Marketing funnel by channel

What it is good for

• Compute CTR, signup rate, purchase rate and overall conversion for each channel.
• Test whether conversion rates differ between channels with a chi-square test on aggregated counts.
• Compute cost per click, per signup and per purchase and rank channels.
• Explain why averaging weekly rates differs from pooling counts.

Methods: proportions, chi-square test of independence, confidence interval for a proportion, funnel analysis, ratio metrics.

Data quality notes

• Aggregated counts: each row summarises many users. Analyse rates with the right denominators (clicks ÷ impressions, signups ÷ clicks, …).
• Email 'impressions' are delivered emails, not ad views, so its CTR is not directly comparable.
• Attribution is last-click, which favours channels close to the purchase.
• No missing values.

Provenance and licence

Impressions per channel-week = channel volume × weekly multiplier (gentle growth and a mid-period sale peak in weeks 10–11) × log-normal noise (sd 0.1). Clicks ~ Binomial(impressions, CTR) (normal approximation when n·p·(1−p) > 400), signups ~ Binomial(clicks, signup rate), purchases ~ Binomial(signups, purchase rate), with channel-specific true rates (exported as FUNNEL_TRUE) and small week-to-week rate variation. Spend = impressions ÷ 1000 × channel CPM.

Synthetic data · CC0-1.0 (synthetic data generated by Statistics Quest) · generated deterministically (seed 20241024). It does not describe real people or organisations; any resemblance is coincidental.