How to use Active Intelligence to analyze and rebuild a campaign
These prompts are based on the Your Database Isn’t Too Big. Your Segmentation Strategy Is Too Simple. webinar, published on the AI Lab by ActiveCampaign.
Get the quick-start guide
What will I accomplish with this guide? By the end, you’ll have a repeatable Active Intelligence workflow that pulls performance data from a past campaign, identifies what worked and what didn’t, and drafts the next version with subject line, copy, and CTA suggestions baked in. It’s based on the workflow Don Purdy demonstrated live in this webinar—the same prompt he runs before every information-session campaign so the next send is built on what the last one taught him.
Before you start, you’ll need:
- An ActiveCampaign account with Active Intelligence enabled
- At least one past campaign with enough send volume to read meaningful open and click rates
- Brand components (logo, color palette, default fonts) already loaded into ActiveCampaign so the rebuilt campaign comes back on-brand
- A specific upcoming campaign to plan—a recurring event, a launch, a re-engagement push—so the analysis has a real next step to feed
Quick reference
- Total time: 10–15 minutes per campaign analysis (down from a half-day spreadsheet exercise)
- Tools needed: ActiveCampaign Active Intelligence; your campaign history; brand components loaded for the campaign rebuild step
- Key output: A written read on the past campaign’s performance plus a drafted next campaign that incorporates the changes Active Intelligence recommends
Watch this section
For full context on the following topics, watch these sections of the webinar:
- The half-day-to-seconds shift Active Intelligence enables—[17:51]–[18:25]
- The live analysis: open rate, CTR, deliverability, unsubscribe—[18:55]–[20:14]
- Brand components feed the rebuilt campaign—[20:14]–[20:47]
- The continuous test → analyze → apply loop—[22:30]–[24:07]
The workflow
Phase 1: Pick the right past campaign to learn from
After this phase, you’ll have: a single campaign identified as the basis for analysis, with enough volume and a clear comparison point.
- Choose a recurring campaign: Don analyzes his last information-session campaign before building the next one. Pick something you’ll send again—a monthly event, a quarterly newsletter, a recurring nurture—so the lessons compound.
- Confirm the segment is consistent: if the upcoming send goes to the same segment as the past campaign, the analysis will translate. If you’re targeting a different group, note that—Active Intelligence can still benchmark, but the comparison loosens.
- Write down what you actually want to know: subject line performance, send-time effects, CTA clarity, copy length. Aim the prompt at the decision you have to make for the next send.
Phase 2: Run the analysis prompt in Active Intelligence
After this phase, you’ll have: a written read on the past campaign—open rate, click-through rate, unsubscribe rate, deliverability—plus a benchmark comparison and specific opportunities for improvement.
- Open Active Intelligence from the recommended-for-you area on your dashboard or wherever your account surfaces it.
- Submit the analysis prompt: see the companion prompt template for Don’s exact structure. The prompt asks Active Intelligence to analyze the past campaign across open rates, click-through rates, subject line, preheader, and CTA, and to surface what worked and what didn’t.
- Let it cross-reference your campaign history: Active Intelligence is reading your data live, not pulling a pre-built report. It also benchmarks against industry peers—Don’s account benchmarks against higher education.
- Read the strengths and gaps section: the output will flag what performed and where the campaign underperformed. This is the analysis you used to do in a spreadsheet.
Things like exporting the data, building the pivot tables, trying to find the story buried in all of the analysis in a spreadsheet — now it’s transformed to something that looks like a prompt and a few seconds.
Phase 3: Let Active Intelligence draft the next campaign
After this phase, you’ll have: a drafted next campaign with revised subject lines, optimized copy, and CTA suggestions, built using your loaded brand components.
- Ask Active Intelligence to build the next campaign: the same prompt that runs the analysis can also draft the next send incorporating the changes. In Don’s run, this happened in seconds after the analysis returned.
- Confirm brand components are applied: the draft pulls from the brand components you’ve loaded—logo, palette, fonts. If the draft looks off-brand, your brand components need an update.
- Review the recommended changes: typically an enhanced subject line, a clearer CTA, optimized copy, and visual improvements. Treat them as a starting draft, not a final.
- Adjust and approve: edit anything that doesn’t fit your voice, then schedule the send.
Phase 4: Close the loop after the campaign sends
After this phase, you’ll have: a documented insight from the campaign you just sent, ready to feed the next analysis.
- Run an A/B test on the new campaign: ActiveCampaign can pick the winner automatically. Don’s pattern: question vs. statement subject lines, urgency vs. specificity, personalization on or off.
- Go back to Active Intelligence after the send: ask why the winning subject line won, or what patterns are showing up across your top campaigns. The model has the new data point already.
- Document the pattern: specificity, urgency, personalization—whichever lever moved the result. Save it somewhere reusable so the next campaign starts from a sharper baseline, not a blank page.
- Apply the learning to the next campaign brief: when you run Phase 1 again next month, the past campaign you analyze is already the better version of itself.
After every campaign, we go back to Active Intelligence. Why did the subject line win? What patterns are showing up across my top campaigns? Wherever the insight, we document it and apply it to the next campaign. So segmentation gets more refined, messaging gets more targeted, and deliverability keeps climbing.
Related
More data from the AI Lab.