Good morning.

Klaviyo recently published benchmark data from 183,000 of their customers.

One number that stood out to me:

Revenue per recipient on automated flows is nearly 18x higher than on campaigns.

Flows account for just 5.3% of email volume but generate nearly 41% of total email revenue.

Most operations spend the bulk of their email effort on the 94.7% that produces the least return.

The behavior-triggered sequences running quietly in the background are where almost half the money lives.

This issue is the playbook for shifting your email architecture from scheduled to responsive.

— Sam

Let’s get into it.

A fixed drip sequence is closer to a broadcast than a flow. It goes out on a schedule. It doesn't know who clicked and who didn't. It doesn't know who bought, who ghosted, or who's been opening every email for six weeks without ever hitting a link.

The core problem: a fixed sequence assumes everyone on your list is the same person at the same stage at the same time. 

The person who signed up from a YouTube video about advanced tactics is a different buyer than the person who found you from a beginner-level blog post. 

Putting them on the same seven-email sequence is lazy marketing dressed up as automation. 

Most teams built their drip once, maybe two years ago, and haven't touched it since.

The drop-off tells the story. Sequences that open at 40%+ on email one routinely land below 10% by email five or six. By the time your actual offer shows up (usually the last email in the sequence) you've already lost the majority of your engaged audience.

There's an invisible cost compounding underneath the drop-off. When subscribers consistently ignore your emails, your sender reputation takes a hit. You start landing in promotions tabs. Then spam. 

Marketo's research found that even top-performing email marketers estimate 51 to 75% of their list is inactive. 

You're paying every month to store and email a list that's majority dead weight, while burning your reputation with the subscribers who actually engage.

MailerLite's data adds a useful nuance: open rate drop-off only really occurs when you email daily. Between monthly and twice-weekly, rates stay fairly consistent. 

The problem isn't actually frequency but irrelevance at any frequency. A fixed sequence has no mechanism for relevance.

For years, the answer to "drip sequences are broken" was "build better branching logic." Map every path. If they click this, send that. If they don't open by day three, try a different subject line. 

The problem was that building and maintaining those decision trees required someone on your team to anticipate every scenario in advance, wire up hundreds of conditional rules, and debug the whole thing when a branch broke or a subscriber fell through a gap.

That's where LLMs and agentic workflows change the equation. AI can now handle three things that used to make behavioral email impractical for a team of 10 to 25. 

The rest of this issue covers the behavioral framework, the tools, and five prompts that let you rebuild the system using AI as the engine.

"Behavior-based email" gets thrown around as a buzzword. Here's what it actually is.

In a fixed sequence, the trigger is time. Email 1 goes out on day 0. Email 2 goes out on day 3. The clock runs regardless of what the subscriber does.

In a behavior-based system, the trigger is action. The subscriber's behavior determines what they get next. Opened email 1 but didn't click? Branch A. Clicked the link but didn't buy? Branch B. Bought? Remove from sequence, move to onboarding. Went quiet for 14 days? Fire a re-engagement email, not email 4.

This has been technically possible for years inside platforms like ActiveCampaign and Klaviyo. 

What's new is that AI is now handling the logic that used to require someone on your team to map out hundreds of branching rules manually.

The old behavior-based model was still brittle. You'd build decision trees (if/then logic) and map out every possible path. 

But you had to anticipate every scenario in advance. Miss a branch and the automation breaks or sends something irrelevant.

The new model is different. Instead of pre-mapping every path, AI systems watch real-time behavioral data (opens, clicks, site visits, purchases, even predictive signals like churn probability) and make a live decision about the next best email to send to that specific subscriber.

ActiveCampaign describes this shift directly: autonomous marketing doesn't follow a fixed script. It reads the room and determines the best messaging, timing, channel, and follow-up on its own. 

Less flowchart, more trained sales rep who knows this particular customer's history.

If you're going to move from a fixed sequence to a behavior-driven system, these are the signals that actually matter.

1. Engagement signals. Opens and clicks are the baseline. More importantly: which links they clicked, how often they engage, and whether engagement is trending up or down. A subscriber who opened 10 of your last 12 emails is in a completely different bucket than one who opened 2 of the last 12.

2. Purchase and conversion signals. Did they buy? What did they buy? Did they abandon a checkout page? A subscriber who hit your sales page three times and didn't convert needs a different email than one who never visited. They need objection handling, not a product introduction.

3. Recency signals. How long since they last engaged? 7 days is different from 60. AI systems use recency as a decay factor. The longer the silence, the more targeted the outreach needs to be before they're permanently tuned out.

4. Predictive signals. This is where it gets interesting. Klaviyo now offers predictive CLV scoring and churn risk scoring built into the platform. Before a subscriber disengages, the model can flag them, triggering a re-engagement campaign before you lose them rather than after.

5. Zero-party data signals. What subscribers have explicitly told you about themselves. Their goals, their stage, what they're working on, what they're struggling with. This is the most underused signal in most email programs. More on this shortly.

