The Concept of Signals
The Signals Report allows you to quickly find events – both (E) interactive and (M) meta events – which have a negative or positive impact on your goals.
For example, you could use Signals to find:
- any (M)
Error=*
events to learn which errors reduce conversion the most and should be fixed first, - (M)
Section=Product-Category/*
events to learn which product categories increase the chances of adding a product to the basket to promote them, - (E)
*.click
events to learn which UI elements users click before they contact your support via live chat and understand what causes them problems, - (E)
button[2]#productID*.addToBasket.click
events to see which products are abandoned after being added to the basket.
Differences Between Signals and Stats
There is a significant difference between events returned by the Signals report and events returned by the Stats report.
The Stats report allows you to find the most popular events in a segment, for example:
- most popular (M)
Error=*
events in visits in which users failed to convert, - most popular (M)
Section=Product-Category/*
events in visits in which users added product to basket.
The Signals report, on the other hand, may surface events that are less popular but have high impact on a goal that you specify and will limit the results to events that happened before the specified goal.
For example, if there were only two visits in your project and in the first visit a user:
- went to Section=Product-Category/Jeans
- added a product to basket
- went to Section=Product-Category/Dresses
- left the website
and in the second visit a user:
- went to Section=Product-Category/Jackets
- added a product to basket
- went to Section=Product-Category/Dresses
- completed a purchase
- left the website.
If you create a segment of visits in which users added a product to basket and pulled the Stats for (M) Section=Product-Category/*
you would get the following results:
Event Value | Count | Share |
---|---|---|
Product-Category/Dresses | 2 | 50% |
Product-Category/Jeans | 1 | 25% |
Product-Category/Jackets | 1 | 25% |
However, if you use Signals to find (M) Section=Product-Category/*
events that increase a chance that user will add a product to the basket, only Jeans and Jackets categories would show up. The Dresses category would only show up among signals indicating that users will likely fail to add a product to the basket because in each of the visits the event (M) Section=Product-Category/Dresses occurred only after a product was added to the basket and never before.
Based on the Stats report you might decide to invest money to drive more traffic to Dresses, but based on data from the Signals report you would avoid that mistake and invest in Jeans and Jackets instead.
Overview of the Signals UI
To use Signals, select a desired segment, click the Signals tab and configure the report:
- 1) Specify event mask to include only events of a certain type or containing a certain string. Mask
"*"
will return all kind of events, both (E) interactive and (M) meta events. A mask like*error*
would return any events containing “error” string, for example (M) Error=Invalid Email and (E) input[1]#email.error.change. Mask*error*.change
would return (E) change events tracked when a user filled in a form field and the field’s selector contained string “error”, for example (E) input[1]#email.error.change, (E) input[1]#phone.error.required.change. - 2) Select whether you want to find events that signal a reduced, increased or entirely eliminated chance of an occurrence of a 3) targeted event.
- 3) Specify the mask of targeted (E) interactive or (M) meta event or events. For example, you could use
Section=Purchase/Completed
to find signals that increase or reduce conversion,Section=Product/*phone*
to find signals that increase or reduce visits to any phone product or evenError=*shipping*
to find out what causes or prevents errors related to shipping.
The Signals Chart
The so called “bubble” chart provides a method to visualize signals. Each signals is displayed as a separate bubble.
- 1) The vertical position of the bubble is based on the signal’s impact – the higher a bubble is positioned the higher its impact is (see below for more details).
- 2) The horizontal position is based on the average distance from the signal to the targeted event. Signals that happen just before the targeted event are displayed farther to the right. For example, if the chart would display signals increasing conversion in an ecommerce website then signal of “going to checkout” would be farther to the right than “adding a product to basket”.
- 3) The color of a bubble matches the value in the signal’s dynamic, indicating whether the signal is trending or receding (see below for more details).
- 4) The size of a bubble is based on signals volume – basically, the larger the bubble the more visits there are with such event.
The Signals List
1) Signal Impact
By default, the signals listed in the table are sorted by impact. The impact of a signal is calculated based on four variables:
- the volume of visits with a targeted event (V1),
- the volume of visits with a signal and with a targeted event (S1),
- the volume of visits without a targeted event (V2),
- the volume of visits with a signal and without a targeted event (S2),
If the query is set to return signals increasing the chance of a targeted event, the impact is calculated using the formula:
impact = MIN ( 100, (S1 / V1) / ( MAX ( 1, S2) / V2 ) ) * V1 / 1000
for signals reducing the chance of a targeted event:
impact = MIN ( 100, (S2 / V2) / ( MAX ( 1, S1) / V1 ) ) * V2 / 1000
and for signals eliminating the chance of a targeted event:
impact = 100 * V2 / 1000
2) Signal Dynamic
The Signal Dynamic tells you if a signal is trending (values 0 to 1) or receding (values 0 to -1) within the time frame specified in your report’s settings:
The angle of the arrow ranges from -45 (value -1) up to 45° (value 1). The color of the arrow ranges from red (value -1, receding) to blue (value 1, trending) if the report shows “increasing” signals, or from blue (value -1, receding) to red (value 1, trending) for “decreasing/eliminating” signals.
3) Event
The 3rd column shows the event name together with the dates of the event’s first and last occurrence in your project. The dates are not affected by your report’s time frame nor by the current segment.
4) In Sessions With The Targeted Event
The percentage value in this column indicates the share of visits with the signal (event) among visits with targeted event(s). The number of visits gives the count of visits with both the signal and targeted event(s).
For example, 10% (10) would mean that there were 10 visits with the signal among 100 visits with the targeted event(s).
5) Events per Session (with the Targeted Event)
This value tells how many times the event was tracked during visits with the targeted event.
6) In Sessions without the Targeted Event
The percentage value in this column means the share of visits with the signal (event) among visits without the targeted event(s). The number of visits gives the count of visits with both the signal and the targeted event(s).
7) Events per Session (without the Targeted Event)
This value tells how many times the event was tracked during visits with the targeted event.