How to Set Up an Effective Tag Taxonomy
Tags allow NEXT to automatically structure unstructured feedback from surveys, interviews, chats, and reviews. When designed well, tags enable powerful analysis such as:
- identifying root causes of problems
- comparing feedback across customer segments
- discovering drivers of satisfaction or dissatisfaction
- exploring topics within the customer journey
However, many teams initially create too many tags or the wrong types of tags. This guide explains a practical way to design tags and tagging jobs based on patterns observed in successful NEXT teamspaces.
Key Principles​
1. Tags should support analysis questions​
Tags are useful only if they help answer questions such as:
- Why are customers unhappy?
- Which product areas generate the most complaints?
- What issues do customers in Germany report?
- What drives detractors in NPS feedback?
If a tag does not help answer questions like these, it is probably unnecessary.
2. Keep the taxonomy simple​
A common mistake is creating hundreds of tags upfront. In practice, only a small subset ends up being used regularly in analysis .
Start small and expand based on real usage.
3. Separate segmentation from insights​
Good tagging systems distinguish between:
Segmentation tags Describe who the customer is
Examples:
- Germany
- Netherlands
- Store
- Web
- New Customer
- Returning Customer
Insight tags Describe what the feedback is about
Examples:
- Checkout issues
- Delivery delays
- Pricing complaints
- Product quality
Mixing segmentation and insight tags in one tagging job makes tagging less accurate and analysis harder.
The Most Useful Tag Categories​
Across multiple teamspaces, the following tag categories consistently enabled the most valuable analysis.
1. Experience Topics (Very Important)​
These describe what part of the experience customers are talking about.
Examples:
- Checkout
- Delivery
- Returns
- Customer Support
- Account Management
- Product Quality
- Pricing
These tags allow questions like:
"What problems occur most often during checkout?"
2. Pain Points (Very Important)​
These capture what went wrong.
Examples:
- Delivery delay
- Payment failure
- Confusing pricing
- Missing product information
- Poor customer support
These tags enable root cause analysis, such as:
"Why are customers unhappy?"
3. Sentiment (Important)​
These describe how customers feel about an experience.
Examples:
- Positive
- Neutral
- Negative
Or sometimes:
- Praise
- Complaint
- Feature Request
These tags help answer questions like:
"Which topics generate the most negative feedback?"
4. Customer Segmentation (Useful)​
These tags describe who the customer is.
Examples:
- Country
- Persona
- Customer type
- Device type
- Channel (Store, Web, App)
They enable questions such as:
"What issues do customers in Germany report?"
5. Business Signals (Optional)​
These tags capture signals relevant to business decisions.
Examples:
- Competitor mention
- Cancellation intent
- Upsell opportunity
- Product request
These can help identify strategic insights.
Tags That Often Add Little Value​
Many teams initially create very detailed tag systems. In practice, these tags are rarely used in analysis and often reduce tagging accuracy.
Several tag types appear frequently in configurations but rarely help analysis.
Overly Detailed Metadata​
Examples:
- Browser version
- App version
- Minor technical attributes
Unless your team regularly analyzes these, they usually add noise.
Highly Granular Product Lists​
Examples:
- Individual product SKUs
- Long lists of product collections
These create hundreds of tags but rarely improve insight.
A better approach is to tag product categories.
Example:
Product Experience
- Packaging
- Personalization
- Gift Options
- Product Quality
Large Demographic Taxonomies​
Examples:
Age 18-24
Age 25-34
Age 35-44
Age 45-54
Age 55-64
Age 65+
These are rarely used unless the team specifically analyzes demographic differences.
Recommended Tag Architecture​
A practical taxonomy typically contains 4–6 tag groups.
Example structure:
Country
- Germany
- Netherlands
- France
Persona
- New Customer
- Returning Customer
- Loyalty Member
Experience Topics
- Checkout
- Delivery
- Returns
- Customer Support
Pain Points
- Delivery delay
- Payment failure
- Missing information
- Technical issue
Sentiment
- Positive
- Neutral
- Negative
Business Signals
- Competitor mention
- Feature request
- Cancellation intent
Recommended Number of Tags​
A good starting range is:
| Tag Type | Recommended |
|---|---|
| Tag groups | 4–6 |
| Total tags | 25–60 |
| Tags per group | 5–15 |
Large taxonomies (100+ tags) often reduce tagging accuracy and are harder to maintain.
Designing Tagging Jobs​
Tagging jobs run AI classifiers that apply tags automatically to highlights.
Separating tagging jobs improves accuracy and maintainability.
Recommended structure:
Tagging Job 1: Experience Topics​
Purpose: classify what the feedback is about.
Example tags:
- Checkout
- Delivery
- Returns
- Customer Support
- Product Quality
Tagging Job 2: Pain Points​
Purpose: identify problems or friction.
Example tags:
- Delivery delay
- Payment failure
- Confusing pricing
- Technical bug
Tagging Job 3: Sentiment​
Purpose: detect emotional tone.
Example tags:
- Positive
- Neutral
- Negative
Tagging Job 4: Persona​
Purpose: identify customer attributes.
Example tags:
- New Customer
- Returning Customer
Optional Tagging Job: Business Signals​
Example tags:
- Competitor mention
- Feature request
- Cancellation intent
A Simple Starter Setup​
Recommended starting setup for most teams:
Tag Groups​
Experience Topics
Pain Points
Sentiment
Country
Persona
Example Tags​
Experience Topics
- Checkout
- Delivery
- Returns
- Customer Support
- Product Quality
Pain Points
- Delivery delay
- Payment failure
- Missing information
- Technical issue
Sentiment
- Positive
- Neutral
- Negative
Persona
- New Customer
- Returning Customer
Tagging Jobs​
Job 1: Experience Topics
Job 2: Pain Points
Job 3: Sentiment
Job 4: Persona
Iteratively Improving the Taxonomy​
The best tagging systems evolve based on how teams use chat and analysis.
A simple process works well:
Step 1: Start simple​
Create a small set of tags.
Step 2: Observe user questions​
Look at the queries people ask in NEXT, such as:
- "Why are customers unhappy with delivery?"
- "What problems occur during checkout?"
- "What feedback do we get from Germany?"
Step 3: Add tags when analysis requires them​
For example:
If people ask:
"What problems do customers experience during checkout?"
Add a tag:
Experience Topic
- Checkout
Step 4: Avoid premature complexity​
Only introduce new tags when they enable new insights.
Summary​
A good tag taxonomy should:
- focus on experience topics and pain points
- keep segmentation simple
- separate tagging jobs by purpose
- start small (25–60 tags is usually enough)
- evolve based on real analysis questions
The goal is not to describe everything in your data.
The goal is to make it easy to discover actionable insights from customer feedback.