Here's how to actually build this. Two versions based on where your operation is right now.

ActiveCampaign at the Plus or Professional tier handles behavioral triggers, list segmentation, and predictive sending. Across their entire customer base in 2025, the platform averaged a 40.4% open rate and 6.7% click rate, well above industry baseline. 

The automation logic is solid, and behavior branching handles the complexity most operations at this level need. If your team built a drip sequence two years ago and hasn't touched it since, this is where you rebuild.

GoHighLevel if you're an agency or managing multiple client accounts. It combines CRM, email, SMS, funnels, and automation in one platform. HighLevel's own November 2025 performance data showed a 4.08% click-through rate across their user base, above the 2 to 3% industry benchmark, with behavior automation cited as the primary driver.

Fathom or Fireflies for sales calls. Auto-transcribes and surfaces what customers actually said. Feed that language into your email copy and your team's understanding of objection patterns.

A simple entry quiz or survey (Typeform, Jotform, or even a plain-text welcome email that asks one question) to start collecting zero-party data from day one.

Klaviyo if you're in ecommerce or have product purchase data. Their 2026 benchmark data confirms what the platform promises: flows generate nearly 18x the revenue per recipient of campaigns. The predictive CLV and churn risk models run in the background without manual setup.

ActiveCampaign with Active Intelligence for the autonomous AI agents that suggest next steps, create automations, and validate results. This is the layer where the system starts making decisions your marketing person used to make manually.

Make.com to wire everything together and create custom trigger logic that native integrations don't support. Useful when you're connecting your email engine to your CRM, your support system, or your agent infrastructure.

Whatever tools you use, the fundamental architecture change is the same.

You don't need a 40-automation system to make this shift. Here's a stripped-down version your marketing person can implement in a weekend, or that you can build yourself if you're still hands-on with email.

Map four behavioral moments, not a linear sequence.

The rule: every email in your system should have a reason to exist based on what the subscriber just did or didn't do. If you can't articulate the behavioral trigger, cut the email.

This separates the operations whose AI email systems actually work from the ones who set up the triggers and still get mediocre results: 

The quality of data going into the system.

Behavioral data (opens, clicks, purchases) tells you what someone did. Zero-party data tells you what someone wants. 

The combination is what lets you send emails that feel like they were written for that specific person, because they effectively were.

The best operators do this at three points.

Zero-party data compounds. The longer you collect it, the more precisely you can segment, the more relevant your emails become, the better your deliverability gets, and the higher your conversion rates climb. A drip sequence decays over time. A well-built zero-party data system gets more effective the longer you run it.

The framework above tells you what to build. These five prompts help you build it. The first one runs standalone to gather audience intelligence. 

The remaining four run in order inside a single Claude or ChatGPT project, where each builds on the output of the last.

Start with Prompt 1 (Audience Research) in a separate conversation. Save that output. Then create a project in Claude or ChatGPT and upload your actual messaging as project files: your landing page copy, your current email sequence (every email with subject lines and send timing), your sales page, your lead magnet if you have one, and the audience research output from Prompt 1. 

Then add this as your project instruction:

The prompts won't produce anything useful without this context. The AI needs to see what your subscribers saw before they hit your list, because your landing page sets the promise and your emails have to deliver on it.

Run this first, in a separate conversation, using an AI tool with web browsing enabled (ChatGPT with browsing, Claude with web search, or Perplexity). It mines real conversations your audience is already having online and pulls out the exact language, frustrations, and objections you need. Save the output and upload it to your project. 

Without this, the remaining prompts fall back on generic marketing language. With it, every email the AI writes uses your audience's actual words.

This prompt diagnoses your current sequence against behavioral criteria. It doesn't rewrite anything. It tells you what's broken, what's working, and what behavioral moments you're missing entirely.

This is the core build. It takes the four moments framework from this issue and turns it into actual emails with entry triggers, exit conditions, and suppression rules tailored to your business.

This prompt takes everything the previous prompts produced and organizes it into a build document. Not a narrative. A blueprint your team can implement from.

This is a standalone prompt for the Ghost module (Moment 4). It writes the single email that determines whether a silent subscriber comes back or gets scrubbed. Run it after the system is built, or run it now if your re-engagement emails are the generic "we miss you" type that everyone ignores.

That’s it for this issue.

The drip sequence had a good run.

It was a real step up from blasting your list once a month and hoping something landed. It introduced the idea that nurture was a system, not a one-off event. For its era, it was the right answer.

But it was built for a world where treating every subscriber differently wasn't feasible.

Fixed sequences ran because that was the best option available. The tools exist today, at price points accessible to any operation with revenue, to build email systems that actually pay attention.

The irony of smarter email is that it means less email. Fewer sends to the right people at the right moment instead of more sends to everyone all the time.

Pick one thing this week. Find one time-based trigger in your sequences and replace it with a behavioral one. Have your team audit the welcome sequence against the four moments framework. Run the audit prompt. See what the data tells you.

Until next time,
Sam Woods
The Editor

.