How AI Donor Follow-Up Actually Works: A Step-by-Step Walkthrough

The most common question we get from campaign managers isn't "does this work?" They've seen enough to believe it works. The question is: "what does it actually look like in practice?"

Fair question. The pitch version of AI fundraising sounds clean. The reality involves call notes, approval queues, voice training, and a few things that need to be set up correctly before the first run. None of it is complicated, but the gap between "AI drafts your follow-up emails" and "here's the step-by-step workflow" is worth closing.

Here's the full process, from end of call session to sent emails, with nothing glossed over.

Step 1: The Call Session

Nothing changes about how the candidate or finance director runs calls. Same dialing routine, same conversation, same call sheet.

The one thing that needs to happen: notes from each call need to get into the system. This is the only new behavior the team has to adopt, and it's the part most campaigns already do in some form.

The notes don't need to be polished. They need to capture:

What was discussed (beyond "called, left voicemail")
Any personal detail the prospect mentioned
Any commitment made, by either side
Whether it's a viable donor and at what level

The format that works best: a quick summary immediately after hanging up, before the next call. "John, mentioned wife is a retired teacher, interested in education funding. Hasn't given before. Seemed genuinely interested, suggested $500 as a starting point. Wants to hear from us about the summer event." That's enough.

Step 2: Triggering the Follow-Up Run

Call session ends. Finance director has notes on 20-40 conversations.

Instead of opening email and starting to compose, the finance director sends a trigger to the system. This looks like a message: a Telegram text, a Slack message, an email, depending on how the campaign's workflow is configured. The message is simple: "Run follow-ups for today's call list." Sometimes it's even simpler than that: just forwarding the day's notes file.

The system picks up the trigger and starts working. While the finance director is doing something else (driving to the next event, eating dinner, prepping tomorrow's call sheet) the drafts are being generated.

Time between trigger and drafts ready: typically 3-10 minutes for a standard call session.

Step 3: What Gets Generated

For each contact in the day's call notes, the system produces a draft follow-up email.

Each draft is:

In the candidate's voice. Not in generic fundraising language. Not in the system's default register. In the specific way your candidate writes when they're following up with someone they just talked to. This requires a voice model to be set up during configuration (more on that below).

Specific to the conversation. The draft references what was actually discussed. The prospect who mentioned his small business gets an email that acknowledges it. The donor who expressed concern about a specific issue gets an email that touches on it. The person who said "call me next week" gets an email confirming you will.

Complete and send-ready. Subject line, greeting, body, ask if appropriate, sign-off. Not a template with blanks. A finished draft.

With a note to the finance director. Each draft comes with a brief note: any flags the system noticed (this contact has given before and the email accounts for that; this prospect mentioned sensitivity to cost and the draft adjusted the ask amount; this one needs a personal review because the notes were ambiguous).

Step 4: The Review Process

The finance director receives the batch. This is where the human check happens: the step that doesn't go away and shouldn't.

For each draft, the finance director can:

Approve as-is
Edit before approving
Hold (flag for the candidate to review first)
Reject (discard, will follow up personally or not at all)

Most drafts in a well-configured system get approved as-is or with minor edits. The finance director is not rewriting emails; she's doing quality control on a batch that's already 90% right. The work shifts from "writing 40 emails" to "checking 40 emails." That's a fundamentally different task, and it's much faster.

Campaigns typically review a 25-contact batch in 20-30 minutes. Versus 3-4 hours to write them from scratch.

Step 5: Sending

Approved drafts send from the candidate's email account, with the candidate's signature. To the recipient, it looks exactly like any other email from the campaign. Because it is.

Responses route to whoever manages the candidate's email. Any reply that requires judgment (a donor asking for a call, someone making a commitment, someone who wants to be removed) gets handled by a human. The AI handled the outreach; a person handles the relationship from there.

What Needs to Be Set Up First

The walkthrough above assumes the system is configured. Getting there requires three things, usually done in the first week of an engagement:

Voice model. The system needs examples of how the candidate actually writes. Emails the candidate has sent themselves. Text messages if the style is appropriate. Speeches, if the campaign emails tend to reflect that register. The more examples, the better the calibration. A well-trained voice model is the difference between drafts that sound like the candidate and drafts that sound like a generic fundraising email.

Call note format. The system needs to understand the format your team uses for call notes, whether that's a shared doc, a CRM field, forwarded voice memos, or something else. This is usually a 30-minute configuration conversation.

Approval channel. Where does the finance director review drafts? Some campaigns prefer email. Some use Telegram. Some use a simple web interface. The workflow gets configured to match how the team actually operates, not the other way around.

What This Doesn't Do

Worth being explicit about the limits:

It doesn't write fundraising ask emails for cold lists. That's a different workflow (the bulk email program) with different parameters.

It doesn't make the ask for major donors. A $25,000 ask still requires a personal call or an in-person conversation. The AI handles the $500-$5,000 follow-up tier, which is exactly the tier that falls through the cracks on most campaigns.

It doesn't work without call notes. If the team doesn't capture notes, the system produces generic emails. Generic emails are better than no emails, but they're not what you're paying for.

And it doesn't eliminate the approval step. Every email that goes out to a donor is reviewed by a human before it sends. That's not a limitation; it's the design. Campaigns are legally and reputationally accountable for every communication they send. The human stays in the loop.

The campaigns that run this well aren't the ones who expected magic. They're the ones who set up the notes process, invested in the voice training, and made the approval review a daily habit. The technology does its part. The operation has to do its part too.

Eric Linder is a former California State Assemblyman (2012-2016) and founder of AutomatedTeams, an AI operations consultancy for political campaigns and advocacy organizations.