MCP Weekly Growth Report

MCP Weekly Growth Report

Turn scattered operating signals into a weekly or daily report with five actions your team can complete quickly.

Customer story: How Gourmet Ads uses Zapier MCP to turn Salesforce and Atlassian into a weekly growth report.

What This Skill Does

This skill instructs an AI agent to:

1. Read the source systems your team already uses: CRM, analytics, email/request signals, and team workspace docs.
2. Find small fixes with outsized impact: broken links, fake traffic, stalled leads, missing CRM fields, neglected content areas, or quiet accounts.
3. Rank recommendations by impact and effort.
4. Write the report into the team's operating surface.
5. Keep each action scoped so someone can complete it quickly.

The result: the team starts the week with a short work queue, not another dashboard to interpret.

When You Need This Pattern

Use this when:

• Your CRM is the system of record, but the team does not live in the CRM all day.
• Marketing, sales, or operations data is split across CRM, analytics, inboxes, spreadsheets, and docs.
• Weekly reports take hours to assemble and rarely surface surprises.
• Team members need a short list of specific fixes, not a broad business review.
• A founder, operator, or manager wants to test reporting ideas before asking engineering to build them.

If your team already has a trusted operating report that creates clear owner-assigned actions every week, you probably do not need this skill.

Architecture

AI client
  -> Zapier MCP
     -> CRM: leads, accounts, owners, opportunities
     -> Analytics: traffic, conversions, source performance
     -> Requests: email, forms, media kit inboxes
     -> Workspace: Confluence, Notion, Docs, Slack, email

Required Connections Via Zapier MCP

• CRM: read accounts, leads, opportunities, owners, last activity, and key custom fields.
• Analytics or source-of-truth metrics: read sessions, conversions, traffic sources, content performance, or campaign data.
• Team workspace: write the final report where people already work.
• LLM: Claude, GPT, or Gemini for analysis, ranking, and summary writing.

Optional:

• Email or shared inbox for request signals.
• Google Sheets or a warehouse if metrics are staged outside the source app.
• Slack or Teams for notifications.
• Calendar or call tool for meeting/activity context.

Core Workflow

1. Define The Report Surface

Pick one audience and one cadence.

Examples:

Every Monday, write a marketing growth report in Confluence for the marketing team.

Every weekday at 7:30 a.m., write a sales action board for each rep in Notion.

Do not start with "analyze everything." Start with the decision the report should make easier.

2. Map Source Systems

Only read fields needed for the report. Avoid pulling full customer records or raw email bodies unless the recommendation depends on them.

3. Set Recommendation Rules

The report should return exactly five action items by default.

Each action must include:

• Action: what to do.
• Owner: person, role, or team.
• Why now: the signal that triggered the recommendation.
• Evidence: source app and key data point.
• Effort: expected time.
• Link: where to act.

Reject recommendations that are vague, unactionable, or too large.

4. Analyze For High-Signal Patterns

Ask the agent to look for:

• Broken or outdated links in high-traffic surfaces.
• Fake, spam, or suspicious traffic that distorts analytics.
• Stalled leads or accounts with no recent engagement.
• Missing CRM fields that block follow-up.
• Content gaps between related brands, products, or audience segments.
• Campaigns with traffic but low conversion.
• Pages, requests, or sources that spiked unexpectedly.
• Repeated manual work that should become a workflow.

Optimize for useful actions, not interesting findings.

5. Write The Report

Use this report format:

# Weekly growth report: {{date_range}}

## Summary
{{2-3 sentences on what changed and where the team should focus.}}

## 5 recommended actions

1. **{{action headline}}**
   - Owner: {{owner}}
   - Why now: {{signal}}
   - Evidence: {{source + datapoint}}
   - Effort: {{time estimate}}
   - Link: {{source link}}

## Watchlist
- {{Signals worth monitoring but not urgent enough for the top 5}}

## Data checked
- CRM: {{objects/date range}}
- Analytics: {{properties/date range}}
- Requests/inbox: {{query/date range}}
- Workspace: {{docs/pages checked}}

Guardrails

• Keep private operational details inside the customer's workspace.
• Use summaries and source links instead of dumping records into the report.
• Treat the report as decision support, not automatic execution.
• Do not write back to CRM fields unless the user explicitly asks for CRM updates.
• Put low-confidence findings in the watchlist instead of the top five.
• Never include API keys, private URLs, or sensitive customer details in shared examples.

Inspired By

Built in production by Benjamin Christie, President of Gourmet Ads. Read the story: How Gourmet Ads uses Zapier MCP to turn Salesforce and Atlassian into a weekly growth report.