Turn scattered feature requests into a prioritized roadmap. This workflow monitors your Zendesk tickets for feature requests, uses AI to extract, categorize, and deduplicate them, scores requests by customer value and frequency, and maintains a living product backlog in Notion that your team can actually use.
Receive closed ticket from Zendesk
The workflow triggers when tickets are closed in Zendesk. It captures the full conversation, customer information including plan type and account value, any tags applied by support, and the resolution outcome.
Extract feature requests with OpenAI
Using OpenAI, the workflow analyzes the ticket to identify any feature requests, whether explicit ('I wish you had X') or implicit ('I had to use a workaround because'). It extracts the core request, the use case behind it, and any specific requirements mentioned.
Match and deduplicate against existing requests
The AI compares the extracted request against your existing backlog to find matches. Similar requests are grouped together, incrementing the request count and adding the new customer to the list of requesters. Unique requests are added as new items.
Score and prioritize requests
Each request is scored based on frequency (how many customers asked), customer value (ARR of requesters), strategic alignment (does it fit your roadmap), and effort estimate (based on AI analysis of complexity). The backlog is automatically sorted by priority score.
Alert product team of trending requests
When a request crosses threshold (e.g., 10 customers or $50k ARR requesting), an alert is sent to Slack. The alert includes the request details, top requesting customers, use cases mentioned, and a link to the Notion backlog item.
Why automate feature request tracking?
Feature requests are buried in support tickets, scattered across spreadsheets, and forgotten in Slack threads. Without systematic tracking, you build what's loudest, not what's most valuable. AI automation creates a single source of truth.
Never lose a feature request again
Every request gets captured, categorized, and tracked. No more discovering that 50 customers asked for the same thing over the past year but nobody connected the dots.
Prioritize by customer value, not volume
Ten enterprise customers requesting a feature might be more valuable than 100 free users. The priority scoring weighs requests by the ARR of customers asking, not just raw count.
Show customers their feedback matters
When you ship a requested feature, you know exactly who asked for it. Close the loop with customers who requested it and turn support interactions into retention opportunities.
How to set up feature request tracking
Setting up this request tracking workflow takes about 15 minutes. You'll connect Zendesk, configure your Notion backlog, and set alert thresholds.
What you need to get started
- Zendesk account with ticket access
- OpenAI API key for request extraction
- Notion workspace for backlog management
- Slack workspace for trend alerts
Setting up your Notion backlog
- Create a database with fields for request title, description, and status
- Add properties for request count, total ARR, and priority score
- Configure linked database of requesting customers
- Set up views for different priority tiers and product areas
Configuring priority scoring
- Set weights for frequency vs. customer value vs. strategic fit
- Define what ARR thresholds matter for your business
- Configure effort estimates (simple, medium, complex)
- Set alert thresholds for when to notify product team
Frequently asked questions about feature request tracking
How accurate is AI at extracting feature requests?
AI is excellent at catching explicit requests and good at inferring implicit ones from workarounds or complaints. It may miss very subtle hints, but catches the vast majority of actionable feedback.
What if the same request is phrased differently?
The AI uses semantic matching, not keyword matching. 'I want dark mode' and 'Can you make the interface easier on the eyes at night' are recognized as the same request and grouped together.
Can I pull requests from sources other than Zendesk?
Yes, you can extend the workflow to monitor Intercom, email, social media, or app store reviews. The AI extraction and Notion management work with any text source.
How do I handle requests we'll never build?
You can mark requests as 'Won't Do' with a reason. This keeps your backlog clean while maintaining the record. Some teams create a separate view for declined requests with explanations.