Marketing deep dive

The power of the funnel is that the conversion between stages, not the raw counts, is what you steer. You rarely control impressions directly - you control the rate at which impressions become clicks (creative and targeting), clicks become leads (landing page), leads become wins (sales), and the size of each win (packaging and price). Own a rate, own a stage.

• Impressions - how many times the ad was served. Bought with budget, capped by audience.
• CTR - click-through rate. Creative and targeting own it.
• Lead-rate - landing-page or form conversion. Web and product own it.
• Win-rate - lead to closed deal. Sales and comp own it.
• ACV - average contract value. Packaging and pricing own it.

Because the funnel is a pure product, a percentage change in any single rate passes straight to revenue as the same percentage - the multiplicative pass-through from b2. Halve CTR and revenue halves, no matter how healthy every other stage is. That is what makes a funnel diagnosable in seconds: you never argue about the whole number, you find the one rate that moved and sum the percentage moves down the guilty path.

• Absolute counts hide the story - "we lost 400 leads" means nothing until you know the rate that caused it.
• Rates are comparable across time - a 2% CTR last month versus 1.4% this month is a clean, ownable signal.
• Deeper stages warn earlier - impressions and CTR move this week; won revenue confirms it weeks later.

There are two legitimate top-lines and teams fight because they conflate them. Marketing revenue is the volume top-line - the bottom of the funnel. ROAS (return on ad spend = revenue / spend) is the efficiency top-line - the same revenue divided by what it cost. You can grow revenue while ROAS falls (you bought volume at a worse price), or shrink revenue while ROAS rises (you cut the wasteful spend). Name which one you mean before you celebrate or panic.

• Revenue up, ROAS down - you scaled by spending into worse audiences. Fine if the payback still clears.
• Revenue down, ROAS up - you trimmed the losing channels. Healthier than it looks on the volume slide.
• Both down - a real leak in the funnel, not a budget choice. That is the case the tree diagnoses.

• CTR falls - the ad is served but nobody clicks. Almost always creative fatigue (the audience has seen it too many times) or a targeting drift. Stage: Impressions to Clicks.
• Lead-rate falls - clicks arrive but do not convert on the landing page. Usually targeting (wrong audience clicking) or a broken or slow page. Stage: Clicks to Leads.
• Win-rate falls - leads pile up but sales cannot close them. Often a sales, comp, or pricing change, or lead quality quietly dropping. Stage: Leads to Wins.
• ACV falls - deals close but smaller. Discounting, downmarket mix shift, or a packaging change. Stage: Wins to Revenue.

The store's orders carry a channel (web or app) - a rough stand-in for "which surface is converting". When the funnel points at a stage, the next question is always which slice. Run this to split completed revenue by channel.

SELECT o.channel,
       COUNT(DISTINCT o.order_id)                  AS orders,
       ROUND(SUM(oi.quantity * oi.unit_price), 0)  AS revenue
FROM orders o
JOIN order_items oi ON o.order_id = oi.order_id
WHERE o.status = 'completed'
GROUP BY o.channel
ORDER BY revenue DESC;

In the tree above, edit CTR from 2% down to 1%. Watch Revenue and the delta chip. Because the funnel is a pure product, revenue should fall by almost exactly 50% - creative fatigue at the top of the funnel halves everything downstream, no matter how healthy the other four stages are. That is the pass-through, felt.

Build a channel performance proxy: for each channel, how many distinct customers ordered, how many orders, and the average revenue per order. This is the closest the store data comes to "which surface converts best".

SELECT o.channel,
       COUNT(DISTINCT o.customer_id)               AS customers,
       COUNT(DISTINCT o.order_id)                  AS orders,
       ROUND(SUM(oi.quantity * oi.unit_price) * 1.0
             / COUNT(DISTINCT o.order_id), 2)      AS rev_per_order
FROM orders o
JOIN order_items oi ON o.order_id = oi.order_id
WHERE o.status = 'completed'
GROUP BY o.channel
ORDER BY customers DESC;

Imagine two campaigns that both produce exactly $192K. Funnel A leaks at the top: a weak CTR, but everyone who does click converts and closes cleanly. Funnel B leaks at the bottom: a strong CTR and lead-rate, but win-rate is poor and deals close small. Same bottom line, very different health.

Write a paragraph: which is healthier, and why? Consider that a top-of-funnel leak is often the cheapest to fix (new creative, better targeting) and that top-of-funnel volume compounds - a better CTR lifts every stage below it. A bottom-of-funnel leak means you are paying to attract interest you then fail to close, which burns spend and demoralises sales. Argue it either way, but ground it in where the leak is cheapest to fix and what compounds.