Metrics and groups

Версия на русском языке

User agent

User agent is a header that a browser, program, or application sends to a server when accessing a website. The user agent typically contains the browser name, its version, the engine on which the browser is based, the operating system and its version. The user agent can also contain the name and model of the mobile device on which it is running and some other information. In other words, it is a browser's business card.

The problem is that this business card is very easy to fake - using extensions or even just in DevTools. Almost any programming language has the ability to pass any information to the user agent when accessing a server. Therefore, user agents should be treated with caution.

Normally, the browser does not lie about itself. There are, of course, exceptions, such as extensions that help change the user agent, or mobile browsers or applications in which an unchanged user agent is embedded because developers are too lazy to change it. But such cases can be ignored - they are either isolated or can be added to the exceptions.

It's another matter when the transmitted information does not match what we learned about the browser as a result of checking, and this pattern prevails on a slice - through the user agent, we see a variety of browsers, but the analysis shows that these are all accesses from the same browser.

Therefore, the user agent is an "unreliable storyteller" and the information we obtain from it should be treated with a healthy dose of skepticism.

Metrics

Metrics are parameters that allow you to answer the question 'how much?' How many visits did the website have? How many of them were made by bots, and how many by real users?

General metrics

Name	Description
Hits	Total number of checks.
Unique users	Number of unique users counted based on IP and user agent.
Good	Number of checks resulting in non-bot, non-suspicious, and non-technical loss traffic.
Tech. losses	Number of clicks that we couldn't verify. Reasons include browsers with disabled JS, outdated browsers, etc.

Bot visits (BV)

A bot is a program or automated script that requests web content (including digital advertising) without user involvement. It could be, for example, a browser on an infected computer that is part of a botnet, or a script written by a programmer that scrapes websites for some purpose.

Name	Description
BV total	Number of clicks made by bots of all types.
Crawler	Legitimate bots that honestly declare their botness through the user agent. These could be search engine bots indexing the web, social media bots, etc. Despite not having malicious intent, they cannot convert and are not the target audience, so they should be ignored.
Spoofing	Bots whose user agent does not match the actual browser version and/or operating system.
Automated	Bots using headless browsers or automation tools like Selenium. The original purpose of such solutions is to automate software, website, application testing, etc. Then bot operators realized they could also use them for click fraud, view fraud, etc., and added them to their inventory.
iFrame	During the analysis, we found that the page is located on the site in a small or inconspicuous element. That is, the impression will be counted, but the real user will not see, for example, the ad and will never be able to click on it.

Suspicious visits (SV)

We see something suspicious in the analysis results, but it is not enough to confidently state that it is a bot. Possible reasons include the user having a harmless extension that somehow affects our checks, or the user accessing through a VPN.

Suspicious clicks require an individual approach, for example, to determine where to attribute suspicious traffic based on conversions - to fraudulent or good traffic.

Name	Description
SV total	Number of suspicious clicks.
Proxy	Users accessing through VPN, proxy, or Tor. These are live clicks, but they come from non-targeted geo. If you are buying traffic from a specific geo, and there are a lot of proxies on it, then the traffic is bad.
Suspicion of spoofing	We suspect that user agent spoofing techniques are being used in the browser. If this metric is low, there is nothing to worry about. But if its percentage is high and you are not buying any specific mobile traffic, then there is a reason to think about it.
Suspicion of automated	We suspect that automation tools are being used in the browser.
Suspicion of fraud	Other types of suspicious activity, such as a too-wide screen, can be the result of poor layout or an attempt to hide ads outside the screen.

Feature

Some browser features that we detect.

Name	Description
AdBlock	Whether AdBlock is installed in the user's browser or not. Note that we only check for the presence of AdBlock; whether it is turned on or off is not checked.
Is mobile	Checking by characteristic features (not by user agent) whether the device is mobile (smartphones, tablets) or not (desktops, TVs).
WebView	Number of visits with Chrome WebView. Chrome WebView is a greatly reduced functionality version of mobile Chrome that is used in apps to view web pages, as well as many mobile browsers for Android, which are customized WebView. Most WebViews cannot display pushes. Therefore, if there is a high percentage of WebView on push traffic, the traffic may be bad.
IFrame	Number of visits made via iframe. <iframe> tag HTML element represents a nested browsing context, embedding another HTML page into the current one.
Push notifications	Number of visits from browsers that support push notifications.

Groups

Groups are parameters that allow you to answer the question "which?". From which countries were the clicks? Which browsers did users use?

Time

Name	Description
Hour	Groups statistics by hours.
Day	Groups statistics by days.
Week	Groups statistics by weeks.
Month	Groups statistics by months. The maximum period that can be selected is two months.

Tags

Name	Description
Integration	Kaminari integration ID.
Bot type	Grouping by bot types.
Referer	URL of the page from which the user landed on the page.
Sub1..Sub7	Tags set by the client.

Geo

Name	Description
Language	User's preferred language, i.e. UI language set in the browser settings.
Timezone	Browser time zone, actually the user's operating system time zone.
Country	Country data obtained from IP address.
City	City data obtained from IP address.
Provider	Internet service provider data obtained from IP address.

Device

Name	Description
Browser	Browser name and version taken from user agent.
Browser engine	All modern browsers are based on three open-source engines: Blink (Chrome, Chromium, Yandex Browser, Samsung Browser, modern Opera and MS Edge, most Android browsers); WebKit ((Safari; versions of Chrome, Yandex Browser, Firefox, etc. for iOS; browsers in PlayStation); Gecko (Firefox). There are also three old unsupported engines that are still present on the internet in the form of Internet Explorer, in MS Edge up to version 18 and in Opera up to version 12. We can determine all 6 engines by their characteristic features, as well as determine the specific version of the engine. It is the browser engine that we write this data to.
OS	Operating system and its version taken from user agent.
Real OS	The current OS without version.
Device	Mobile device manufacturer obtained from user agent.
Device type	Device type (desktop, smartphone, smart TV, etc.) obtained from user agent.
Connection type	The type of connection that the device uses to connect to the network (Bluetooth, Wi-Fi, mobile internet).

Screen

Name	Description
Screen width	Screen width in pixels.
Screen height	Screen height in pixels.
Screen orientation	Screen orientation is relevant for mobile devices - portrait, upside-down, turned left, turned right.
Pixel density	How many screen pixels will be used to render one CSS pixel, i.e. this is actually the screen resolution.