On the “Settings” tab you can not only set model parameters and set channels that should not get value assigned, but set up a funnel of your project.

Funnel setup

To start the setup, choose one of the step edit options:


In the step editor you can set its name and conditions. You can identify sessions of the step with page path, event, e-commerce action or domain name.


The checkmark under each step show if the step is enabled. Sessions of a disabled step don't get value assigned and don't influence probability of other steps.

“Calculate the funnel” button lets you preview metrics of the steps to find out mistakes in steps conditions or data.

Probability and value

Each funnel step has three metrics: Step conversion rate, Probability and Value.

Step conversion rate is a percentage of visits, where transition to a particular step occurred, out of all visits.
For example, conversion rate — 9% for step “Checkout” means that checkout page was visited in 9% out of all visits.

Probability is a percentage of visits, where transition to a particular step occurred, out of visits with the previous step action. It shows how difficult is for visitors to pass to the step from the previous one.
For example, probability — 78% of step “Checkout” means that transition to this step occurs in 78% of visits where “Add to cart” action occurs.

Probability of the first step — VisitIf you start with a step that selects all visits, its probability equals 100% and value 0%. To assign value to channels targeted at visitor acquisition, you can set “Non-bounce visit” as the first step, its probability is equal to share of sessions without bounce.

Data is segmented during probabilities calculation During calculation probability of each step, data is grouped by user type, geo regions and previous steps. This lets the model consider that probabilities of users pass though the funnel significantly depend on these parameters.
User type — new or returned — is defined whether a visitor made purchase during last year or not.
Device category — desktop only, tablet only, mobile only, cross device — is defined category of device a visitor used during last year.
Geo region — is defined by Google Analytics session dimension.

Value is attributed only to sessions that lead a visitor to a purchase and only to sessions where the first visit of a step occurred. To calculate the value firstly we define a score of each step, it's inversely proportional to probability of the step. Then score of the step is divided by sum of all steps scores.

The more difficult for users to pass thought the step, the more value it gets.

Here's an example of value calculation (the table is also available in Google Sheets):

Step Step conversion rate Probability Score Value How value was calculated
Visit 60% 60% 40 18% 40/(40+30+81+38+40)
Product page 42% 70% 30 13% 30/(40+30+81+38+40)
Add to cart 8% 19% 81 35% 81/(40+30+81+38+40)
Checkout 5% 63% 38 16% 38/(40+30+81+38+40)
Purchase 3% 60% 40 18% 40/(40+30+81+38+40)
Was this article helpful?
0 out of 0 found this helpful
Have more questions? Submit a request


Please sign in to leave a comment.