System events are a group of events available to all Bloomreach Engagement customers by default without the need for a specifically-tailored integration. As opposed to [custom events](🔗), system events should be the same for all Bloomreach Engagement users. In this article, we will list and explain the attributes of these events.

# session_start, session_end, first_session

🚧

Ghost sessions

A `session_start` and `session_end` (but not `page_visit`) can occasionally be tracked without the customer being active. This can happen when the customer leaves your website running, their computer falls asleep, and then their browser gets randomly active and generates a session ping without any `page_visit`. Read more about this occurrence and our solutions for it in [Ghost sessions](🔗).

The duration of the session will be the difference between the timestamps of `session_start` and `session_end` in seconds. The timestamp of the `session_end` will have a timestamp within 2 minutes after the customer closed the last page (the 20-minute timeout is ignored for the purpose of calculating the session duration).

In other words, the timeout ensures that if someone closes the page but returns within 20 minutes, this will still be counted as a single session. However, if they don't return, those 20 minutes will not be counted into the session duration.

The events are tracked with the following properties (you can add custom ones using [the Javascript SDK's ping options](🔗)):

📘

Using Google Analytics ID for `first_session`

If you have just started using Bloomreach Engagement and you had tracked your customers using Google Analytics before, you can use the Google Analytics ID to immediately distinguish between old and new customers also in Bloomreach Engagement. You can get the Google Analytics ID from the [Google Analytics tracker](🔗). Google Analytics ID is written in the form `1.2.286403989.1366364567` and it's last part (behind the last dot) is the timestamp of the first session. In Bloomreach Engagement, you can extract it from Google Analytics ID using an expression and store it as a customer attribute.

🚧

Customer's location inaccuracies

As all the attributes that localize the customers are based on their IP, they depend on the quality of the IP infrastructure. The `country` and `city` attributes are generally very reliable, however, the `latitude` and `longitude` are often not (as they are a more precise measure of the customer's location).

## session_start custom attributes (deprecated)

`session_start` is often complemented with these manually-added custom attributes. These attributes are deprecated. If you want to try to add some of them as custom attributes, read on how to add default attributes to events & set up advanced tracking parameters in the **[Web tracking article](🔗)**.

## session_start and session_end activity properties

`activity` properties are sent by the [Javascript SDK](🔗) in session pings.

# page_visit

`page_visit` happens each time the customer opens one of the pages of your website. Note that these events are disabled by default but you can enable them using the [track options of the Javascript SDK](🔗).

🚧

Issues with the tracking

If you are using our CDN page outside of your own domain for giving out surveys/getting consent, no `page_visit` or `session_start` events will be generated.

# campaign

The `campaign` event is generated when working with emails, text messages, push notifications, or webhooks.

## Values of the status attribute

## Which status attributes are generated by what channel

## Values of the code attribute

`Code` attribute gives you the SMTP error or reply code received from a server after you send out a campaign email. The `status` attribute is based on these codes, however, it is useful to also know the code itself as it is more precise in assessing what happened to the email. The following [link](🔗) contains a list of codes and their corresponding meanings. For these to work correctly, you need to have a [Mailgun integration](🔗). Even though SMTP codes are reliable for most customers, beware that small email providers often send alternative SMTP codes that will not be recognized.

📘

Tracking clicked event

Links to your webpage sent out in a campaign firstly lead to our CDN domain so that the `clicked` event is tracked. After that, the customer is immediately redirected from the CDN domain to your webpage.

🚧

Issues with the opened event

`Opened` event tracked for Gmail does not guarantee that the recipient had indeed opened your email. Gmail's servers sometimes download external images from emails into their cache in advance. As we use these images to track whether the customers opened the email, we get a fake open event every time Gmail's servers do this. Moreover, email providers sometimes block the loading of the images we use for the tracking of the `opened` event so even if your email seems to be just `delivered` it might, in fact, be opened.

### Bounce management

When an email does not get delivered, there might be multiple reasons why it happened. We collect the failure feedback from our email providers and automatically assign them to standard bounce categories such as soft bounce (e.g. full inbox) or hard bounces (e.g. invalid email). Bloomreach Engagement also works with our custom type of bounce, what we call a “cumulative bounce”! Read all the details in our article for [Bounce Management](🔗).

## Status name change and filters migration

In 1.173 we unified the statuses of email campaign events from email providers and renamed them to meet the industry standards and improve user experience.

In the following table you can find the naming changes:

Along with renaming, an ** automatic migration** was carried out, in order to update event filters as defined below. The new naming along with the original one was used in the migration to ensure backwards compatibility. So don't worry, you probably don't have to update your analytics and campaigns manually, but keep an eye on your business critical use cases after the change to make sure everything works as expected.

The following change was done in all filters that use the renamed statuses:

• Operators `equals` have been changed to `in` and `does not equal` to `not in` and extended with new statuses.

• Operators `in` and `not in` have been extended with new statuses.

🚧



Rest of the string filters `contains`, `does not contain`, `starts with`, `ends with` and `matches regex` were not changed as they are unlikely to be used for this purpose. **Manual change** is needed in case you want to use those filters in combination with the renamed statuses.

For any other custom usage such as jinja or webhooks a **manual change** is needed as well.

_Example:_ `equals dropped` ->`in [dropped, soft_bounced]`

`in [dropped, x, y]` -> `in [dropped, soft_bounced, x, y]`

## Automatic bot detection in emails

Some portion of opens/clicks that you might track in your email campaigns can be generated by automated bots that are scanning content and links of the email to identify spam and phishing emails. To tackle this problem we have developed Automatic bot detection that automatically identifies such bots.

We use the _honeypot link_ solution. It is based on adding an invisible link that real humans can’t click but bots do. Every tracked click on the honeypot link is identified with a status containing a special value `clicked_honeypot`. This feature does not _suppress_ any events but rather _provides the necessary data_ to let your analytics sort it out.

It is important to state that it won’t identify 100% of bots as this is technically not possible. There is no universal solution and bots are constantly changing and trying to identify themselves as real humans - to prevent being identified by spam creators.

This feature can be turned on/off in project settings under `Campaigns` -> `Channels` -> `Email` -> `Automatic bot detection`. After turning on, it automatically starts identifying bots with specific status.

# survey

`survey` happens when the customer is shown a survey question you had created. [More on surveys](🔗)

# merge

`merge` when two customer profiles are merged into one

# ab test

`ab test` happens when the customer sees one of the variants of a page/banner/recommendation you had created in an A/B split. [More on A/B testing](🔗)

# experiment

`experiment` happens when an experiment is shown to the customer. [More on experiments](🔗)

# anonymization

`anonymization` happens when you remove the customer's identification but still keep their data for analytics purposes. You can read more about anonymization in [Data management](🔗).