Exercise: Funnel Drop-Off Analysis

The marketing funnel dataset is the most complex of the three — seven stages from impressions to closed-won deals, five channels, three months, plus spend data for ROI calculation. The analytical challenge is multi-dimensional: you need to show absolute volume, conversion rates between stages, channel comparisons, and cost efficiency — all without overwhelming the viewer. This is where chart selection and prompt engineering matter most.

Here is marketing funnel data across five channels and three months.

Before visualizing, analyze:
1. What is the conversion rate from each stage to the next, by channel?
2. Where is the biggest drop-off in the funnel for each channel?
3. Which channel has the best cost per closed-won deal?
4. Is there a channel where top-of-funnel volume is misleading — high impressions but poor conversion?

Then create an interactive dashboard Artifact with the charts. I need to present this to a CMO who cares about: where to invest more budget, where the funnel is leaking, and which channel produces the cheapest customers.

[Paste marketing-funnel.csv here]

The data contains a deliberate pattern: Webinars have the lowest impressions but the highest conversion rates at every stage. LinkedIn Ads have massive impressions but the funnel leaks badly between clicks and leads. Google Search has the highest absolute closed-won count. Email is the cheapest per acquisition. Content/SEO has the lowest cost but the slowest volume. Each channel tells a different story depending on which metric you emphasize — and that is exactly the trap. The CMO needs all five stories in one coherent narrative.

Upload marketing-funnel.csv to ChatGPT and ask for a Python script:

Prompt 1: "Write a Python script that creates a funnel chart for each channel showing the drop-off from Impressions to Closed Won. Use a horizontal funnel layout with stages as rows and channel volumes as bars. Aggregate across all three months."

Prompt 2: "Add a chart showing cost per closed-won deal by channel. Calculate it as total spend divided by total closed-won deals across all months. Sort from cheapest to most expensive. Add dollar labels."

Prompt 3: "Add a conversion rate heatmap — channels as rows, stage-to-stage conversions as columns (Click Rate, Lead Rate, MQL Rate, SQL Rate, Opp Rate, Close Rate). Color scale from red (worst) to green (best). Include percentage values."

Prompt 4: "Add a scatter plot with cost per acquisition on the x-axis and total closed-won deals on the y-axis. Each point is a channel. Size the points by total spend. Label each point with the channel name. Add quadrant lines at the median of each axis. Combine all four charts into a single figure and save as PNG."