What Is Unwrap?

Unwrap is a customer intelligence platform that brings all of your customer feedback together, from support tickets and app reviews to surveys and sales calls. It shows you what customers are saying and what's changing.

You can use Unwrap to analyze feedback from every channel in one place. It uses natural language processing (NLP) to group feedback into themes, label its sentiment, and flag feature requests, product defects and churn risk.

Product, customer experience (CX) and voice of customer (VoC) teams use it to act on what matters most.

Here's an overview of how Unwrap works, the parts of the platform and the terms you'll see along the way.

How Unwrap Works

Unwrap works in 3 steps: it collects feedback from every source you connect, reads and groups every entry, then brings what it finds to you.

Collect: Bring In Feedback From Every Source

First, Unwrap ingests your written feedback wherever it lives. That includes:

• Public sources, like app store reviews, social media sites such as Reddit, X and Facebook, and review sites such as G2, Capterra and Trustpilot

• Internal sources, like Zendesk tickets, Intercom conversations, Net Promoter Score (NPS) surveys and sales call transcripts from tools such as Gong and Zoom

• Data warehouses, like Snowflake and BigQuery

There are hundreds of direct integrations, plus 3,000+ more through a partnership with Zapier. We can also build new integrations as requested. Once a source is connected, new feedback comes in automatically each night. See Adding Data.

Understand: Group and Score Every Entry

Next, natural language processing, a form of machine learning, "reads" all of this text for you and pulls out the most actionable trends and insights.

Every piece of feedback is split into entries. Entries that describe the same issue are grouped together, even when customers word it differently.

New patterns are picked up each time feedback comes in. You can also build your own Groups by describing what you want to track.

Every insight links back to the original feedback, so you can always check what customers actually said.

Your Taxonomy: How Groups Are Organized

Groups sit in a hierarchy called your taxonomy, so you can quantify both broad areas of your product and the specific problems inside them:

• Level 1 (L1) Groups are high-level areas, like "Account Management" or "Social Sharing."

• Level 2 Groups get more specific, like "Difficulties creating accounts" or "Errors sharing to Facebook."

• Level 3 and 4 Groups get even more specific and use exact customer language, like “I’m getting an error message every time I enter my name”

A parent Group's count includes everything in its child Groups. When a new pattern turns up, Unwrap creates a Group and nests it under the right parent automatically.

You stay in control. You can build your own Groups, change where a Group sits in the tree, and add or remove feedback. Each correction teaches Unwrap to make similar decisions better next time.

During onboarding, Unwrap can also work from your existing issue or feedback taxonomy. See the Taxonomy Page and Taxonomy Best Practices.

Labels on Every Piece of Feedback

Every piece of feedback is labeled for:

• Feedback type, as a Feature Request, a Product Defect or both

• Business impact, as Churn Risk (customers at risk of leaving), Upsell Opportunity (customers looking to expand, upgrade or buy more) or both

Act: Get Insights in Dashboards, Alerts and Reports

You'll find what Unwrap uncovers in several places:

• The Home page shows insights picked out for everyone in your View.

• Boards are dashboards of charts you build to track the issues and trends you care about.

• Real-time alerts tell you, in Slack or email, when a new issue appears or an existing one spikes.

• Subscription reports send an updated analysis of your feedback to email or Slack on a schedule you choose.

• Reports turn an Unwrap Assistant conversation into a shareable write-up with charts. You can send one once, schedule it or download it as a PDF.

• Linked Actions connects Jira, Asana or Linear, so you can create a ticket straight from a Group.

With Linked Actions, you'll also see the change in related complaints once your team resolves an issue, so you know whether the fix worked.

The Parts of Unwrap

Unwrap's Data Model

Here are a few terms used to describe the parts of the Unwrap platform.

Entries, Sentiment and Groups

Example: One Feedback Entry, Broken Down

Take this fictitious support ticket about Spotify:

Administration and Access

Users and data are organized in 2 layers:

• An Organization is the set of people you want to collaborate with. Everyone in it has access to the same data.

• A View is where you integrate one or more data sets (your Zendesk feed would be one data set) that you want insights on. An Organization can have many Views, such as one for public mobile app feedback and another for NPS data.

Every user is either an Admin or a Member of the Organization. Admins invite and remove users, manage permissions, and are the only users who can create or change integrations and custom fields.

Within each View, users also get a role:

• Contributors can edit the taxonomy and correct AI decisions.

• Viewers can analyze feedback, build Boards and use the Assistant without changing the taxonomy.

• Repliers can respond to customers where replying is enabled.

Teams can sign in with single sign-on through SSO providers like Okta or Microsoft Entra. We support SSO through both OIDC and SAML. See Data organization and permissions.

Get Started With Unwrap

1. Create an account and join your team's Organization. If you're the first person at your company to use Unwrap, contact your Unwrap representative to activate your Organization.

2. Read Data organization and permissions to learn how Organizations, Views and roles work.

3. Add data by connecting your first feedback source.

4. Connect Slack so alerts and reports reach your team where it already talks.

5. Ask the Unwrap Assistant your first question about your feedback.

Need help? Email us at contact@unwrap.ai.