Dear all,
I just wanted to check with the group, which approaches / methodologies might be possible for my problem.
We send a DVD to around 5,000 subscribers. On the DVD there is a film the user has already paid and there is a second (encryted) film with a preview. If the subscriber likes the preview, he/she sends us an e-mail, pays and gets the decryption key. The question is which method to use to increase the number of subscribers to send us an e-mail.
Constraints:
1. There can only be one extra film per DVD.
2. The films, which are encrypted and the not encrypted films are from a common set of films. It doesn't make sense to send an encypted film, which the subscriber either has already received or will anyways be sent to him un-encrypted later in the year.
3. For all films we have some charateristics like actors, type of film, duration, producer, ...
4. We already saw, that there are some subscribers, who always decrypt the extra film and there are other subscribers, who never decrypt the extra film.
5. There is some extra cost to put the extra film on the DVD.
6. Instead of sending one DVD with an extra film we think about sending a second DVD for heavy users. Then on one DVD there would be one un-encypted film and one encyrpted film, on the other DVD there would be two encypted films.
I thought about different approaches:
- Check, which films were paid by other customers, who have the same set of subscribed films
- Do a standard analysis with the CAR including the subscribed films, RFM indicator and the current encrypted film and create a model. In the model the list of already sent encrypted films and the response (either decrypted or not) is made part of the CAR. Using this model each time a film is sent the list of possible encryted films are put into the model and the film with the highest score is sent to the subscriber.
- I heard from special basket analysis methodologies, but have never used them. Which might be useful?
Best regards,
Jan
I just wanted to check with the group, which approaches / methodologies might be possible for my problem.
We send a DVD to around 5,000 subscribers. On the DVD there is a film the user has already paid and there is a second (encryted) film with a preview. If the subscriber likes the preview, he/she sends us an e-mail, pays and gets the decryption key. The question is which method to use to increase the number of subscribers to send us an e-mail.
Constraints:
1. There can only be one extra film per DVD.
2. The films, which are encrypted and the not encrypted films are from a common set of films. It doesn't make sense to send an encypted film, which the subscriber either has already received or will anyways be sent to him un-encrypted later in the year.
3. For all films we have some charateristics like actors, type of film, duration, producer, ...
4. We already saw, that there are some subscribers, who always decrypt the extra film and there are other subscribers, who never decrypt the extra film.
5. There is some extra cost to put the extra film on the DVD.
6. Instead of sending one DVD with an extra film we think about sending a second DVD for heavy users. Then on one DVD there would be one un-encypted film and one encyrpted film, on the other DVD there would be two encypted films.
I thought about different approaches:
- Check, which films were paid by other customers, who have the same set of subscribed films
- Do a standard analysis with the CAR including the subscribed films, RFM indicator and the current encrypted film and create a model. In the model the list of already sent encrypted films and the response (either decrypted or not) is made part of the CAR. Using this model each time a film is sent the list of possible encryted films are put into the model and the film with the highest score is sent to the subscriber.
- I heard from special basket analysis methodologies, but have never used them. Which might be useful?
Best regards,
Jan