← Back to the results
The next campaign, played out
Short List
Three short emails over three weeks, to the people the last campaign already warmed up. The email is the signal. The destination does the selling.
Working title. Not a product name, and not for external use.
Who it goes to
Warm first. Cold in the same format, second.
The last campaign did not produce a demo, but it did produce a sorted audience. Start there, in three waves, smallest and warmest first.
People worth a call
Reached the landing page, or read it, or already known to Civis from a previous event. Security-scanner accounts are stripped out.
Repeat engagers
Opened or clicked across two or more separate messages. Repeat engagement over weeks is the signal email security software cannot fake.
Fresh cold audience
Nonprofits from the public-filings and hiring-signal pipeline, in the same short format, so the cold read is comparable to the warm one. Sized at build time.
What comes out before anyone is mailed
Every layer is applied when the list is built. The last campaign learned this the expensive way: 467 people who could not receive mail sat on the lists through five sends, and only one of them would have been caught by send-time suppression.
Mailability is checked against every candidate up front, not against whoever survives the other filters. Otherwise the people we drop have unknown reachability and we can never answer "who did we leave out, and could we even have reached them".
The emails
Three sentences, one link, one small ask.
Rendered as they would arrive. Two subject lines per touch, because the subject is now the only thing being tested. Everything else is identical between the arms.
After the open, the plain and branded lanes were indistinguishable on clicks, and the branded one won the Leadership persona and produced fewer opt-outs. There is no evidence a light wrapper suppresses response, and a real cost to sending unbranded mail from a company this audience cannot name.
Who fronts the sends. The named human sender is the variable most likely to have earned the open rate in the last campaign, and it puts replies in a specific inbox. Whoever it is, the reply address has to reach somebody who reads it.
How it plays out
Six weeks becomes three.
The last campaign spent six weeks per group on four long emails, and the three after the first bought no incremental engagement.
Scroll the timeline sideways to read it.
Scroll the timeline sideways to read it.
A click. One link to a two minute overview, no form in the way.
The same click, from a different angle and in new words. Never the same body twice.
A yes or a no, by reply. That is the whole ask, and it is the one we can count.
What triggers the next step
Day 22, one week after the last touch, opens have finished accruing and the result is read on clicks and replies. A yes goes straight to a person. The winning subject line carries into wave 2 unchanged, so waves 2 and 3 are not re-running the same test. One send time, one wave size, every wave: the last campaign's worst-performing send was its only morning send and its largest, and the same first email, unchanged, swung from 29.3% open to 16.8% depending on the day it went out.
How we would know
The instrumentation, fixed before send one.
Every gap below is one the last campaign is still carrying. None of them can be fixed after a send.
What the last campaign could say
- Opens and clicks per send, inflated by security software we identified in week six
- Landing page arrivals for the recipients we could identify, which was fewer than half the traffic
- Form opens that the page fired for people on arrival, whether they wanted the form or not
- Demos: zero, and that number is correct
- Replies: unknown, on both arms, for the whole campaign
What Short List would say
- Replies counted from send one, because the reply address is one somebody reads and labels
- Clicks per person, per touch, per subject arm, with per-person tracking live on the first email
- Human clicks and machine clicks reported as separate numbers from the first send, using the scanner shape we can now describe
- Page reading, because the page will not open its own form and manufacture intent
- Which subject line won, read on clicks and replies rather than on opens
The funnel we would instrument
Open rate stays on the report and stops being a success metric. Apple is 62% of all email opens and its privacy feature affects 55 to 60% of them, so an open is a signal about the subject line and nothing else.
In one line
Same audience, a tenth of the words, an ask a person can answer.
The last campaign proved we can reach these people and that they will not book a demo with a company they cannot name. Short List keeps the reach, drops the ask to something a stranger can complete alone, and counts the one signal we have never been able to count.