Reference · built on requestOperator by FibricCustomer support

Sentiment Spike

Tracks tone across support conversations and reviews by product and topic. Proposes an alert and a holding reply when a spike appears.

Request accessPricing ↗

About

A bad batch, a broken checkout, a courier having a bad week: the first sign is usually a cluster of unhappy messages about one thing. Sentiment Spike reads the sentiment your support tool and contact center already attach to messages and calls, and the star ratings and review text arriving from review platforms. It groups them by product, topic, or location and compares each group with its own recent history.

When a group moves past the threshold you set, it proposes an alert to a named channel that lists the conversations behind it, and a holding reply agents can use until someone knows the cause. A lead approves both. Replies to individual reviews are proposed one at a time.

This is a reference listing. It documents what Fibric would read from Sentiment Spike and what it could propose, based on the vendor's published interfaces. Fibric builds it under a managed deployment when you request it; selecting it here installs nothing.

Inputs

  • Sentiment scores in Kustomer, assigned per message on create and averaged for the conversation and the customer
  • Contact sentiment from Amazon Connect conversational analytics: OverallSentiment per participant and SentimentByPeriod across the call
  • Service reviews from the Trustpilot Business Units API, filtered by stars, tags, startDateTime, and responded
  • Reviews from Yotpo with score, sentiment, sku, and created_at, paged with count and page
  • Reviews per location from Google Business Profile with starRating, comment, createTime, and any reviewReply
  • Tags, product fields, and topic fields on conversations in the support tool, so a spike is attributed to one product or one topic

Proposed actions

  • Target capability: propose an alert to a Slack channel through chat.postMessage, naming the product or topic, the window, and the conversations behind it
  • Target capability: propose a holding reply for agents to use on the affected conversations, saved as a draft rather than sent
  • Target capability: propose a tag on each conversation in the spike, so the count can be followed and closed out later
  • Target capability: propose a reply to a review in the spike through Trustpilot or Google Business Profile, for approval one review at a time

Proposed actions are target capabilities. Every action runs propose-first and needs a validated deployment and the appropriate permissions.

What you can build

  • Catch a product complaint wave in support

    Conversations tagged with one product in Kustomer turn negative over an afternoon. The operator proposes a Slack alert listing them and a holding reply; the lead approves, and agents use the reply while the cause is found.

    With Kustomer, Slack

  • See reviews turn on one sku

    Yotpo reviews for a single sku drop in score across a few days. The operator proposes an alert naming the sku and the reviews, and a tag on the support conversations that mention it.

    With Yotpo, Slack

  • Watch call sentiment by queue

    OverallSentiment for the customer side falls across calls in one Amazon Connect queue. The operator proposes an alert with the contact ids so a supervisor can open the transcripts.

    With Amazon Connect, Slack

  • Reply to a burst of one-star reviews

    Several one-star reviews arrive on Trustpilot and Google Business Profile for one location. The operator proposes a reply per review from the wording you approved, each for separate approval.

    With Trustpilot, Google Business Profile

Requirements

  • A support or contact center connector with sentiment on messages or calls: Kustomer or Amazon Connect
  • A review connector: Trustpilot, Yotpo, or Google Business Profile
  • A messaging connector for the alert: Slack or Microsoft Teams
  • A baseline window and a threshold set by you, per product or topic
Authentication
Runs on the credentials of the support, contact center, review, and messaging connectors you attach. Nothing is posted to a channel or a review without a person's approval.

Limits

  • Sentiment engines differ. Kustomer averages per-message scores; Amazon Connect scores each turn from -5 to +5. Each source is compared only with its own history
  • Trustpilot's public review endpoints return at most 100,000 records per query. Older history is reached with startDateTime and endDateTime
  • Yotpo recommends up to 100 reviews per request, so a spike on a rarely reviewed sku is seen later than one on a busy sku
  • A spike is a count over the baseline you set. Sarcasm, quoted text, and one-word messages are scored by the source engine, not the operator

Access and pricing

Reference listing. Fibric builds the operator under a managed deployment when you request it. Your quote covers the build, capabilities, usage, and support.

Request Sentiment Spike ↗

Questions and answers

What does the lead approve when a spike fires?
The alert text and its channel, the holding reply wording, and each review reply on its own. The lead can edit the wording, drop conversations from the list, or dismiss the spike. Nothing is posted to Slack, to a conversation, or to a review before that.
What is recorded when a spike is approved?
The spike itself: the product or topic, the window, the baseline, the count, and the conversation and review ids behind it. Plus who approved the alert and the reply, when, and the tag on each conversation so the spike can be closed out.
Does it post to customers or reviewers on its own?
No. The holding reply is a draft agents choose to use. Review replies are proposed one at a time and each needs approval. The only automatic act is reading; every write waits for a person.
Ask about Sentiment Spike

Ask about the capabilities and requirements in this listing.

For project-specific requirements, contact Fibric.

This operator is developed, published, and supported by Fibric. Third-party names and logos identify the systems an integration connects to; they are the property of their respective owners, who are not affiliated with Fibric and do not sponsor or endorse this listing. Trademark policy