# Competitive Research Digest

> A weekly briefing on what your competitors actually shipped

| | |
|---|---|
| **Category** | Research |
| **Type** | AI Agent |
| **Difficulty** | Intermediate |
| **Automation** | Mostly automated |
| **Runtime** | ~30 min |
| **Est. cost** | $1.1 |
| **Version** | 1.0.0 |

## Purpose

Competitive monitoring degrades into a folder of unread alerts. This agent checks a defined set of sources weekly, compares against last week's snapshot, and reports only what actually changed — with a link to the evidence for each item.

The filtering is the product: an alert stream nobody reads is worse than no monitoring.

## Problems solved

- Alerts fire on noise and get muted
- Manual competitor checks slip when everyone is busy
- Pricing and positioning changes are noticed late
- No record of how a competitor evolved over time

## Benefits

- Reports changes, not activity
- Every item carries a source link to verify
- Week-over-week snapshots build a change history
- One briefing replaces many alert streams

## Limitations

- Only sees public information
- JavaScript-heavy sites need a rendering fetch to read correctly
- Cannot judge strategic intent — it reports what changed, not why
- Materiality assessment needs your context to be useful

## Success metrics

- Material changes surfaced per month
- False-positive rate in the digest
- Time from competitor change to internal awareness
- Digest read rate

## Prerequisites

- **Api key** — LLM API key
- **Account** — Web fetching service
- **Software** — Tracked source list
- **Account** — Storage for snapshots

## AI tools used

- **ChatGPT**
- **Claude**
- **Perplexity**

## Steps

### 1. Weekly trigger

Scheduled run kicks off against the tracked source list.

- _Time:_ 1 min

### 2. Fetch the tracked sources

Retrieve each page with rendering, normalise the content and store this week's snapshot.

- _Time:_ 6 min

> ⚠️ Respect robots.txt and rate limits — aggressive scraping gets your IP blocked and can breach terms of service

### 3. Diff against last week

Compare snapshots and discard cosmetic differences — timestamps, rotating testimonials, session tokens.

- _Time:_ 4 min

### 4. Assess materiality

Judge each change against your positioning and drop the immaterial ones.

- _Model:_ Claude Sonnet
- _Time:_ 8 min

```text
Assess these competitor changes for materiality to us.

Our positioning: {{our_positioning}}
Our target segments: {{segments}}

Changes detected:
{{change_set}}

For each change:
- what changed, in one line
- source URL
- materiality: high|medium|low|noise
- why it matters to us specifically, or why it does not

Be ruthless with "noise". A reworded headline is noise. A new pricing tier is not. If everything this week is noise, say so — an empty digest is a valid and useful result.
```

### 5. Write the digest

Compose the briefing: headline changes first, then detail, each with its source.

- _Model:_ GPT-4o
- _Time:_ 6 min

### 6. Distribute and archive

Post to the team channel and file in the competitive intelligence archive.

- _Tool:_ Slack
- _Time:_ 3 min

### 7. Team review

The team marks items as actioned or irrelevant, tuning next week's materiality judgement.

- _Time:_ 2 min

## FAQ

**Is monitoring competitor websites legal?**

Reading public pages generally is, but terms of service and rate limits still apply, and some jurisdictions treat aggressive automated access differently. Respect robots.txt and keep request rates modest.

**How do we stop the digest becoming noise?**

The materiality step exists for exactly this. Give it your real positioning, be strict about the "noise" category, and feed the team's actioned/irrelevant marks back in.

**Can it tell us why a competitor made a change?**

No, and any workflow claiming to is guessing. It reports what changed and links the evidence — the interpretation is yours.

---

_Exported from AIToolsay — https://invitationbuddy.com/workflows/competitive-research-digest_