Get Slack alerts for low Dixa customer satisfaction scores

Every 30 minutes, check Dixa for conversations that just received a low customer satisfaction rating and alert our support leads in Slack with enough context to act. Run this on a cron schedule that fires every 30 minutes.

Two settings should be easy to change at the top of the workflow: the score threshold (default: flag any rating that is at or below 2 out of 5) and the Slack channel where support leads receive alerts (a support-leads channel).

First, use Dixa's Search conversations with filters to pull conversations that were recently closed or rated, looking back over roughly the last hour so nothing is missed between runs. This is the set of conversations to inspect.

For each conversation, use Dixa's List ratings for a conversation to read its satisfaction scores. Ratings are reported per survey question, so treat the conversation as low-rated if any question scored at or below the threshold. Skip conversations that have no rating yet or where all scores are above the threshold.

For each low-rated conversation, use Dixa's List messages to read what actually happened in the conversation before writing anything.

Then summarize, in a few short lines: the issue, the likely reason the customer was unhappy, and one suggested recovery step. Post that summary to the support-leads Slack channel using the Slack Bot integration's Send a Message, and include a direct link to open the conversation in Dixa (built from its conversation id) so a lead can jump straight in. Include the low score and the conversation id in the message as well.

Do not ping leads twice about the same conversation. Keep a record of the conversation ids you have already alerted on, and skip any conversation that is already in that record before posting. Add each conversation id to the record once its alert has been sent.