Marketing Automation Consultant: What to Have Them Build First

What a marketing automation consultant should build first for a service business: lead response, missed calls, reminders, reviews, and how to check the work.

By Carlos Familia, CRM & AI automation expert. Updated October 14, 2026. 7 min.

If you run a small med spa and pay for Facebook leads, this one is for you.

The marketing agency behind a med spa in Pennsylvania wanted every Facebook form lead to get a call soon after the form was sent. Before I built anything, I checked each link in the chain against live data. In the previous 14 days, 58 form leads had left a phone number, and not one had been called. Of the 45 most recent Facebook contacts, 42 had no tags when they were created, and the feed meant to announce new leads had been dead for weeks.

None of that needed AI. It needed someone to look before building.

What should a marketing automation consultant build for you?

For a service business, a marketing automation consultant should build five things, in this order: the first reply to a new lead, missed-call recovery, appointment reminders with no-show recovery, review requests, and old-lead reactivation. Each one should come with a test that proves it works.

goLance, a freelance marketplace, puts freelance marketing automation experts worldwide at $40 to $80 an hour at mid level, $80 to $140 at senior level, and $140 to $225 for specialists with ten or more years. Those are its 2026 figures, not my prices, so use them as a range to compare quotes against.

If your customers speak two languages, ask for both branches to be tested. My demo systems have English and Spanish branches, and a language router sends each lead down the right chain.

Why should you measure before they build?

Because one number can hide your real problem. A med spa in New Jersey looked fast on paper before I built its booking bot: half of new leads got a first reply within 36 seconds. But 1 in 10 waited 73 minutes or longer, and over 14 days, 2 leads never heard back at all.

If you had only looked at the median, you'd have been told the response time was fine. Ask your consultant for the median, the slow tail and the count of leads never answered, measured before any build. Then ask for the same numbers a month after launch.

What can quietly stop your leads from arriving?

Several things, and none of them show an error. These came from real builds:

• An unmapped Facebook lead form drops its leads. For an insurance agency in Puerto Rico, mapping the form backfilled 39 stuck leads in about three minutes. Those old leads did not trigger any workflow, so nothing followed up on them automatically.
• Facebook lead-form leads can create two contacts that GoHighLevel merges 1 to 2 seconds after the webhook. A fast reply can go to a contact that no longer exists.
• Moving a stage and then adding a guard tag fires the next workflow before the tag exists. On a wellness clinic funnel, that sent the same follow-up WhatsApp message to leads two or three times. Tag first, then move.
• A "tag added" trigger fires only when the tag is newly added. Re-adding a tag the contact already has does nothing, so a tag that signals "call this lead" is an event, not a state.

Ask the consultant how they confirm that a test lead from each form really arrives, and how they handle a form that was never mapped.

What should happen to your old leads and no-shows?

Old leads are the cheapest pipeline you have, and the easiest to mishandle. At a med spa in New Jersey, a reactivation tagger tagged 94 contacts on its first run. Eight were wrong, and I cleaned them up. The second run tagged 1 contact, with zero leaks.

For a facial spa in the same state, the first run tagged 68 backlog contacts, with a maximum of 4 touches per lead and a gate for Meta's 24-hour messaging window.

No-shows need a sequence that stops itself. For an HVAC company in California, the recovery runs weekly for 12 weeks and exits as soon as the person rebooks, replies or changes stage.

Why do your texts need registration first?

Because in the US, business texts from local numbers need carrier registration (A2P 10DLC), and any text step you build before approval is a step you can't count on. HighLevel's own help center says registration is required when a business sends SMS or MMS to US recipients from standard 10-digit local numbers.

I've run into this twice. For a roofing contractor in Kentucky, I held back a rule that cancelled the appointment when nobody replied to a text. Before approval the text never arrives, so every booking would cancel. For an HVAC company in California, the registration was rejected, so email carried the follow-up and the SMS steps stayed parked.

A consultant who skips this will deliver a system that looks finished and sends nothing. My guides on GoHighLevel A2P and missed call text back cover the details.

What should you see when the work is done?

You should be able to open the account and see what was built, in order, with a test report next to it. As an example, I built 8 demo sub-accounts for different industries, with about 69 workflows each. They are my own demo builds, tested end to end, not client results.

Across the 8 accounts, 268 test journeys passed. Each journey is a test contact walking one path through the workflows, and the logs prove which steps ran. The first version passed 104 of 104 journeys, with about 3,300 executed steps.

A good handover has five things: a list of workflows with plain names, folders that match your pipeline stages, a test report, a list of the tags and fields in use, and logins in your name.

Ask your consultant for the same kind of evidence. A folder per pipeline stage, a list of workflows, and a test report beat a screen recording. I describe the testing method in how I test GoHighLevel workflows.

Which small details cost you the most?

Small settings that change what customers receive. Four I would ask any consultant about:

• Editing a saved email template changes what already-published workflows send.
• Gmail fetches the tracking pixel on delivery, so an email nobody touched can count as "opened".
• Bounce and complaint workflows in one of my builds turned email Do Not Disturb off, the opposite of what they were meant to do.
• Asking only happy customers for reviews is review gating. Filter by rating when thanking, never when asking.

More of these are in my notes on GoHighLevel workflow mistakes.

What should you ask before you sign?

• What will you measure before you build, and what will you measure a month after launch?
• Which lead sources will you test, and how will you prove a test lead from each one arrives?
• What is the plan if text registration isn't approved yet?
• Where is the test report, and can I read the execution logs myself?
• Who owns the accounts when you finish?

Should you hire a marketing automation consultant, a CRM consultant or a GoHighLevel expert?

Choose by where the work lives. A marketing automation consultant designs follow-up, reminders and reviews. A CRM consultant fixes the pipeline, fields and data underneath. A GoHighLevel expert builds it inside that platform.

If part of the project is an AI agent on the phone or in chat, add an AI automation consultant to the list. Many people cover two of these roles, so ask which parts they test themselves. If you want a second opinion on your plan, book a call in the calendar below.

P.S. If you'd rather install a ready-made system yourself, the niche Automation Packs are $22 each (dental, roofing, HVAC and more). The HVAC pack includes a missed-call workflow, and I tested it the same way as the rest.

About the author

Carlos Familia, CRM & AI automation expert. GoHighLevel, n8n, AI agents, servers. Full-stack developer.

I build CRM systems in GoHighLevel, n8n automations on my own servers and AI voice and chat agents for service businesses. Full-stack developer: I write the workflow, the API integration, the database and the server it runs on, and I test every build with real journeys before handoff.