This training chapter shows how to work with user-centric data in Webtrekk. You can download this training chapter as PDF file here.
Content:
- User-centric data
- Calculation of frequencies
- Calculation of intervals
- Overview dimensions and metrics
- Automatic cleanup of URM data
- Integration of additional data
- Predictions
User-centric data
For every user Webtrekk calculates data, that depict frequencies and intervals.
There are different kinds of user-centric data:
- Focus on frequencies
- How many visits did a user make in total?
- How many pages did a user access?
- …
- How many orders did a user make in total?
- What order value did a user generate?
- How much discount was given to a user in total?
- …
- Analysis of accesses
- Analysis of orders
- Focus on intervals
- How many days pass between visits?
- How many days pass between orders?
- …
Calculation of frequencies
Which information can be generated by a user?
For the example of frequencies of visits the following information could be relevant:
- What was the number of the visit at a date (e.g. of the first order)?
- How many visits did a user make in total?
- How many visits did a user make in the analysis time period?
Besides the regular metric customer and customer profile views are differentiated:
- Customer: The value a user had at a specific date.
- Customer Profile: Focusses on the current status of the users (until his last visit). For doing so, the last visit does not necessarily have to be within the analysis time period.
- [Regular]: How often did the value occur in the analysis time period.
Example: "Customer Visits" vs. "Customer Profile Visits" vs. "Visits"
Even though the last visit was not made within the analysis time period, "Customer Profile" shows the value of the last visit.
Example:
Customer profile views are not suitable for a time-based analysis.
Calculation of intervals
Webtrekk calculates intervals of visits and orders.
Calculation of visit intervals
- For each visit the amount of days that passed since the previous visit is calculated.
- The days are calculated based on the date, the time is not taken into account.
- The first visit is not considered for calculation.
By using cohorts Webtrekk provides the possibility to analyze days/weeks/months since the first visits (e.g. "Cohorts (Lifespank Weeks)").
By using the filter engine the analysis can be limited to specific events/visits.
Example: Calculation days between contacts with filtering
How many days prior to the last order users accessed the website?
How many days prior to the last visit users accessed the website?
Calculation of order intervals
- For each order the the amount of days that passed since the previous order is calculated.
- The days are calculated based on the date, the time is not taken into account.
- For the first order the amount of days passed since the first visit is depicted.
Overview dimensions and metrics
The following objects and metrics are calculated automatically:
Traffic view
Order view (1/2)
Order view (2/2)
Why is much information available as dimension, even though figures are depicted?
For lots of user-centric data it can be helpful to use it as dimensions, even though figures are depicted. Only by doing so further metrics and formulars can be viewed in this context.
Example for using the dimentions "URM – Customer Profile Visits":
Allows the analysis of the question: "How many users did make only 1 visit in total?"
Example for using the metric "URM – Customer Profile Visits":
Allows the analysis of the question: "How many visits did users make in their lifetime, when they had at least one access via the channel "Direct"?"
Automatic cleanup of URM data
User-centric data of users who have no login information (i.e., no Customer ID) are cleaned up automatically.
By default, user-centric data is automatically cleaned up if a user has
- exactly one visit and exactly one page impression: after 30 days
- exactly one visit and more than one page impression: after 180 days
- more than one visit: after 360 days
Consequently, users who are registered are not deleted automatically.
Integration of additional data
Additional data can be integrated arbitrarily via URM categories.
Several options arise on how to submit this data:
- Onsite data collection: Submission of the information directly on the website.
- Data import: Via SOAP, JSON, Excel, Datafeed information can be imported.
Predefined URM categories
Numerous categories are already predefined in Webtrekk.
- These categories and key figures do not have to be set up separately in Webtrekk.
- Some categories can be populated automatically with the survey widget available in Tag Integration.
The following predefined categories exist: (1/2)
The following predefined categories exist: (2/2)
Individual URM categories
Individual information from CRM can be integrated with URM categories.
- Here the transferred customer ID serves as the key.
Examples:
- Creditworthiness
- Qty of offline orders
- Lead status
Predictions
Predictions are used to make forecasts based on the usage behavior.
- What is the likelihood that a specific visitor will never come back to the website?
- What is the likelihood that a specific visitor will buy something during one of the subsequent visits?
- What order value will a specific visitor generate with his next order/in the next 30 days/lifetime?
This information is relevant in order to derive individual marketing measures.
- The costs of marketing measures should not exceed the expected profit.
- Only spend money on customers with high potential!
A self-learning system continuously analyses influences and takes them into account for the calculation.
- The system is configured individually for each customer.
- In the first 2-3 months in particular, the predictions are automatically adapted on a customer-specific basis.
Various criteria are analyzed:
- Number of visits
- Number of orders
- Purchases / visit
- Visit duration Avg
- Page impressions / visit
- Product views / visit
- Value of product views / visit
- Order value Avg
- Order value
- Days since first / last visit
- Days since first / last purchase
- …
The following prediction metrics / dimensions are available:
- To calculate the churn probability, all that is needed is for each page to be captured in Webtrekk.
- To calculate the probability of all other metrics and objects, orders and products have to be measured as well.
Calculating the churn probability
Example:
Day 1
5 percent of all visitors come back to the website on the day of the first visit.
The churn probability on day 1 is: 100 % – sum of all returning visitors (Day 1 – Day 8) = 100 % - 22 % = 78 %
Day 2
7 percent of all visitors come back to the website one day after the first visit.
The churn probability on day 2 is: 100 % – sum of all returning visitors (Day 2 – Day 8) = 100 % - 17 % = 83 %
The overall churn probability for a period takes into account when visitors were last active on the website.
Individual behavior patterns are now considered in the calculation.
For example:
- Visit duration Avg
- Page impressions / visit
- Number of visits
- …
For example, if there were ten page impressions during the first visit, only visitors with a similar number of page impressions would be included in the calculation.