The Feedback Pile
Your team already knows what customers want. It is in the support tickets, the NPS comments, the Slack threads where someone said “three people asked for this last week.” The problem is connecting it to what you are building.
When a PM asks “what are customers saying about onboarding?”, nobody should have to dig through five tools to answer that. This workflow puts customer feedback on your story map, next to the cards it relates to.
What You Need
- Cartographer with customers set up in the space.
- An AI assistant, such as Claude, ChatGPT, or similar, connected through the MCP server.
- Feedback in any format: support tickets, NPS exports, Slack threads, spreadsheet rows, or a summary you type out.
The Workflow
Paste or describe feedback, let your AI assistant categorize and place it, then tag the originating customer. The map accumulates evidence over time.
Feed in the Feedback
Use /import-research and paste the feedback into the conversation. Your AI assistant categorizes what it finds: feature requests become story cards, complaints and pain points become risk cards, unanswered questions become question cards.
You can also describe feedback in plain language: “Acme Corp has asked three times for bulk export, and two support tickets mention CSV formatting issues.” Your AI assistant creates the matching cards from your description.
Tag with Customers
After cards are created, attach them to the originating customer using cards_customers_set. If the customer does not exist yet, your AI assistant creates them with customers_create, including any context you provide about account tier, segment, or importance.
A card tagged with three enterprise customers tells a different story for prioritization than one with no customer attached.
Place on the Map
Your AI assistant reads your existing board structure and maps each piece of feedback to the right journey and step. An onboarding complaint lands under the onboarding journey. A billing feature request goes to the payment step.
Feedback that does not fit any existing step is a signal. It means customers are talking about parts of the experience your map does not cover yet, a gap worth investigating.
Filter by Customer
Once feedback is tagged, use board views to filter by customer. This shows you everything tied to one account: every card they have influenced, every risk they have raised, every feature they are waiting on.
Create saved views for key accounts. Before a QBR or renewal conversation, pull up the customer’s view to see where things stand. Share the view externally with a password-protected link so the customer sees their roadmap items without accessing your full board.
Keeping the Map Current
Import feedback on a regular cadence: weekly, after each support review, or whenever a batch of NPS responses comes in. The map accumulates evidence over time, and patterns that were invisible in a spreadsheet become clear once you see clusters of customer cards on specific steps.
Instead of debating which feature matters more, you can point to the map: “Four enterprise accounts have cards on this step, and two of them flagged it as a risk.”
