abhiroopb/feature-request-scanner

Feature Request Scanner

Daily automated scan of customer/seller feedback channels for feature requests. Classifies, scores urgency, deduplicates, and writes to a tracking spreadsheet — then promotes high-priority items to your roadmap tool.

Channels to Scan

Configure the channels relevant to your product area:

| Channel ID | Name | Workspace | Content Type | |---|---|---|---| | CHANNEL_001 | #customer-feedback | your-workspace | Raw in-app "Suggest a Feature" submissions | | CHANNEL_002 | #churn-signal-digests | your-workspace | AI-generated daily digests of churn signals | | CHANNEL_003 | #feedback-intake | your-workspace | GTM team structured submissions | | CHANNEL_004 | #strategic-churn-threats | your-workspace | Account Manager escalations for high-value customers |

Tracking

• Google Sheet: YOUR_SHEET_ID — tab Feature Requests
• Columns: Feature | Priority | Product Area | Detailed Description | Source | Requester | Status | Notes
• Last-run timestamp: ~/.config/feature-request-scanner/last-run
• Roadmap tool base: Configure your Airtable/Linear/Jira project for deduplication

Workflow

Step 1 — Determine Time Window

mkdir -p ~/.config/feature-request-scanner
LAST_RUN=$(cat ~/.config/feature-request-scanner/last-run 2>/dev/null || echo "")

If LAST_RUN is empty or older than 7 days, default to the last 24 hours. Otherwise scan from LAST_RUN to now. Use the Unix timestamp for Slack API's oldest parameter.

Step 2 — Fetch Messages from All Channels (in parallel)

For each channel, fetch messages since the last run using your Slack CLI:

slack-cli get-channel-messages \
  --channel-id <CHANNEL_ID> --workspace your-workspace --limit 50

Filter messages by timestamp — only keep messages newer than LAST_RUN.

For #customer-feedback: Messages follow the format "New feedback received:" with structured fields (Customer, Vertical, Platform, Feedback). Extract the Feedback field.

For #churn-signal-digests: These are daily digest posts. Parse the digest for individual items with links back to the feedback channel.

For #feedback-intake: Messages follow the format "Submission from @user" with structured fields (Team Name, Customer Type, Product, Request Details). Extract Product and Request Details.

For #strategic-churn-threats: Messages follow the format with structured fields (AM Name, Customer Name, Annual Revenue, Why is this customer a churn threat, What specific outcome...). Extract the churn reason and outcome fields.

Step 3 — Filter for Relevant Product Areas

Only keep messages that match your configured product area keywords (case-insensitive matching). Define keyword groups for each product area your team owns.

Example product areas and keywords:

E-commerce / Online Store:

• ecom, e-commerce, online store, online ordering, website, web store, digital storefront, online checkout, shipping, delivery, pickup

Customer Profiles:

• profile, profiles, customer profile, business profile, public page, landing page

Checkout:

• checkout, payment flow, tipping, tip, gratuity, receipt, payment screen

General POS:

• pos, point of sale, register, kiosk, items, catalog, inventory

Always include messages from #strategic-churn-threats regardless of keywords — these are high-signal by definition.

Discard messages that are clearly support issues (can't login, charge failing, etc.) rather than feature requests.

Step 4 — Auto-Classify Each Request

For each filtered message, determine:

| Field | Logic | |---|---| | Feature | Short name (≤ 10 words) summarizing the request | | Priority | See urgency scoring below | | Product Area | Tag with one or more from your configured product tags | | Detailed Description | 2-3 sentence summary of the request and rationale | | Source | Slack permalink to the original message | | Requester | Customer feedback (in-app), GTM / Sales (intake channel), AM escalation (churn threats), or Internal feedback | | Status | Needs triage | | Notes | Customer revenue if available, country, vertical, churn signal if present |

Product Tagging

Every request gets tagged with one or more product tags. Apply primary tags first — these are the areas your team owns. A request can have multiple tags.

• In the spreadsheet, write tags comma-separated in the Product Area column: Profiles, ECOM
• In the Slack summary, show the primary tag first: Profiles · ECOM
• If none of the primary tags match, use the best-fit secondary tag
• Requests tagged with at least one primary tag get a slight priority boost (they're in your domain)

Step 5 — Urgency Scoring

| Priority | Criteria | |---|---| | P0 | Explicit churn threat from high-value customer (>$1M revenue), OR blocking a deal, OR multiple customers requesting same thing in 24h | | P1 | Churn signal (mentions switching/leaving), OR strategic AM escalation, OR feature gap causing significant revenue impact | | P2 | Clear feature request with business value, multiple customers affected, or improving competitive position | | P3 | Nice-to-have, single customer request, or exploratory |

Step 6 — Deduplicate Against Existing Sheet & Roadmap

Before adding any row, check if a similar feature already exists in the tracking sheet and your roadmap tool.

6a. Check Google Sheet

Read existing rows from your tracking sheet and compare the new request's Feature name against existing rows (semantic match, not just exact string).

