What types of data can be collected from browsers to perform web analytics

Nowadays there are not question about "Do we need to have data about our digital customer?" But there are questions about 1- what kind of data could be available and 2 - how to collect and interpret to perform business needs efficiently. In this article I'll try to figure out with first question.

Let's take a look at schema of data exchange between client device and web site server. What we should notice is that before browser reveal the site information there are 2 type of interactions. It is browser web API and HTTP protocol

1. https://developer.mozilla.org/en-US/docs/Web/HTTP/Overview

2. https://developer.mozilla.org/en-US/docs/Web/API

Both of them could provide important information. First, let consider HTTP and for demonstrate i will use node js http client and Chrome browser installed on Windows.

As you can see the both side information is similar. And now we have valuable client information about watched url, connection status, browser type,  language, operation system. Though we can get more additional information like client IP address if we use advanced headers like  'x-forwarded-for'. For analytical purposes we are interested in next headers, 'Cookie', 'Referrer' , 'Sec-Fetch-User', 'Location' - they allows us to get more information about sequence of interaction with site pages on client side.

Collecting and analyzing information coming from HTTP already could give us the picture of what site pages clients saw, in what order , what browser type etc. But it not enough for more deeply understanding client behavior and particular study of interaction with content on the page. To collect that type of information we need to use browser API.

Right after client browser have connected to the web site server via HTTP, it return html page. And now browser try to draw this html. After this we have html models that allow us to interact with them and collect additional information like clicks, navigations, scrolls, history etc This models - are browser API.  For more information visit spec documentation https://html.spec.whatwg.org/#webappapis

We could send that information each time we want register an user interact with content and emit javascript events https://html.spec.whatwg.org/multipage/webappapis.html#globaleventhandlers. However it coud work just if user set to allow javascript on his browser.

Indeed, site visiting wont get info about client phone number address or another personal information. But it provide enough information for quantitative analyses and could answer the questions, what kind information more appropriate to what user has behavior for marketing reasons