Prophet Forecasting: Predict Sales & Trends with Confidence — Lead Magnet

Here is a free lead magnet outline designed to build authority and provide immediate, actionable value.

**Title Suggestion:** **The Prophet Forecasting Quick-Start Guide: 7 Steps to Predict Sales & Trends**

**Format:** Standalone PDF Checklist / Reference Card (2-3 pages)

**Value Proposition:** This guide gives the reader the exact “plumbing” they need to get Prophet running. It solves the frustration of installation errors and data formatting issues, but stops short of teaching advanced tuning (seasonality, changepoints, validation), which is covered in the full skill.

—

### Lead Magnet Outline: 7-Step Checklist

**Step 1: Install & Import the Right Tools**
– [ ] **Install:** Run `pip install prophet` in your terminal or Jupyter notebook.
– [ ] **Import:** Add `from prophet import Prophet` and `import pandas as pd` to your script.
– [ ] **Verify:** Run a test import to confirm no dependency errors (e.g., `print(Prophet.__version__)`).

**Step 2: Prepare Your Data (The “Secret Sauce”)**
– [ ] **Two Columns Only:** Your DataFrame must have exactly two columns: `ds` (date) and `y` (value).
– [ ] **Date Format:** Ensure `ds` is a datetime object (e.g., `pd.to_datetime(df[‘ds'])`). Prophet hates strings!
– [ ] **No Missing Dates:** Check for gaps (e.g., `df[‘ds'].diff().value_counts()`). Fill missing dates with `NaN` or `0`.

**Step 3: Create & Fit Your First Model (The “Default” Run)**
– [ ] **Instantiate:** `model = Prophet()` (using default settings for now).
– [ ] **Fit:** `model.fit(df)` – This trains the model on your historical data.
– [ ] **Watch for Warnings:** If you see “Optimization terminated successfully,” you're good. Ignore “Discrete seasonality” warnings for now.

**Step 4: Build a Future Horizon (The “Crystal Ball”)**
– [ ] **Create Future DataFrame:** `future = model.make_future_dataframe(periods=30)` (predicts 30 days ahead).
– [ ] **Check Periods:** Change `periods` to match your business cycle (e.g., 90 for quarterly, 365 for yearly).
– [ ] **Note:** This only extends the date index; it doesn't predict yet.

**Step 5: Generate Your Forecast**
– [ ] **Predict:** `forecast = model.predict(future)`.
– [ ] **Extract Key Columns:** Focus on `ds`, `yhat` (predicted value), `yhat_lower`, and `yhat_upper` (confidence intervals).
– [ ] **Quick Check:** Print the last 5 rows of `forecast[[‘ds', ‘yhat', ‘yhat_lower', ‘yhat_upper']]`.

**Step 6: Visualize the Trend (The “Aha!” Moment)**
– [ ] **Plot Forecast:** `fig1 = model.plot(forecast)` – Shows historical data + future prediction.
– [ ] **Plot Components:** `fig2 = model.plot_components(forecast)` – This is your goldmine. It splits the forecast into **Trend**, **Weekly**, and **Yearly** components.
– [ ] **Interpret:** Look for strong weekly patterns (e.g., weekends dip) or a rising/falling trend line.

**Step 7: Spot Check for Changepoints (The “Hidden Insight”)**
– [ ] **Find Changepoints:** Run `model.changepoints` to see where the trend changed direction.
– [ ] **Visualize:** Add `changepoint_prior_scale` to your model (e.g., `Prophet(changepoint_prior_scale=0.5)`) to see how flexible the trend is.
– [ ] **Action:** If you see a sharp drop during a known event (e.g., a product launch failure), you've found a changepoint.

—

### Call to Action (CTA) – The “Full Skill” Offer

**Headline:** **Ready to Master Prophet? Go Beyond the Basics.**

**Copy:**
“You've just built your first forecast. But to get *accurate*, business-ready predictions, you need to tune the model. In the **full Prophet Forecasting Skill**, you'll learn how to:

– **Add custom holidays** to account for Black Friday, promotions, or closures.
– **Detect and fix bad changepoints** that ruin your trend line.
– **Cross-validate your model** to measure error (MAPE, RMSE) and avoid overfitting.
– **Apply Prophet to a real