Pulse Notes #1
What 264 first messages to money creators taught us
Data cut: Oct 12, 2026, 12:00 UTC
Since late September we have been sending first messages to people who teach money online: creators, coaches who already sell, and tax preparers. We kept a ledger. Here is what it says, small-sample caveats included.
The numbers (all first contacts up to Oct 12)
| Channel | First messages | Human replies | Rate (95% interval) |
|---|---|---|---|
| 177 | 4 | 2.3% (0.9% to 5.7%) | |
| 32 | 1 | 3.1% (0.6% to 15.7%) | |
| 55 | 0 | 0.0% (0.0% to 6.5%) | |
| Total first contacts across channels | 264 | 5 | — |
Intervals are 95% Wilson confidence intervals: they show uncertainty in the observed rates and do not establish differences between channels. "Human reply" means a person we first contacted answered, read by hand; we did not separate replies that came after follow-ups. We classified 15 replies: 5 human, 8 bots or auto-responders, and 2 false positives. Two more email replies are still unclassified. Counting all 15 classified replies as human would have tripled our human-reply count.
Our ledger contains no verified paid customers as of Oct 12. Payment verification is incomplete, so the real total is unknown and we make no claim either way.
What we learned
1. Check who answered before you celebrate. More than half of the classified replies were bots or auto-responders. Classify replies by hand before you report a rate.
2. One campaign looks better, and we still cannot say why. One email batch got 3 human replies from 37 sends (8.1%). Two other batches got 1 from 55 (1.8%, interval 0.3% to 9.6%) and 0 from 77 (interval 0.0% to 4.8%). The first batch's interval is 2.8% to 21.3%, so they overlap. These small, non-randomized batches do not establish a winner. It could be the copy, the week, or the list. With numbers this small, "no winner yet" is the honest answer.
3. Instagram produced no human replies, and 6 of 55 messages were never delivered. Check delivery before blaming the channel: a silent failure looks like being ignored.
4. Bounces matter early. 4 of 177 emails bounced (2.3%), just above the 2% brake in our own internal rules. Verify addresses before sending.
5. Count people, not messages. We sent 157 follow-ups on top of the 177 first emails. If you mix them into one number, you will think you contacted more people than you did. Our 264 is first contacts per channel, so deduplicate people across channels before you report a unique-person total.
A useful thing to steal: the 5-line outreach ledger
Keep one row per person per channel, with these five fields. It is the reason this issue exists.
- Who and which channel (public handle or business name only)
- Date of first message and which version of the copy
- Delivered? (yes / bounced / not delivered / unknown)
- Reply type: human / auto-reply / false positive / unclassified / none
- Outcome you can verify (a booked call, a payment ID) or "unknown", plus the date of the first human reply, if any
Then divide human replies by people contacted, per channel, and write the interval next to it. Report the sample size next to the rate. A small sample or a wide interval limits what you can conclude, so say so and keep going.
Same idea for your own audience
Your audience also tells you what they need, in their own words, and you probably have not counted it. The free Audience Money Check takes about two minutes. It turns the questions your followers repeat into a score and a one-page plan. It shows its math, and it is a planning tool, not an income promise.
Next Tuesday: what tax preparers can do in the off-season.
Reply to any Pulse Notes email with one thing you would like us to measure. We read every reply.
Pulse Notes is a weekly note from Pulse. Unsubscribe anytime. No financial advice; nothing here promises results.