Work in progress. Many questions can be answered by reading the Web App User's Manual.

If you have questions:

IRC freenode.net #cogswell
email: support@cogswell.io

Questions

How do rules work? What is the difference between Filters and Conditions?

Filters

Filters are the way that you control the set of data that conditions are applied to. Filters are applied in order. Example:

Let's say 500 events have been sent to Cogs, and the schema and data looks like:

id	first_name	last_name	state	logged_in	purchase_amount
0	bob	jones	CA	false	10.99
1	jane	doe	NY	false	32.98
...	...	..	..
499	anakin	skywalker	NM	true	62.11

If your first filter was: state is NM

Then we would apply that filter to the full dataset of 500 records. Let's say we got 200 records back.

If you had a second filter: logged_in is true

Then that would be applied to the result set of 200 records from the first filter result. Let's say you had 40 records as a result.

If you had a third filter, then it would be applied to the result of the second filter, which was 40 records.

Conditions

Conditions always evaluate to true or false. It is the sum of these true or false statements that are evaluated to determine if the rule passes. Cogs currently ANDs all conditions together, therefore, all conditions must evaluate to true for a rule to pass. All rules are ANDed together and thus all Rules must pass in order for a Campaign to trigger.

Each condition is applied to the resultant data set from the result of applying filters (if any were defined).

Following our above example, lets say we had a third filter, which reduced the data set to the following three (3) records:

id	first_name	last_name	state	logged_in	purchase_amount
99	james	jones	NM	true	77.22
114	sarah	amhearst	NM	true	2.99
373	jonah	nichols	NM	true	9.99

Let's say you had two conditions:

Condition 1: count purchase_amount greater than 2

Condition 2: average purchase_amount less than 10.00

Then it will first evaluate the first condition against the data set. In this case, since there are three records in our dataset, then this is a pass. Please note that on the count operator, the selected field has no bearing. This is simply just a way to count the number of records in the resultant dataset.

It will then evaluate the second condition, in which the average of the purchase_amount of the three records turns out to be larger than 10.00. This evaluates to false. So the rule would not pass.