If your team is under an AI mandate and you are not ready to hand the whole campaign over to a recommendation engine on day one, agentic secondary CTAs are the lowest-risk first play you can ship. The idea is simple: leave your campaign’s primary call to action exactly as it is, and add a second CTA block to the email template that an agent writes dynamically for each individual person. You keep the proven offer. You bolt on a personalized layer that an LLM composes on the way out. And you can build the whole thing in Marketo today.

This is the move for the marketer who hears “use AI” from the top and wants a concrete, shippable answer that does not bet the campaign. Edward Unthank walks through it on a glassboard in a few minutes, so let’s break down the anatomy, the infrastructure, and why it earns a seat as your team’s first agentic capability.

What an Agentic Secondary CTA Actually Is

Picture a standard marketing email. There is a headline, a picture, a paragraph or two of copy, maybe three bullet points, a “really appreciate it,” and then one primary call to action: click the link, take the action, do the thing the campaign was built to drive. That primary CTA is the campaign. It does not change.

An agentic secondary CTA sits underneath that primary offer as part of the template every send uses. Instead of being hand-written once and frozen, it is composed dynamically per recipient by a large language model. You are not replacing the recommendation logic of the campaign. You are adding a personalized second ask that the agent calculates for the individual who is about to receive the email.

The distinction matters because it lets you adopt agentic AI without going all in. The campaign’s primary CTA stays human-controlled and predictable. The secondary CTA becomes your sandbox for per-person personalization at scale.

The Anatomy of the Secondary CTA Block

When you picture what the agent actually produces, it is a small, self-contained block with four parts. Compile them together and you get a fully embeddable, fully personalized secondary call to action for one specific person.

• Super headline: the attention line that opens the block, written for this individual.
• Body paragraph: the message content, roughly one paragraph (Edward notes you might run two, but one is the working default).
• CTA button: the button label the agent generates to match the offer.
• CTA link: the destination URL the agent calculates for that person.

Each of those four elements is computed for the individual recipient. The super headline, the body, the button, and the link are all populated by the model, then assembled into one block the template drops into the email. That is the entire payload: four fields, dynamically written, embeddable, personalized.

How You Build It in Marketo Today

The Marketo side of this is deliberately boring, and that is the point. You build the secondary CTA into the overall template you already use, then have an agent on the outside write to person fields or program fields so the block populates per individual.

Here is the infrastructure shape Edward describes:

• Template: add the secondary CTA block to the email template in Marketo, with tokens that pull from the four fields (super headline, body, button, link).
• Fields: map each part of the block to person fields or program fields so Marketo has somewhere to read the dynamic values from.
• The agent on the outside: an orchestration layer (LangChain or n8n) runs the agentic approach that generates the four values and writes them back into Marketo’s person or program fields.
• Compile and send: at send time, the template assembles the four fields into the embeddable block, and each person gets their own calculated secondary call to action.

That is the whole loop. The model writes the headline, body, button, and link; the agent writes those to fields; Marketo renders the block. Nothing about the primary campaign has to change to make this work.

Why This Is the Right First AI Play

The reason agentic secondary CTAs work as a starting point is risk. If you are not going full all-in immediately on letting an AI recommendation engine drive whole campaigns, this gives you a real agentic capability that you can stand up basically right away, without touching what already converts.

For a team operating under an AI mandate and trying to figure out where AI is most beneficial, that combination is rare: it is genuinely agentic, it is personalized per individual, and it is immediate. You are not waiting on a six-month platform migration or a wholesale rebuild of your campaign logic. You add a block, wire an orchestration layer, and ship.

• Low risk: the primary CTA, the part that drives the campaign, is untouched and human-controlled.
• Immediate: the Marketo build is a template change plus field mapping, not a re-architecture.
• Genuinely agentic: an LLM composes net-new copy and destinations per person, which is exactly the kind of capability an AI mandate is asking for.

This is the kind of fun, concrete experiment Edward likes to come up with: small enough to try this quarter, real enough to count as progress on your AI roadmap. Building this kind of agentic layer into Marketo is exactly the work Etumos does, from agentic marketing operations through to agentic revenue operations.

Where It Goes From Here

Once a secondary CTA block is live and the orchestration layer is reliably writing to fields, you have proven the pattern: an agent on the outside, a template on the inside, and a clean handoff through Marketo fields. From there you can widen the agent’s remit, feed it more signal about each person, or eventually let the recommendation engine reach into the primary CTA once you trust it. But you start small, on the secondary layer, where a miss costs you nothing and a win compounds. If you want help standing up that first agentic capability, the team at Etumos agentic operations can build it with you. Let’s talk.

Frequently Asked Questions

What are agentic secondary CTAs?

Agentic secondary CTAs are a second call-to-action block added to your email template that a large language model writes dynamically for each individual recipient. The campaign’s primary CTA stays the same; the secondary block (super headline, body paragraph, CTA button, and CTA link) is composed per person and compiled into a fully embeddable, personalized unit.

How do you build agentic secondary CTAs in Marketo?

You add the secondary CTA block to your existing Marketo email template, map its four parts to person fields or program fields, and run an agent on an orchestration layer (LangChain or n8n) that generates the values and writes them back to those fields. At send time, Marketo assembles the fields into the block, so each recipient gets a personalized secondary CTA.

Why are agentic secondary CTAs a low-risk first AI play?

Because they leave your proven primary call to action untouched. You are not handing whole campaigns to a recommendation engine. You add a personalized second ask on the secondary layer, where a miss costs you almost nothing, which makes it ideal for teams under an AI mandate that want a real, shippable agentic capability immediately.

What is the anatomy of an agentic secondary CTA block?

Four parts: a super headline, a short body paragraph (about one paragraph), a CTA button, and a CTA link. An LLM writes all four for the individual, an agent writes them to Marketo person or program fields, and the template compiles them into one embeddable, fully personalized block.