6b. Check Roadmap Tool

Search your roadmap tool (Airtable, Linear, Jira, etc.) for existing projects that match the request. Try 2-3 keyword variations from the feature name to catch matches.

6c. Handle Duplicates

If a match is found in either source:

• Don't add a duplicate row to the sheet
• Note it as "+1" in the presentation
• If the new request has a higher priority signal, flag it for priority upgrade
• Capture the existing status — is it In Progress, Backlog, Delivered, etc.?
• Build a link if a roadmap record exists

6d. Notify the Original Poster (with approval)

For each duplicate where the original Slack message has a clear poster (not a bot), offer to reply in-thread letting them know the request is tracked:

"This request matches an existing roadmap project. Reply to the poster? (y/n)"

If approved, reply in the original Slack thread with a status-appropriate message:

• In Progress: Let them know it's being worked on with a link
• Backlog: Acknowledge it's tracked and being evaluated
• Delivered: Good news — it's shipped! Share the link
• Sheet only (no roadmap record): Acknowledge it's tracked

Rules:

• Always ask for approval before posting any Slack reply
• Show the draft reply in chat first
• Don't reply to bot-posted messages — only to human posters
• Don't reply if the message is older than 7 days (stale threads)
• Batch the approvals: present all planned replies together so the user can approve/skip in bulk

Step 7 — Present Results for Review

Present all classified requests in a single table, grouped by priority:

## 🔍 Feature Request Scan — [Date]
Scanned [N] channels · [N] messages · [M] relevant requests found · [D] duplicates skipped

### 🔴 P0 — Critical
| # | Feature | Product Area | Source Channel | Customer/Revenue | Action |
|---|---------|-------------|----------------|-----------------|--------|

### 🟡 P1 — High
| # | Feature | Product Area | Source Channel | Signal | Action |
|---|---------|-------------|----------------|--------|--------|

### 🔵 P2 — Medium
| # | Feature | Product Area | Source Channel | Notes | Action |
|---|---------|-------------|----------------|-------|--------|

### ⚪ P3 — Low
| # | Feature | Product Area | Source Channel | Notes |
|---|---------|-------------|----------------|-------|

### 🔄 Duplicates (already tracked)
| # | Feature | Status | Roadmap Link | New Signal | Reply? |
|---|---------|--------|-------------|------------|--------|

Step 8 — Write to Google Sheet

For each new (non-duplicate) request, append to the tracking sheet with columns: Feature, Priority, Product Area, Description, Source, Requester, Status, Notes.

Step 9 — Promote to Roadmap (user-driven)

Nothing is filed to the roadmap without explicit approval. All items are written to the sheet first. Promotion happens interactively.

After presenting the results table, go through P0 and P1 items one at a time:

[Feature name] — [Product Area] · [Priority] · [1-line description] Promote to roadmap? (y / n / skip all)

For P2/P3 items, just mention they're in the sheet and can be promoted later.

Step 10 — Post Summary to Slack (with approval)

After presenting results and completing any roadmap filings, offer to post a formatted summary to a Slack channel. Only post after explicit user approval. Never auto-post.

Step 11 — Update Timestamp

date -u +%Y-%m-%dT%H:%M:%SZ > ~/.config/feature-request-scanner/last-run

Start-of-Day Integration

When run as part of start-of-day, present a condensed summary before other triage sections:

## 🔍 Feature Request Scan (since yesterday)
[N] new requests found across [N] channels · [P0] critical · [P1] high · [P2] medium

🔴 **P0:** [Feature name] — [Customer] threatening churn over [issue] → Promote to roadmap? (y/n)
🟡 **P1:** [Feature name] — [N] customers requesting [thing] → Promote to roadmap? (y/n)
📝 **P2-P3:** [N] items added to tracking sheet
🔄 **Duplicates:** [N] already tracked · [N] replies drafted

Weekly Rollup

Auto-generate a "Top 10 Feature Requests This Week" summary every week (or on demand). Write it to a shared doc for leadership review.

Trigger

• Automatic: During start-of-day on Mondays
• Manual: "weekly rollup", "feature request rollup", "top feature requests this week"

Rollup Workflow

1. Gather the week's data from the tracking sheet
2. Rank and select top 10 by priority, duplicate count, primary tag match, and signal strength
3. Write to shared doc with a formatted table and theme summary
4. Update timestamp for next rollup

Manual Trigger

The user can also trigger this skill manually:

• "scan feature requests"
• "check feedback channels"
• "run feature intake"
• "what's new in customer feedback?"
• "weekly rollup" / "feature request rollup"

Notes

• Digest channels produce daily AI digests — use these as a shortcut when available, but still scan raw channels for items the digest may have missed
• Bot/app messages in feedback channels follow consistent formats — parse the structured fields
• For strategic churn threats, always capture the revenue and customer name — these are high-signal for prioritization
• The tracking sheet and roadmap tool are complementary: the sheet is the intake funnel, the roadmap is where items get staffed and tracked
• When in doubt about product area classification, default to Other and let the user re-classify during triage