In Tweakwise, attributes can be configured to be used to build a customer profile.
For example, to build a personalization profile for our demo merchandising shop, where you want to add weight to color and fit. Configure color, fit or gender as measures and dimension these on product-group. The personalization matrix in our demoshop visually represents these segments.
Weight
For each event, we assess its impact by assigning a weight to relevant measures and dimensions.
Event | Weight (initial) |
---|---|
Click product | 1 |
View product | 1 |
Purchase | 1 |
Search | 1 |
Half life
The weight of each measure or dimension is influenced by the frequency of events associated with them.
To ensure that our recommendations remain relevant, we apply the concept of half-life to these weights. Over time, the relevancy of a measure or dimension decreases according to its half-life.
If the relevancy of a measure or dimension falls below a threshold of 0.0001, it is considered no longer significant. At this point, the measure or dimension is removed from the personalization matrix.
This approach ensures that only the most relevant attributes are used in creating personalized experiences for our customers.
Example: View Tweakwise Hoody 2 times in a day
Day | Action | Weight |
---|---|---|
0 | productView Tweakwise Hoody | 1 |
0 | productView Tweakwise Hoody | 2 |
0 | Calculate relevancy | 2 |
1 | Calculate relevancy | 1 |
2 | Calculate relevancy | 0.5 |
3 | Calculate relevancy | 0.25 |
... | ||
15 | Calculate relevancy | 0.0001 |
Retention
Profiles are automatically deleted after a certain time of inactivity. The retention of profiles is determined by the maximum number of profiles allowed in your account. For instance, if your account allows up to 100,000 profiles, the addition of a new profile will trigger the deletion of the longest inactive profile once this limit is reached.