What this answers

The funnel report asks where do people drop off. This one asks what did we pay to get them, and did it pay back. Same fictional home-services company, viewed from the media side: one row per day per campaign, with impressions and clicks in front of the funnel and won revenue behind it.

Metrics

• Spend — ad spend. Avg / Day divides by the days in range.
• Impressions — ads served. CPM = spend ÷ impressions × 1,000.
• Clicks — ad clicks. CPC = spend ÷ clicks. CTR = clicks ÷ impressions.
• Leads — quote requests. CPL = spend ÷ leads. Lead CVR = leads ÷ clicks.
• Consults — in-home consultations booked. Cost / Consult = spend ÷ consults.
• Jobs Won — signed contracts. CAC = spend ÷ jobs won.
• Won Revenue — contract value of those jobs. Avg Job = revenue ÷ jobs.
• ROAS — won revenue ÷ spend.

Every ratio is computed on the totals of whatever is in view — never as an average of daily ratios, which would let a thin Sunday count as much as a heavy Tuesday. A ratio shows — when its denominator is zero.

Paid and non-paid together

Google Ads and Meta Ads carry spend, impressions and clicks. Organic Search, Direct and Email have no media cost, so their spend and impressions are zero and every cost ratio (CPM, CPC, CPL, Cost/Consult, CAC) plus CTR and ROAS reads — for them. Their "Clicks" column is sessions rather than paid clicks, which keeps Lead CVR and everything downstream comparable. The Paid vs. rest of marketing section shows what each side actually contributes.

Attribution

Leads, consults, jobs and revenue are first-touch — credited to the campaign that first brought the person in, and recorded on that arrival date. That makes channel comparisons fair: spend and outcome sit in the same row for the same campaign.

One thing to know about the demo data: every step is booked on the arrival date itself, so there is no settling lag here — the last few days are complete, just short (the range ends mid-week). Against a real warehouse the same query would lag, because a job signed in August is credited back to a June click, and recent weeks would keep filling in for months. Read this dashboard for shape and mechanics, not for how fresh real data behaves.

Controls

• Date range — applies to every card, table and chart. Presets are anchored to the last day of demo data, not to today.
• Channel and Campaign Type — multi-select; click a chip to toggle it. Everything except the Paid vs. rest section respects both.
• Day / Week / Month — re-buckets the table and each trend chart independently.
• Metric chips on each chart toggle individual series on and off.

What the data is built to show

• Mid-June — paid social scale-up. Meta spend nearly triples. CPM climbs as the audience thins, lead quality drops sharply, and Meta's CAC roughly doubles while its ROAS halves. Cheap leads, expensive customers.
• Early July — quote form redesign. Lead CVR steps up on both paid channels with no change in CPC, so CPL improves without buying anything differently.
• Across the quarter, search CPCs drift up ~12% from auction pressure.
• Search vs social CPM is an order of magnitude apart ($500+ vs ~$30) — high CPC × high CTR versus the opposite. Blending CPM across channels is meaningless, which is why it's always shown per channel here.