- Published
- Byline
- GoHighLevel Partners Editorial
- Filed under
- GoHighLevel Automation
How to design email automation that reads the CRM record rather than a stale export, and the settings that quietly break it once it is live.
CRM Marketing Email Automation When the CRM Is Also the Sender
CRM marketing email automation is the practice of letting the record that already holds a person’s history decide what they are sent next, instead of exporting a list and mailing it from somewhere else. The difference is not cosmetic. When the contact record and the sending engine live in one system, an email can be conditioned on a booked appointment, a pipeline stage or a form answer at the moment it fires, and it can be stopped by the same events. When they live apart, every automation is running on a snapshot that was accurate when the export ran.
For an agency operating client sub-accounts, that unified model is what makes email work sellable at all. You are not maintaining a separate list per client; you are maintaining a pattern, and the pattern reads whatever the client’s own record says.
The four objects every automated email depends on
The contact record
Everything else hangs off it. If duplicates exist, a person receives the sequence twice and unsubscribes once, which quietly removes them from both records. Deciding early which field is the identity — email address, phone number, or both together — costs an hour at the start and prevents a class of problem that is painful to unpick once thousands of records exist.
The field or tag that carries state
An automation cannot read intent, only data. Whether someone is a new enquiry, a no-show or a past client has to be written somewhere before a workflow can branch on it. Tags are fast to add and easy to sprawl; custom fields are slower to set up and far easier to audit a year later. Most agency builds end up using tags for transient state and fields for anything a human will need to read.
The trigger
The trigger is the event that starts the sequence: a form submitted, an appointment booked, a stage changed, an inbound reply received. The most common design error is triggering on something too broad, such as any contact created, and then filtering heavily inside the workflow. Narrow triggers are easier to reason about and much easier to test.
The workflow itself
This is where waits, conditions and the emails live. It is also where the same build gets rewritten three times, because the first version encodes a specific client’s offer instead of the shape common to the vertical. Build the shape, then let fields supply the specifics.
Four kinds of automated email, and where each belongs
| Type | What starts it | Good for | Main risk |
|---|---|---|---|
| Broadcast | You, manually, to a segment | Announcements, events, one-off offers | Sending to a stale segment nobody re-checked |
| Nurture sequence | A single entry event, then timed waits | New enquiries, lead magnets, long consideration cycles | Running on regardless of what the person has since done |
| Behaviour-triggered | An action or a change of state | Abandoned forms, no-shows, stage movement, reactivation | Firing repeatedly when the state flips back and forth |
| Transactional | A specific transaction or booking | Confirmations, reminders, receipts | Being buried inside a marketing sequence that can be paused |
Keeping transactional mail separate from marketing mail matters more than it looks. A confirmation that lives inside a nurture workflow inherits that workflow’s pauses, quiet hours and unsubscribe behaviour, so a client who opted out of the newsletter stops receiving appointment confirmations too. Build them as their own workflows.
A build order that avoids rework
- Write the journey in plain sentences first. One line per step, in the order a real person experiences it. If a step cannot be described without naming the tool, it is not yet a step.
- Decide the exit conditions before the emails. What should stop this sequence — a booking, a reply, a purchase, a tag? A sequence with no exit is the one that mails a customer an offer for something they bought last week.
- Set the technical sending foundation. Authentication and sender reputation determine whether any of this is read at all, so the SPF, DKIM and DMARC setup comes before the copy, not after the first client complaint.
- Build one branch end to end. Take the most common path, get it working with a test contact, and only then add the alternatives.
- Add the scoring or prioritisation last. Ranking contacts is useful once the sequence itself is stable; adding it early makes every debugging session twice as long. The mechanics of incrementing a score inside a workflow are straightforward once the branches are settled.
The traps that break a working sequence quietly
- Re-entry behaviour. Whether a contact can enter the same workflow twice is a setting, and the wrong choice produces either a duplicate sequence or a contact who never gets the follow-up they should. The re-entry rules are the first thing to check when a workflow appears not to fire.
- Time zones on waits. A wait of one day and a wait until nine the next morning behave differently for a contact three time zones away. Sequences that looked fine in testing send at three in the morning in production.
- Quiet hours applied to the wrong channel. Sending windows configured for text often do not govern email, or vice versa. Confirm which channel a window actually covers rather than assuming it covers both.
- Unsubscribes handled per sub-account. An opt-out has to remove the contact from every marketing sequence, not only the one that prompted it. Test this deliberately, with a real address you control.
- List hygiene left undone. Repeatedly mailing addresses that bounce damages the sending reputation the whole sub-account shares. Suppress hard bounces automatically rather than reviewing them monthly.
What to measure, and what to leave alone
Open rate has become the least reliable figure on the report. Privacy features that pre-fetch images register opens nobody performed, which inflates the number in a way that varies by audience and by mail client. It still has some use as a trend line on one list over time, and almost none as a comparison between clients.
The figures worth putting in front of a client are the ones tied to something that happened: replies, bookings, pipeline stages entered, and the unsubscribe and complaint rates that tell you whether the frequency is tolerable. Complaint rate deserves particular attention because it is the one metric with an external consequence — mailbox providers act on it, and the damage lands on every client sending from the same infrastructure.
Finally, measure at the sequence level rather than the email level. A single message that gets few replies may still be the one that keeps a contact warm for the message that gets the booking, and pruning it because it looked weak in isolation is how a working sequence gets slowly dismantled.
Frequently Asked Questions
What makes CRM email automation different from a standalone email tool?
The condition and the send happen against the same live record. A sequence can branch on a booking made ten seconds ago, or stop because a pipeline stage changed, without an import or a sync delay. A standalone tool works from whatever was true when the list was last exported to it.
How many emails should a nurture sequence contain?
There is no correct count, only a correct exit. Decide first what event should end the sequence, then write enough messages to cover the period a person realistically takes to reach that event. A short sequence with a clean exit outperforms a long one that keeps mailing people who already converted.
Should transactional emails live in the same workflow as marketing ones?
No. A confirmation or reminder placed inside a marketing sequence inherits its pauses, sending windows and opt-out behaviour, so someone who unsubscribes from the newsletter can stop receiving appointment confirmations. Build transactional messages as their own workflows with their own triggers.
Why did a workflow not fire for a contact who clearly qualified?
The usual causes are re-entry rules preventing a second pass through a workflow the contact already completed, a trigger filter that is narrower than expected, or the contact being unsubscribed from the channel. Check those three in that order before rebuilding anything.
Is open rate still worth reporting to a client?
Only as a rough trend on one list over time. Image pre-fetching by privacy features registers opens that no person performed, and how much it inflates the figure varies by audience and mail client. Replies, bookings and pipeline movement are the numbers that survive scrutiny.
How does one client’s email behaviour affect the others in an agency account?
Sending reputation is attached to the domains and infrastructure used to send. If one sub-account mails an unclean list and collects complaints or hard bounces, deliverability can degrade for anything sharing that sending path. Separate sending domains per client and automatic bounce suppression both reduce the exposure.
We build your GoHighLevel account for you
CRM, pipelines, funnels, SMS and automation — configured for your business, not left for you to figure out.
- Setup (one-time)$850
- Ongoing (per month)$38.80
No obligation quote · reply within 1 business day
This site is independently operated and is not officially affiliated with, endorsed by, or sponsored by HighLevel LLC. We do not receive referral commissions — this is our own flat-fee implementation service.