Feature Adoption Analysis

Analyze how features are being adopted across your user base, identify champions and laggards, and track adoption trends over time.

Tools

This skill uses Pendo MCP tools exclusively. All tool references below (e.g., searchEntities, activityQuery, segmentList, visitorQuery) refer to the Pendo connector tools (prefixed Pendo: in the tool list).

Before starting, call Pendo:list_all_applications to get the available subscription IDs and app IDs. Every subsequent Pendo tool call requires a subId (subscription ID) and most require an appId. If the user hasn't specified which app or subscription to use and there are multiple options, ask them to confirm before proceeding.

Parameters

• feature_name: The feature to analyze. Can be a partial name, exact name, or feature ID.
• timeframe: The time period for analysis (default: last 30 days).
• comparison_period: Optional previous period for trend analysis (defaults to the equivalent prior period).

Step 0: Find and Confirm the Feature

Users often provide partial or ambiguous feature names. Never silently pick a feature — always confirm when there's any ambiguity.

How to find the feature

1. Use Pendo:searchEntities with itemType: ["Feature"] and the user's search term to find matching features.

2. If the search returns exactly one result, confirm with the user: show the feature name, ID, and any relevant metadata, and ask "Is this the right feature?"

3. If the search returns multiple results, present a disambiguation list so the user can pick the right one. For each result, show:

   • Feature name and ID
   • Tagged app (if available)
   • Any description or page association

   Mark the most likely feature with a ⭐ Recommended tag based on:

   • Whether it belongs to the user's primary app
   • Whether it has recent activity
   • Name similarity to the search term

   Example:

   I found several features matching "dashboard". Which one are you looking for?
   
   1. ⭐ **Dashboard Main View** (ID: abc123) — Recommended
      - App: Web App | Last active: Feb 2026
      → *Closest name match with recent activity*
   
   2. **Dashboard Settings Panel** (ID: def456)
      - App: Web App | Last active: Jan 2026
   
   3. **Admin Dashboard** (ID: ghi789)
      - App: Admin Portal | Last active: Feb 2026
   

4. If no results are found, let the user know and suggest alternative search terms or ask them to double-check the name.

5. Do not proceed to the adoption report until the user has confirmed which feature to analyze.

Step 1: Collect Adoption Metrics

Once the feature is confirmed, gather core adoption data. Run these queries together to save time:

Tools: Pendo:activityQuery

• Unique visitors who used this feature in the timeframe
• Unique accounts with users of this feature
• Total event count for the feature
• Previous period metrics for the same duration (for trend comparison)

When constructing activity queries, filter by the confirmed feature ID and use the appropriate time period. Request both numVisitors and numEvents where possible.

Step 2: Identify User Segments

Understand who is and isn't using the feature:

Tools: Pendo:activityQuery, Pendo:visitorQuery

• Power Users (Champions): Top 10 visitors by event count for the feature. These are your best candidates for case studies, beta testing, and internal advocacy.
• Recent Adopters: Visitors who first used the feature in the last 7 days. Useful for understanding current momentum.
• Top Accounts: Accounts with the most feature users, ranked by unique visitor count.

For each power user, try to include their account name so the data is actionable. The Pendo:activityQuery grouped by visitorId may not always return a clean account name — if needed, cross-reference by running a separate Pendo:activityQuery grouped by accountId for the same feature to get account-level context, or infer the account from the visitor's email domain.

Step 3: Calculate Adoption Rate

Compare feature users against the total active user base to get a meaningful adoption percentage:

Tools: Pendo:activityQuery, Pendo:segmentList

1. Get total unique active visitors across the entire product in the same timeframe.
2. Calculate adoption rate: (feature users / total active visitors) × 100
3. If segments are available (via Pendo:segmentList), calculate adoption rate per segment for additional insight — this often reveals that adoption is strong in one segment but weak in another. Note: some subscriptions have thousands of segments, which will cause Pendo:segmentList to fail without a substring filter. If this happens, note in the report that segment-level breakdown is available if the user specifies a segment of interest, and move on — don't let this block the rest of the report.

The adoption rate relative to active users is more useful than against all visitors, since inactive users aren't a realistic adoption target.

Step 4: Analyze Trends

Look at how adoption is changing over time:

Tools: Pendo:activityQuery

• Query daily or weekly unique visitors for the feature over the timeframe
• Identify growth or decline patterns
• Note any significant spikes (possible correlation with launches, guides, or announcements) or drops (possible bugs, UX issues)
• Compare the current period's total against the previous period to calculate percent change
• Partial periods: The first or last week/day in a trend may be incomplete (e.g., a 30-day range ending mid-week). Flag partial periods in the report so the user doesn't misread a shorter bucket as a decline.

If the timeframe is 14 days or less, use daily granularity. For longer periods, use weekly.

Output Format

Generate a structured feature adoption report:

## Feature Adoption Report: {feature_name}
**Feature ID**: {feature_id}
**Period**: {timeframe}

### Adoption Overview
- **Total Users**: {unique_visitors} visitors across {unique_accounts} accounts
- **Adoption Rate**: {adoption_rate}% of active users ({feature_users} / {total_active_users})
- **Total Events**: {event_count} interactions
- **Trend**: {trend_direction} ({percent_change}% vs previous period)

### Power Users (Champions)
| Rank | Visitor | Account | Events |
|------|---------|---------|--------|
| 1    | {visitor_1} | {account_1} | {events} |
| 2    | {visitor_2} | {account_2} | {events} |
| ...  | ...     | ...     | ...    |

### Top Accounts by Adoption
| Rank | Account | Users | Events |
|------|---------|-------|--------|
| 1    | {account_1} | {user_count} | {events} |
| 2    | {account_2} | {user_count} | {events} |
| ...  | ...     | ...   | ...    |

### Adoption Trend
{weekly_or_daily_trend_summary — describe the trajectory in words, noting any inflection points}

### Segment Adoption (if available)
| Segment | Adoption Rate | Users |
|---------|--------------|-------|
| {segment_1} | {rate}% | {count} |
| ...     | ...          | ...   |

### Insights & Recommendations
- {insight based on the data — e.g., "Adoption is concentrated in 3 accounts, suggesting broad rollout hasn't happened yet"}
- {actionable recommendation — e.g., "Consider targeting accounts with high overall activity but zero feature usage for outreach"}
- {trend insight — e.g., "Week-over-week growth has been steady at ~5%, indicating organic discovery"}

Rules

• Never skip feature confirmation. If the search returns more than one result, always present options with a ⭐ Recommended tag and wait for the user to choose. This is the most important rule.
• Always explain the recommendation. Don't just star a feature — briefly say why.
• Always pass the confirmed feature ID explicitly in all subsequent queries.
• Default timeframe is 30 days if not specified.
• Calculate adoption rate relative to total active users, not all visitors — this gives a more honest and useful number.
• When identifying power users, include their account name so the data is actionable for outreach.
• If a query returns no results (e.g., zero usage, no segments), note it briefly in the report rather than leaving the section blank. Zero adoption is itself a meaningful finding.
• Provide actionable recommendations based on the data — not generic advice but specific next steps tied to what the numbers show (e.g., which accounts to target, whether to invest in guides or announcements).
• If the user asks about multiple features, generate separate reports for each, with a brief comparison summary at the top.