview data is assigned an anonymous ID and saved. If cookies are available, ID consistency can be maintained across multiple sessions.

Step 2: Identify the user with a unique URL
When a user first visits a website, they are anonymous. In this anonymous state, you have two options to know 'who is on your website':

1. 1. Use a unique, non- PII (personally identifiable information) URL
2. Use the identification function of an existing page view tracker

If you are using '2', Narrator recommends '1' because you may get stuck in Chrome extensions. Methods include adding a unique 'order_id' or 'subscription_id' to the thank-you page that appears after purchase or subscription, or adding a unique 'contact_id' to the URL in the email. I have.

Step 3: Connect users with anonymous pageviews
Through the work of Step 1 and Step 2, 'Page views assigned unique and consistent anonymous identifiers for each user' and 'Identifiable user behavior data such as orders, subscriptions, emails, etc.' You can get it.

Using these, first search the data warehouse for page views that include 'order_id' in the URL and get the customer's email address. Find the anonymous ID from the pageview with order_id and overwrite the one used as the identifier from the 'email address' to that 'anonymous ID'. This will allow you to assign a user-identifiable ID to all page views.

in Web Service, Posted by logq_fa