operations workflow

Customer Signal Miner

Find recurring customer problems without losing the evidence behind each theme.

Setup: 12 minutes
Difficulty: Beginner
Tested: 2026-07-26

# Customer Signal Miner

Turn scattered feedback into a ranked, traceable product-opportunity brief.

**Outcome:** Find recurring customer problems without losing the evidence behind each theme.

**Category:** operations

**Setup:** 12 minutes · Beginner

## Requirements

- Anonymized feedback items with stable IDs
- A product-area list and scoring scale

**Tools:** claude, n8n

## Expected output

A ranked opportunity table with evidence IDs and unanswered questions

## Steps

### 1. Normalize the inputs

Assign an ID, channel, date, customer segment, and exact quote to each item.

**Prompt**

```text
Normalize these feedback items without paraphrasing the quote. Return missing metadata as null.
```

### 2. Cluster by customer job

Group items by the blocked outcome rather than by requested feature wording.

**Prompt**

```text
Cluster by customer job and problem. Attach every item ID once and do not name a feature solution.
```

### 3. Score the signal

Score frequency, urgency, and strategic fit separately and retain contradictory evidence.

**Prompt**

```text
Produce a score table, explain each score from evidence IDs, and list disconfirming feedback.
```

### 4. Prepare a decision brief

Rank opportunities and identify the smallest next validation action for each.

**Prompt**

```text
Write a one-page brief with ranked problems, evidence, uncertainty, and one next interview question per problem.
```

## Limitations

The ranking reflects the supplied sample and can amplify collection bias or overrepresented customer segments.

## Privacy

Anonymize names, email addresses, account IDs, and confidential product details before processing.

## Sources

- https://claude.ai/
- https://n8n.io/

Tested 2026-07-26 · Updated 2026-07-26

# Customer Signal Miner

Turn scattered feedback into a ranked, traceable product-opportunity brief.

**Outcome:** Find recurring customer problems without losing the evidence behind each theme.

**Category:** operations

**Setup:** 12 minutes · Beginner

## Requirements

- Anonymized feedback items with stable IDs
- A product-area list and scoring scale

**Tools:** claude, n8n

## Expected output

A ranked opportunity table with evidence IDs and unanswered questions

## Steps

### 1. Normalize the inputs

Assign an ID, channel, date, customer segment, and exact quote to each item.

**Prompt**

```text
Normalize these feedback items without paraphrasing the quote. Return missing metadata as null.
```

### 2. Cluster by customer job

Group items by the blocked outcome rather than by requested feature wording.

**Prompt**

```text
Cluster by customer job and problem. Attach every item ID once and do not name a feature solution.
```

### 3. Score the signal

Score frequency, urgency, and strategic fit separately and retain contradictory evidence.

**Prompt**

```text
Produce a score table, explain each score from evidence IDs, and list disconfirming feedback.
```

### 4. Prepare a decision brief

Rank opportunities and identify the smallest next validation action for each.

**Prompt**

```text
Write a one-page brief with ranked problems, evidence, uncertainty, and one next interview question per problem.
```

## Limitations

The ranking reflects the supplied sample and can amplify collection bias or overrepresented customer segments.

## Privacy

Anonymize names, email addresses, account IDs, and confidential product details before processing.

## Sources

- https://claude.ai/
- https://n8n.io/

Tested 2026-07-26 · Updated 2026-07-26

{
  "slug": "customer-signal-miner",
  "title": "Customer Signal Miner",
  "description": "Turn scattered feedback into a ranked, traceable product-opportunity brief.",
  "outcome": "Find recurring customer problems without losing the evidence behind each theme.",
  "category": "operations",
  "tags": [
    "customer-feedback",
    "product",
    "research"
  ],
  "setupMinutes": 12,
  "difficulty": "Beginner",
  "featured": true,
  "testedAt": "2026-07-26",
  "updatedAt": "2026-07-26",
  "tools": [
    "claude",
    "n8n"
  ],
  "inputs": [
    "Anonymized feedback items with stable IDs",
    "A product-area list and scoring scale"
  ],
  "expectedOutput": "A ranked opportunity table with evidence IDs and unanswered questions",
  "steps": [
    {
      "title": "Normalize the inputs",
      "instruction": "Assign an ID, channel, date, customer segment, and exact quote to each item.",
      "prompt": "Normalize these feedback items without paraphrasing the quote. Return missing metadata as null."
    },
    {
      "title": "Cluster by customer job",
      "instruction": "Group items by the blocked outcome rather than by requested feature wording.",
      "prompt": "Cluster by customer job and problem. Attach every item ID once and do not name a feature solution."
    },
    {
      "title": "Score the signal",
      "instruction": "Score frequency, urgency, and strategic fit separately and retain contradictory evidence.",
      "prompt": "Produce a score table, explain each score from evidence IDs, and list disconfirming feedback."
    },
    {
      "title": "Prepare a decision brief",
      "instruction": "Rank opportunities and identify the smallest next validation action for each.",
      "prompt": "Write a one-page brief with ranked problems, evidence, uncertainty, and one next interview question per problem."
    }
  ],
  "limitations": "The ranking reflects the supplied sample and can amplify collection bias or overrepresented customer segments.",
  "privacy": "Anonymize names, email addresses, account IDs, and confidential product details before processing.",
  "sources": [
    "https://claude.ai/",
    "https://n8n.io/"
  ],
  "related": [
    "support-inbox-triage"
  ]
}

Requirements

Bring these inputs.

Anonymized feedback items with stable IDs
A product-area list and scoring scale

Output contract

Know what done looks like.

A ranked opportunity table with evidence IDs and unanswered questions

Execution

Run the playbook.

01
Normalize the inputs
Assign an ID, channel, date, customer segment, and exact quote to each item.
```
Normalize these feedback items without paraphrasing the quote. Return missing metadata as null.
```
02
Cluster by customer job
Group items by the blocked outcome rather than by requested feature wording.
```
Cluster by customer job and problem. Attach every item ID once and do not name a feature solution.
```
03
Score the signal
Score frequency, urgency, and strategic fit separately and retain contradictory evidence.
```
Produce a score table, explain each score from evidence IDs, and list disconfirming feedback.
```
04
Prepare a decision brief
Rank opportunities and identify the smallest next validation action for each.
```
Write a one-page brief with ranked problems, evidence, uncertainty, and one next interview question per problem.
```

Field note

How to use this route.

This is the primary acceptance workflow for Indie Agent Hub. Every conclusion stays attached to source IDs so a solo founder can inspect the evidence instead of trusting a polished summary.

Bring these inputs.

Know what done looks like.

Run the playbook.

Normalize the inputs

Cluster by customer job

Score the signal

Prepare a decision brief

How to use this route.

Keep moving through the atlas