How to build an ActiveCampaign automation in 2026

Write the job in one sentence before you open the builder: "when somebody requests the guide, send it, then follow up differently if they show buying intent." An automation you cannot describe in a sentence is a strategy problem, and ActiveCampaign's canvas will happily let you build it anyway.

This builds that nurture workflow end to end, and flags the three things that go wrong.

One prerequisite: this needs Plus. Starter caps an automation at five actions, and the workflow below is seven before the branch. Our plan comparison covers why that ceiling catches people.

Start with the trigger, and only one

Go to Website → Forms if the journey starts with an ActiveCampaign form. Build an inline form, collect only the fields you will actually use, and set the form action to apply a tag — guide-requested — rather than only subscribing to a list.

Tags make better triggers than list subscriptions because they are specific. A list says where someone is; a tag says what they did.

Then in Automations, create a new automation and choose that tag as the single start trigger.

The mistake to avoid here: adding two or three overlapping triggers because each seems relevant. Somebody who submits the form and gets the tag and joins the list enters the automation three times and receives the sequence three times. The symptom shows up days later as an embarrassed email from a subscriber. Start with one trigger and add others only once you understand the re-entry setting.

Worth a billing note too: on accounts created from 3 November 2025 every contact counts toward your paid limit, including spam signups. Put a CAPTCHA on the form — form spam is now a cost as well as a nuisance.

Deliver what you promised, immediately

First action: Send an email with the guide. No pitch, no newsletter signup push, nothing else. Somebody asked for a thing; give them the thing.

This sounds obvious and is the most common place a nurture sequence sours. The first email sets whether the next four get opened.

Add a wait, and understand what it is for

A wait is not padding. It is the window in which the contact produces the behaviour you are about to branch on — reading the guide, visiting the pricing page, doing nothing.

Three days is a reasonable default for a guide. A longer consideration cycle wants longer.

A gotcha worth knowing: wait steps run on the account's time zone, not the contact's. A "wait until 9am" step sends at 9am where your account thinks it is, which for an international list means the middle of somebody's night. Check the account time zone under settings before you rely on timed waits.

Branch on intent, not on opens

Add an If/Else condition. The temptation is to branch on whether the first email was opened. Do not.

Open tracking is unreliable now. Privacy features in mail clients pre-fetch images, which registers opens that never happened, and other setups block tracking entirely, which hides opens that did. Branching on that produces logic that is confidently wrong in both directions.

Branch on something that required a decision instead: a click on a specific link, a tracked visit to the pricing or services page, a booked consultation, a reply. Those are harder to fake and mean something.

Make the two branches genuinely different

The high-intent branch should do something a nurture email does not: send proof — a case study, a comparison, a specific answer to the objection that stage of the funnel has — raise the lead score, or create a task for a person to follow up.

The low-intent branch sends another useful thing and waits again.

If both branches send essentially the same email with a different subject line, the branch is complexity with no payoff. Delete it and send one email.

Add the exit condition before you turn it on

If somebody buys, they should stop receiving "still thinking it over?" emails that afternoon. Use a goal or an exit condition on the purchase event or the tag your checkout applies.

This is the step people add after the first complaint. Add it first — it costs one action and it is the difference between an automation that feels attentive and one that feels like it is not listening.

Test it properly, which takes fifteen minutes

Before it goes anywhere near real traffic, submit the form yourself with a fresh address and check all of this:

• The right tag and list were applied.
• The first email arrived, rendered correctly and the download link works.
• Force yourself down the high-intent branch by clicking the tracked link, and confirm the branch fires.
• Run a second test contact down the low-intent branch by doing nothing.
• Check the wait durations against the account time zone.
• Trigger the exit condition and confirm the contact leaves.
• Submit the form a second time with the same address and confirm you do not start a duplicate journey.

That last one is the test nobody runs and the one that catches the expensive mistake.

Three automations beat one big one

The biggest error in this product is treating complexity as sophistication. A sixty-step workflow handling acquisition, onboarding, upsell and winback is impressive on a screen and miserable to debug — and somebody will have to, probably not you, probably at the worst moment.

Three automations with one job each are easier to read, easier to test, and easier to switch off when one of them stops matching the business. Use tags to hand contacts between them rather than drawing everything on one canvas.

Name them for what they do, not what they are: "Guide request → nurture" beats "Automation 4 (new)".

When something does not fire

• Nobody is entering. Check the trigger actually matches what the form does — a form that subscribes to a list will not fire a tag trigger.
• Contacts entered but no email went. Check the send allowance. Plus allows 10× your contact limit per month across everything, automations included, and hitting it skips sends.
• Everybody took the same branch. Usually the condition looks at something that is never true — a tag applied later in the flow, or a page visit that site tracking is not installed for.
• Somebody got the sequence twice. Overlapping triggers, or the re-entry setting allowing it. This is the one to check first when a subscriber complains.

Our full review covers where this product fits, and if you are moving an existing setup across, the migration guide has the order to do it in.