Routing rules are flexible and personalised. They enable you to accurately define how and to which respondent your conversations should be distributed via the assignation methods.
Here are the 3 assignation methods available within our platform:
1. Assignation in escalation
Overflow occurs depending on the occupation of the selected routing groups and the order defined by the administrator.
Here's how to distribute conversations in escalation between 3 routing groups which will be A > B > C:
When the threshold for conversation queues is set to 0 :
You only engage visitors if your agents are available to receive conversations in real-time.
For a conversation to be distributed to routing group B, agents from routing group A must have either reached their maximum occupancy, or be offline, or be unavailable.
For routing group C to receive a conversation, routing groups A and B must have either reached their maximum occupancy, or be offline, or be unavailable.
When the threshold for conversation queues is greater than 0 :
You engage visitors even if your agents are not available to receive conversations in real-time.
For a conversation to be distributed to routing group B, the routing group A's conversation queue must be full.
Here is an example with the following configuration:
- global conversation queue threshold set to 100
- Routing group A's threshold set to 50
- Routing group B's threshold set to 30
- Routing group C's threshold set to 20
Conversations will first arrive in routing group A's conversation queue. If group A's conversation queue is full (full when reaching 50 conversations), new conversations will arrive in group B's queue, and so on.
When all the conversation queues of each group associated with your routing rule are full, i.e. the number of 100 pending conversations has been reached, then no more visitors will be engaged.
NB : It is not recommended to choose the cascade distribution method for conversations with third party channels (WhatsApp, Facebook, Apple...), as the availability of the groups is not checked and the conversations are by default queued to the first distribution group (here A). Prefer the custom distribution rule to make a distribution.
2. Equitable assignation
Conversations are equally distributed between several routing groups.
Here's how the process of a fair routing between 3 routing groups A = B = C works :
When the threshold for conversation queues is set to 0 :
You only engage visitors if your agents are available to receive conversations in real-time.
Conversations are distributed in turns to routing groups A, B, then C. When a routing group is not available (maximum occupancy, offline, unavailable), conversations are sent to the available routing group furthest from its target percentage.
When the threshold for conversation queues is greater than 0 :
You engage visitors even if your agents are not available to receive conversations in real-time.
Conversations are distributed in turns to the conversation queues of routing groups A, B, then C.
For example, if you have 3 groups and have defined a threshold of 100 conversations, incoming conversations are split equally between the groups:
- 1 for group A's conversation queue
- 1 for group B's conversation queue
- 1 for group C's conversation queue
- 1 for group A's conversation queue
- and so on, so that group A's conversation queue receives a maximum of 34 conversations, and the queues of groups B and C 33 conversations each.
When all the conversation queues of each group associated with your routing rule are full, i.e. the number of 100 pending conversations has been reached, then no more visitors will be engaged.
N.B. Whether a conversation queue is set to 0 or not, if a routing group is increasingly lagging behind according to the target percentage, the routing engine favours it in the distribution for the group to catch up (conversations are distributed in priority to the groups furthest from its target percentage).
3. Personalised assignation
Conversations are distributed between several routing groups according to a ratio defined by the administrator.
For a personalised routing defined between 3 routing groups A, B, C:
Conversations are distributed in turns to routing groups A, B, and C and then distributed to groups furthest from their target percentage.
When the threshold for conversation queues is set to 0 :
You only engage visitors if your agents are available to receive conversations in real-time.
When the threshold for conversation queues is greater than 0 :
You engage visitors even if your agents are not available to receive conversations in real-time.
For example, if you have defined a threshold of 100 conversations for your routing rule and with 3 routing groups whose distribution is as follows: A:50% B:35% C:15%; new conversations will then be distributed in turns to the conversation queues according to the routing system selected.
- 1 for group A's conversation queue
- 1 for group A's conversation queue
- 1 for group A's conversation queue
- 1 for group B's conversation queue
- 1 for group B's conversation queue
- 1 for group C's conversation queue
- And so on and so on
When all the conversation queues of each group associated with your routing rule are full, i.e. the number of 100 pending conversations has been reached, then no more visitors will be engaged.

N.B. Whether a conversation queue is set to 0 or not, if a routing group is increasingly lagging behind according to the target percentage, the routing engine favours it in the distribution for the group to catch up (conversations are distributed in priority to the groups furthest from its target percentage).
NB: The counters for the number of conversations distributed per routing group are reset at the start of every hour (based on the project timezone). This is valid for both fair and personalized distribution.
How are conversations handled?
When the threshold for conversation queues is set to 0 :
Agents from each group receive the conversations in real-time when logged in to their conversation panel and available.
When the threshold for conversation queues is greater than 0 :
Agents from each group handle conversations coming from queues when logged in to their conversation panel and available. Whenever they snooze or close a conversation, it frees up a spot in their group's conversation queue, and they receive the next conversation to be handled.
4. Impact of asynchronous on assignation methods
Assignation in escalation
If a rule 1 exists with
- Group 1 - 50 places
- Group 2 - 20 places
Conversations will fill the queue in Group 1, before being able to flow into Group 2
Equitable / Custom assignation
If a rule 1 exists with
- Group 1 - 50% - 50 places
- Group 2 - 50% - 20 places
Conversations will be allocated according to the 50/50 target over each full hour: the counters start again from zero at the start of every hour.
NB : In our example, for group 1, it is possible to compensate for the equitable distribution that would deprive group 1 of conversations by placing 30 agents in group 1 and 50 in group 2.