Jobber to GoHighLevel: Lead Response and Review Automation

A local service business had two real gaps. New leads got no response until someone had time to notice and reply, and review requests only went out when someone manually remembered to tag a contact after a job wrapped. Both problems traced back to the same root cause: Jobber, the field-service platform the business ran on day to day, and GoHighLevel, the CRM handling marketing and communication, were not actually talking to each other.

Connecting them was not a single step. GoHighLevel offers two separate Jobber integration paths with overlapping but different trigger sets, and one of them only installs at the agency level and gets pushed down into the sub-account, it cannot be installed from inside the sub-account at all. That is not obvious until you go looking for it, and picking the wrong one would have meant building the whole system on a trigger that does not actually exist where you need it.

Once the right integration was in place, the real work was finding out where the platform's own claims did not hold up. A widely cited third-party write-up listed quote and request triggers for this integration that simply do not exist when you check the live trigger list directly, so anything planned around them had to be rebuilt around what was actually there: a trigger for new clients and a trigger for completed jobs, nothing else. The one contact-matching action available only supports exact-match filtering with no OR logic, so a lead who gave a phone number but no email, or the reverse, could not be matched with a single lookup. That got solved with two chained lookups instead of one, phone first and email as a fallback, each carrying its own check so the workflow never fires on someone already tagged as an existing customer.

The client already had a review-request system, and a genuinely good one, but it was not the only one in the account. A vendor-supplied starter template had come bundled with the Jobber marketplace app, and a third, unused template sat in a separate test account from an earlier evaluation. Two of the three had a workflow with the exact same name doing different things, which meant publishing the wrong pair would have sent the same customer two separate review requests. Every workflow across all three systems got read node by node before anything was touched, existing send history and all, so the decision on what to keep was based on what each one actually did, not on its name.

The strongest system was the one already built internally: it had duplicate-prevention, a proper multi-touch follow-up sequence, and referral logic the vendor template never had. Rather than throwing that away, the vendor template got stripped down to the one thing it did have that the internal system was missing, a live trigger for completed jobs, and rewired to just add a tag. That tag is what the internal system was already listening for, so the fix closed a real gap (the whole system depended on a tag that nothing was actually adding) without duplicating a single piece of working logic.

Along the way, testing against real sends rather than trusting the workflow canvas caught two things that would have shipped broken otherwise. A leftover joke message a previous builder had left inside a live template, apologizing for a coffee-deprived typo, was one accidental send away from going out to a real customer as a second automated text. A mapped data field also turned out to sync correctly for some values and come back silently blank for others, only caught by creating several real test contacts with different values and comparing what actually landed on each one.

A fourth automation the client asked about, a win-back agent for stale, unanswered quotes, turned out to be backed by real money sitting on the table, tens of thousands of dollars across a handful of old quotes. But there is no live event anywhere in the platform for a quote going stale, confirmed directly rather than assumed, so it would need its own scheduled job rather than a workflow trigger. Scoped it clearly and paused it rather than shipping something half-built just to call it done.