CCA: Contact selection explained

The aim of this document is to explain how the next contact is selected from the list(s) to best suit the requirements defined in the CCA. 

Please note that this quota methodology is relevant for v5.5 and earlier versions of Askia, but is now superseded with a new methodology from v5.6. The new methodology is described in the following article. 

High level explanation

A survey will have a list or lists assigned to work with an outbound group of agents. There are 2 major parts in selecting the best contact from these lists:
 - The best type of contact to call next.
 - The specific contact of that best type.
Categorizing the contacts in terms of the different types is done according to the quota definition. The label for each category or type is called the "Peneo".

The origin of the Peneo

From the quota definition, the CCA calculates a unique number for each possibility in the quota. This unique number is called the peneo value.

For a simple quota definition example like gender (male & female), there are three possibilities: male, female and unknown. The latter occurs when there is no appropriate data available, either because there has been no interview yet, or no answer has been provided for this particular question.

Here is the formula:

Peneo = resp_q1 + count_q1 * resp_q2 + count_q1 * count_q2 * resp_q3
with:
- resp_q = index response + 1 (for DK/NA)
- count_q = count responses (including linked) + 1 (for DK/NA)
Order questions: as encountered in the quota tree (from top to bottom):

This peneo value is written to the list database for each contact and is updated when the quota definition changes or when new data is available from filling in the interview or from the dynamic data coming from the list's (external) source database. Depending on quota definitions the number of peneo possibilities can be extremely high.

Fishing out the highest peneo from the pool

Not all possible peneos are swimming in the pool of the lists associated with the outbound group. Since not all peneos are available, we need the best match from all available peneos to meet the quota definition. The best match is of course dependent on the interviews collected so far and the specifics of the quota tree. The best possible match is found through statistical analysis, by using Khi2 tests on the current distribution, compared to the acquired distribution, defined by the quota.

Tuning the fishing method

Through the priorities in the quota definition you can define if some quota cells should be reached faster, slower or not at all. The blocked priority means a quota line will never be used. The ignored priority will still try to get to the desired distribution but will allow going over the desired target. The behaviour will only define the client-side behaviour when an undesired quota occurs. The automatic behaviour will not allow extra data to be filled in once over target. Semi-automatic behaviour lets the agent decide what to do in the case of undesired data. Manual behaviour leaves the choice of action up to the survey routings.

Finally. . .

Once the best possible peneo is found, the next stage of contact selection can start. This is getting the best contact from this peneo category. The best contact from the lists is the contact with the highest priority for this peneo. The contact priority depends on the position in the list, callbacks, appointments made, appointment importance and list priority. Once we have chosen this best contact for the best possible peneo, we can celebrate and start the process over again for the next best contact.

Please note that this quota methodology is relevant for v5.5 and earlier versions of Askia, but is now superseded with a new methodology from v5.6. The new methodology is described in the following article.