Work out which channel really drove your signups when every tool claims credit

You are a skeptical marketing analyst. Below are signup numbers by week, what each channel's own dashboard claims, and any self-reported "how did you hear about us" answers. The claims add up to more signups than we got. Work out what each channel actually earned.

1. THE OVERCOUNT - table: Channel | Signups claimed | Attribution model it uses | Lookback window | Likely double-counted with.
2. THREE VIEWS - per channel: last-touch share, self-reported share, and what happened to total signups in weeks the channel was paused or scaled. Show the arithmetic.
3. BEST ESTIMATE - a range per channel (low to high), not a single number, with the view you trusted most and why.
4. WHAT WOULD SETTLE IT - the cheapest test to narrow the widest range: a holdout, a pause, a code, or a survey change.
5. BUDGET READ - one paragraph: which channel is most likely over-credited and what that means for next month.

Rules: no preamble. Do not invent numbers or weeks not in the data. Say "not enough data" rather than force a range.

WEEKLY SIGNUPS AND SPEND BY CHANNEL: {{paste}}
EACH TOOL'S CLAIMED CONVERSIONS AND SETTINGS: {{paste}}
SELF-REPORTED SOURCE ANSWERS: {{paste}}