Turn 30 support tickets into a prioritized spec

Most teams sit on a goldmine of support data and never turn it into shipped fixes. The tickets get tagged, filed, and forgotten. Nobody can say later which ticket justified which decision.

This use case walks the loop Pathmode actually runs: raw tickets in, unreviewed signals in the Inbox, and a spec where each claim states what backs it and who judged it.

The input

A CSV export from Zendesk, Intercom, or HelpScout. Three columns minimum: timestamp, customer message, tag. The example below is anonymized from a real B2B SaaS team.

2026-04-12, "Tried to invite my designer but the email never arrived", onboarding
2026-04-12, "Invite link expired before I could click it, had to ask again", onboarding
2026-04-13, "How do I add someone to the team? Can't find the button", onboarding
...

The flow

1. Import the export. Open the product's Evidence tab and use Import for CSV, TSV, XLSX, or PDF files. Each row becomes its own signal with a type (friction, quote, observation, metric, request) and its source preserved.

2. Every imported row lands Unreviewed. Importing is not endorsement. The signals go to the Inbox and stay there until a signed-in person judges them. Nothing an import produces backs a claim on its own.

3. Write the spec. Open the Intents tab, start a new intent, and describe the problem the tickets point at. The spec's objective, outcomes, constraints, and edge cases are the claims you will have to back.

4. Work the Inbox. Each signal gets exactly one of three judgments: Use as backing, Keep unreviewed, or Dismiss. Use as backing asks you which intent and which exact claim, then reviews and anchors the signal in one action, stamped with your name and the time.

5. Read the claim states. Open the spec's Evidence section. Every claim shows Backed, Unreviewed, Assumption, or No evidence. Expand one to see the exact items behind it, each with its tier, origin, source, and latest reviewer.

The output

An IntentSpec where:

• Every claim states its basis rather than implying one
• Reviewed backing is separated from unreviewed signals and AI-drafted assumptions
• Each judgment carries a name and a time, so "why did we build this?" has an answer
• No evidence is descriptive, not a failure. It tells you where to look next

Try it yourself

1. Sign in to Pathmode
2. Open any product, switch to the Evidence tab
3. Click Import and drop in your support export
4. Open Intents, start an intent for the problem the tickets describe
5. Return to the Evidence Inbox and use the strongest signals as backing for the exact claims they support

Related

• Playbook: From Support Ticket to Shipped Feature
• Use case: Spot the conflict between two stakeholder requests
• Use case: Find the friction pattern across 5 user interviews