Direct Conversation and Direct Message

Some flow automations allow you to apply filters (language, sentiment) to the whole conversation or to individual messages.

This is why it is important to understand differences between a direct conversation (DC) and a direct message (DM).

Direct conversation

Direct conversation is a body of messages exchanged between two profiles on a social network.

In other words, when your customer sends you a direct message to your Facebook profile, they are starting a direct conversation.

Conversations are, technically speaking, endless. Any interraction between any two profiles on a given social network is considered a single direct conversation.

For example, imagine that follower John sent you a direct message on Facebook 10 years ago. You replied and that was it. If John sends you another message tomorrow, the new message will be a part of the conversation that started 10 years ago.

Direct message

Direct message is a single ‘unit’ of direct conversation that was sent by to your profile (you) by another profile (your customer).

How we analyze conversations and messages

Understanding the differences between direct conversation and direct messages is important for handling communication with your followers, whether in immediate sorting or in complext flow automations.

Community feeds (all incoming, sent, etc.) only work with direct conversation (not with with individual direct messages).

• Filters related to the notion of ‘message’ work with aggregated values from direct messages. This applies to the following filters

  • Language

  • Sentiment

  • Message type

  • Media type

  • Chatter messages

  • Fulltext

• Fulltext search ocurs in the last 100 direct messages including brand and customer messages.

• Some filters allow you to make a choice to apply them on the message lever or the conversation level.

Direct messages

On the direct message level, a filter is applied to a single message.

When a new customer direct message arrives, it is evaluated against filters. If it meets the conditions, your rule is triggered.

Direct conversations

On the direct conversation level, things are interesting 🙂

Only your customers' messages are considered in the filters.

Example: Your agent reply in German to your customers French messages would have no effect on the message/conversation being filtered as French.

Every time a new direct message is received in a direct conversation, this is what happens:

1. Automation selects one or multiple customer messages that arrived after the previous agent's responses.
   Selection logic:

   

   • the latest direct message from your customer

   • up to 10 previous customer’s direct messages or all previous customer’s direct messages up your own latest message, whichever comes first

2. We analyze the applicable messages - first language, then sentiment.

3. Based on this analysis, language and sentiment are determined.

Only the most recent detected language and sentiment are stored. In other words, the language of the direct conversation is always determined by the last analysis.

Instagram anomaly

When your follower mentions you (your brand’s profile) anywhere on Instagram in any story, Instagram API will treat that mention as a direct message between that follower and your profile. That mention will, as a result, slot in the queue of direct message within the direct conversation between you and that specific follower.

This, however, re-classfies the entire dicrect conversation to a Story. When an actual direct message arrives, the story turns into a direct conversation again.

1. direct message → becomes a part of direct conversation

2. story mention → becomes a part of direct conversation → direct conversation turns into a story
   Flow automation rules that are applied over direct conversation are not executed.

3. direc message → becomes a part of direct conversation → story turns into a direct conversation

Community feeds

Community feeds (All incoming, Sent, etc.) work with direc conversations, not with individual direct messages.

Implications:

• Filters related to message work with aggregated values from direct messages

  • language

  • sentiment

  • message type

  • media type

  • chatter messages

  • fulltext
    Fulltext search takes place in the last ten direct messages including both the customer’s AND the brand’s direct messages.