The Loadstar — Ad Funnel

House ads are our own inventory. Impressions track newsletter send volume; clicks are what actually reaches the microsite.

Months where GA4 was unavailable and no backfill source exists are shown as a gap ("no data"), never as zero. Backfilled months are marked °.

Stack total equals Total Inquiries exactly. Qualified Sales Inquiries (TF1) are included from 27 Sep 2026 — they were previously left out of the total, which understated April, May and June, the only months that route was reachable. June was 32 inquiries at 0.8%, not 22 at 0.5%.

Count, not quality. The Contact bucket is a general enquiry form on a general contact page, and most of what arrives is not an advertising lead. A full sample of June 2026 found 1 of 7 genuine: four were subscription support, two were spam. September's submissions look the same. Treat a rise in Contact volume — and any conversion rate computed from it — as a measurement signal, not a performance one, until the source-side form split separates intent at the point of submission.

Signups are audience growth, not inquiries — they are never counted toward the funnel's inquiry total.

This is the missing piece, and it is the one that matters most right now. Everything above this stage counts how many people raised a hand. This stage is the only place that asks how good they were — and we cannot currently answer it. That gap is why a rise in Contact-form volume tells us almost nothing: a full sample of June found 1 genuine advertising lead in 7, so counting inquiries without qualifying them can move in the wrong direction and still look like progress.

Read the bars below as indicative only. They come from a Lead Score typed by hand into the Loadstar Sales Funnel Tracker in Notion, not from anything the funnel measures. Four specific reasons not to lean on them:

· Unscored rows count as Low. A blank score reads as 0, which lands under 20 — so "Low" mixes genuinely poor leads with leads nobody got round to scoring. Those are opposite findings.
· The population does not reconcile with the stage above. April shows 29 scored leads against 22 total inquiries. A funnel cannot widen through a qualification stage, so these are counting different things.
· The month is the wrong month. Rows are attributed by when the Notion page was created, not when the lead made contact, so anything written up late lands in the wrong column.
· It has lapsed. Used March to August, then effectively stopped — September has one row. The tracker itself is healthy (68 rows, 60 scored); the process around it stopped.

Media Kits Sent reads 0 in every month for the same reason: it counts the Response Sent checkbox, which has never been ticked on any of the 68 rows. It appears in the all-metrics table below.

What would fix it: score leads from behaviour the funnel already sees — which form they used, which pages they read, whether they are at a named target account — rather than from a judgement typed in afterwards. Until then this stage is a question mark, and it is deliberately left visible as one.

Deal value is attributed to the month a deal is won, not the month the inquiry arrived — so revenue and inquiries in the same column are not a cohort.