Prompt templates to analyze email performance
These prompts are based on the What Happens When You Feed AI 21,000 Emails and Ask What to Do Next webinar, published on the AI Lab by ActiveCampaign.
Get the prompt template
How to use these prompts
These prompts are pulled from the workflow Evan and Mason from Scrappy ABM walked through live in the webinar.
Use with: ActiveCampaign Active Intelligence—applies to every prompt below unless noted. Copy, paste, and swap in details where you see [BRACKETS]. Every AI model behaves a little differently, so review the first output and refine from there.
Prompt 1: Ask the AI what it needs before you analyze
Best for: Starting any analysis on the right foot so you get specific feedback instead of a vague summary.
Why this prompt: Evan opened with this when he was preparing his next webinar and wanted real analysis rather than a generic readout. Asking what the AI needs first means it tells you exactly what data to provide, instead of guessing.
Primary prompt
What do you need from me in order to assess my latest [EVENT TYPE]? My goal is to [GOAL].
Variables to fill in:
- [EVENT TYPE]—the campaign or event you want analyzed, for example “webinar” or “April newsletter.”
- [GOAL]—what you’re trying to learn or improve, for example “understand why open rates dipped” or “find content worth reusing.”
What to expect: A short list of the data it needs from you, such as the relevant tag, the campaign name, and the metrics you care about. Provide those, then move to the analysis.
Follow-up prompt:
Here is the tag for [CAMPAIGN]: [TAG NAME]. We sent emails and promoted on LinkedIn.
Based on this, what are the two or three most important things I should focus on for the next one?
Prompt 2: Improve a subject line for a specific campaign
Best for: Sharpening the subject line and content of a campaign tied to a known tag or goal.
Why this prompt: Mason described reaching for this when working from a specific event tag with a clear goal in mind. Naming the tag and the goal up front is what separates a useful answer from a generic one.
Primary prompt
Here’s the tag for our [EVENT] workshop: [TAG NAME]. Here’s my goal: [GOAL].
Can you help me figure out a better subject line and suggest how to optimize the content?
What are your thoughts?
Variables to fill in:
- [EVENT] and [TAG NAME]—the event and the exact tag that identifies it in your account.
- [GOAL]—the result you want, for example “lift the open rate above our 42% average.”
What to expect: Several subject line options and content suggestions tied to that campaign. Treat them as a starting set, not a final answer.
Follow-up prompt
Give me three more options that lean into [ANGLE], and tell me which one you’d test first and why.
Prompt 3: Find your best-performing content on a topic
Best for: Reusing what already worked when you’re building a new sequence on a familiar theme.
Why this prompt: Evan ran this while building an email sequence on ICP development. He wanted to see every email that referenced the topic, sort the winners from the rest, and carry the strongest into the new send.
Primary prompt
How many emails have we sent that reference [TOPIC]?
List them, and show me which performed well and which didn’t.
Variables to fill in:
- [TOPIC]—the theme you’re building around, for example “ICP development” or “post-event follow-up.”
What to expect: A list of past emails on that topic with performance flagged, so you can see what to reuse and what to leave behind.
Follow-up prompt:
From the ones that performed well, what do they have in common?
Use that to draft a subject line and opening for a new email on [TOPIC].
Tips for better results
- Ask, don’t command. Evan’s rule is to ask a question rather than issue an instruction: “Don’t say I need this. Ask a question.” A question invites the AI to tell you what it needs and returns more useful detail.
- Give it context. Name the tag, the campaign, and the goal every time. A little context goes a long way, and the more you supply, the better the response.
- Don’t go off the first answer. Give it a couple of iterations. The first output is a draft you refine, not a finished recommendation.
- Set guardrails. Tell the AI what to focus on so it doesn’t wander, the way you’d direct a capable new intern who has access to everything.
Related
More data from the AI Lab.