How a Cleaning Company Fixed a Broken Zapier Automation in a Day

A cleaning company's Zapier automation for review requests stopped after a sender-address change. One call traced the fault upstream, and the Zap never changed.

Oct 19, 2026
How a Cleaning Company Fixed a Broken Zapier Automation in a Day
💡
TL;DR: A Zapier automation that had quietly added every new customer to Mailchimp and triggered a review request since 2023 stopped cold after the company changed the address its receipts were sent from. One diagnostic call and one follow-up email traced the failure upstream of Zapier, into the software that generates the receipts, and the company's own developer had it fixed the next day. No rebuild, and the paid development session that had been booked was cancelled.
Laptop on a small-business desk, where the broken Zapier automation was traced. Photo by Kari Shea on Unsplash
Laptop on a small-business desk, where the broken Zapier automation was traced. Photo by Kari Shea on Unsplash

The situation

The client is a small cleaning company that dispatches crews by truck and bills customers from an in-house, Microsoft Access–based job database. When a job is completed, that database emails the customer a receipt. Behind the scenes, a Zapier automation built for them in 2023 turned each completed job into two marketing actions: a new contact in Mailchimp, and a review request sent through their review platform.
It ran for more than two years without anyone touching it.

The problem: a Zapier automation that went silent

The address the receipts went out from started landing in customers' spam folders. The fix seemed obvious: send receipts from a different company mailbox. The moment they switched, the automation stopped. No new Mailchimp contacts, and no review requests going out after completed jobs.
That's the expensive kind of failure for a local service business. Nothing errors out loudly. Receipts still go out and customers still get billed, so the only symptom is a slow drop in fresh reviews and a mailing list that stops growing.

What they'd already tried

The client's first move was the sensible one: go looking for the rule that forwarded receipts to Zapier, assuming it lived in Outlook. They checked their Microsoft 365 mail rules and found nothing that matched. They also asked the developer who maintains the job database, who said the problem couldn't be on his side. With nobody left to ask, they tracked down the team that built the original automation.

What we did

A Connex consultant walked through the setup live on a quick consultation call:
  • Zapier side: the Zap's trigger was a "new inbound email" step listening on a Zapier forwarding address, followed by the Mailchimp "Add/Update Subscriber" step and the review-request step. The Zap itself was intact and needed no changes.
  • Mail side: nothing in the Outlook rules was forwarding receipts. So the trigger email wasn't a forward at all.
  • The tell: a sample of the trigger email contained the customer's name, phone number, and email, but the customer-facing receipt didn't include a phone number. That meant a second, separate email was being generated by the job database itself, not by Outlook.
After the call, the Connex team dug into the 2023 project history and found the original setup notes. Back then, the client's IT contact had been asked to have the database send a second email to a dedicated address that ultimately fed the Zap. The follow-up email to the client said it plainly: the database is likely sending both the receipt and the Zapier email, so "the fix might be with that system, not with Zapier."
With that pointer, the developer found it by the next day. The email the database sent to Zapier still carried the old "from" address inside its body, and that mismatch was why the automation stopped firing. He corrected it and the automation resumed.

The results

  • Back online within about a day of the first call. The diagnostic call happened one afternoon, and the fix was confirmed the next.
  • Zero changes to the Zap, and no rebuild. The 2023 automation kept running as designed.
  • The paid development session that had been booked to fix it was cancelled because it wasn't needed.
  • Mailchimp contact creation and post-job review requests resumed, with the receipts now going out from the new, non-spam-flagged mailbox.
There's no dollar figure here, and we're not going to invent one. The client didn't share job volumes, so we can't honestly estimate them. The real result is that a quiet revenue leak (missing reviews and a list that had stopped growing) was found and closed in a day instead of turning into weeks of trial-and-error rework.

In the client's words

"With the information [Connex] provided our software writer was able to determine where in his code the problem was happening. Previously, he claimed it could not have been part of his code." — the client, by email
The client followed up asking to bank the unused time toward future work, including an AI phone-answering automation she's considering.

FAQ

Why would changing a sender address break a Zap that never mentions that address? Email-triggered automations often depend on a second, machine-generated email, not on the message the customer sees. If any system along the way filters on, copies, or embeds the old address, the chain breaks even though the Zap looks fine.
My Zap stopped firing but shows no errors. Where do I look first? Check whether the trigger is even arriving. If Zapier's trigger history is empty, the problem is upstream: a mail rule, a forwarding setting, or the app that sends the email. Rebuilding the Zap won't help.
What access did this need? A screen-share of the Zapier account and Outlook rules on the first call, plus the original project notes. The actual fix was made by the client's own developer in their own system, so they stayed in control of their data.
Do you only support automations you built? No. We regularly take over and repair Zaps and Make scenarios built by others. Having the original history just makes it faster.

Have an automation that quietly stopped working?

If a workflow that used to run on its own has gone silent, we can usually find where it broke on one call. Book a time with the Connex team.