📬 Automated Weekly Client Report

English | Polski 🇵🇱

Marketing agencies waste 1–2 hours every week manually compiling client reports: copying sales numbers, re-typing Google reviews, formatting emails. This automation does it all — the report lands in the client's inbox every Monday at 7:00 AM, before anyone starts their workday.

▶️ View a live sample report · Setup guide · Demo script

---

🔄 How it works

flowchart LR
    A(["⏰ Schedule<br/>Mon 7:00 AM"]) --> B["⚙️ Config<br/>(single edit point)"]
    B --> C["📊 Google Sheets<br/>sales data"]
    C --> D["⭐ Firecrawl<br/>Google reviews scrape"]
    D --> E["🧮 Code<br/>week-over-week deltas"]
    E --> F["🤖 OpenAI<br/>summary + recommendation"]
    F --> G["📄 Code<br/>render HTML template"]
    G --> H(["✉️ Resend<br/>email to client"])

    D -. "scrape fails →<br/>report without reviews" .-> E
    F -. "AI fails →<br/>numeric fallback" .-> G

    style A fill:#111827,color:#fff
    style H fill:#16a34a,color:#fff

Loading

One n8n workflow, 8 nodes, zero manual clicks. Every new client = editing one Set node (business name, Google Maps URL, e-mail addresses).

✨ What the report contains

Section	Source	Example
💰 Revenue + WoW change	Google Sheets (POS export)	14 280 zł (+4.6%)
🛒 Transactions & avg. basket	Google Sheets	98 transactions, 145.70 zł
⭐ New Google reviews & rating	Firecrawl scrape of public Maps listing	3 new, 4.6/5
🤖 AI summary (2–3 sentences)	OpenAI gpt-4o-mini	"Revenue grew 4.6% to 14 280 zł..."
💡 One actionable recommendation	OpenAI gpt-4o-mini	"Check if the June promo is still live..."

Note

Why scraping instead of the official Google Business Profile API? The official API requires a business-use-case approval process that can take weeks. Scraping the public listing via Firecrawl ships the demo in hours — and the architecture is designed so the real API can be swapped in later without rebuilding anything.

🛡️ Failure-resilient by design

flowchart TD
    S{"Sheet has<br/>sales data?"} -- no --> X["❌ Stop with clear error<br/>(a report without sales data is useless)"]
    S -- yes --> R{"Firecrawl<br/>scrape OK?"}
    R -- no --> R2["⚠️ Continue without reviews section"]
    R -- yes --> AI{"OpenAI<br/>responds?"}
    R2 --> AI
    AI -- no --> AI2["⚠️ Numeric fallback summary"]
    AI -- yes --> OK(["✅ Report delivered"])
    AI2 --> OK

    style X fill:#dc2626,color:#fff
    style OK fill:#16a34a,color:#fff

Loading

All 5 scenarios are covered by logic tests that run the actual code from the workflow file (not a copy): happy path, scrape failure, AI failure, empty sheet, single week of data.

🚀 Deploy it yourself (~2 hours)

1. Get API keys — OpenAI, Firecrawl (free tier), Resend (free tier)
2. Create a Google Sheet with a sales tab — columns as in mock-data/sales-example.csv
3. Import workflows/weekly-report.json into any n8n instance (cloud or self-hosted) and attach credentials
4. Edit the Konfiguracja node — business name, Google Maps URL, recipient & sender e-mail
5. Execute manually to test → activate the workflow

Full step-by-step instructions: docs/setup.en.md (wersja polska)

📦 Report data schema (JSON)

{
  "period": { "from": "2026-07-06", "to": "2026-07-12" },
  "sales": {
    "revenue": 12450,
    "revenue_change_pct": -12.3,
    "transactions": 87,
    "avg_basket": 143.1
  },
  "reviews": {
    "new_count": 3,
    "average_rating": 4.6,
    "latest": [
      { "author": "Anna K.", "rating": 5, "text": "Bardzo miła obsługa." }
    ]
  },
  "ai_summary": "Sales dropped 12% vs the previous week, yet customer reviews remain excellent...",
  "recommendation": "Check whether the June promo is still displayed on the homepage."
}

📁 Repository structure

workflows/
  weekly-report.json        # n8n workflow export (validated, ready to import)
templates/
  report-template.html      # e-mail HTML template with {{placeholders}}
mock-data/
  sales-example.csv         # 10 weeks of sample sales data
docs/
  setup.en.md / setup.md    # deployment guide (EN / PL)
  demo-script.en.md / demo-script.md   # 10-min client demo script (EN / PL)
  sample-report.html        # rendered sample report (live on GitHub Pages)
.env.example                # required environment variables

💸 Running costs

Service	Cost
OpenAI (gpt-4o-mini, 1 call/week)	~$0.01/month — negligible
Firecrawl	free tier is enough
Resend	free tier: 100 emails/day
n8n	free self-hosted or cloud trial

Business case: replaces ~4 hours of manual work per month. At any agency rate, ROI is immediate.

🗺️ Roadmap (extensions, not in MVP)

• Real social media stats via Meta Insights API (requires Meta business verification)
• Official Google Business Profile API instead of scraping
• Branded PDF attachment instead of inline HTML
• Multi-tenant panel — one workflow serving many clients

📄 License

MIT — use it, fork it, deploy it for your clients.