How a Google Ads Agency Matched Leads to Revenue in 4 CRMs

A Google Ads agency for home-services firms swapped fragile Zapier lookups for Make + Airtable lead-to-revenue matching across four CRMs. See how it works.

Nov 18, 2026
How a Google Ads Agency Matched Leads to Revenue in 4 CRMs
TL;DR: A small Google Ads agency serving home-services companies needed every call and form lead tied to the quotes and jobs it turned into. The data lived in each client's own CRM. Connex replaced a set of Zapier lookups that kept breaking with a lead-to-revenue matching layer built in Make + Airtable. It now covers clients on Pipedrive, ServiceTitan, Jobber, and Workiz and holds about 162,000 records across five bases, with old records cleaned out automatically after 18 months. One fix to how estimates were matched raised a single client's tracked quote value by about $50K.
Photo by Raze Solar on Unsplash
Photo by Raze Solar on Unsplash

The situation

The agency is a small U.S. shop that runs Google Ads for home-services businesses: plumbing, heating and cooling, roofing, junk removal. Its whole pitch depends on proving which ads produce paying jobs, not just phone calls.
Lead tracking runs through WhatConverts, a call- and form-tracking platform. When a lead in WhatConverts carries a quote value and a sales value, the platform can pass those values back to Google Ads, so bidding learns which clicks turn into revenue.
The catch is that the revenue doesn't live in WhatConverts. It lives in each client's field-service CRM, and every client runs a different one. Some use Pipedrive, some ServiceTitan, some Jobber, and later Workiz.

The problem: revenue lived in a dozen systems the ad platform couldn't see

On the first call in early 2023, the agency described it plainly. They were trying to "bring in quote/sales value from CRMs including Pipedrive/ServiceTitan or via Google spreadsheet into WhatConverts," and they were "not sure how to update a lead's quote/sales value from the CRM automatically." Some clients simply sent spreadsheets or CSVs.
The hard part wasn't moving one number. It was the shape of the data:
  • One lead, many jobs. A homeowner who calls from one ad might book a repair, then a replacement a month later. Both jobs came from the same click, so their values need to be added together, not overwritten.
  • Quotes vs. sales. An estimate isn't revenue. The quote value and the sales value had to land in different fields and change as the job moved through its stages.
  • No shared ID. The CRM doesn't know the WhatConverts lead ID. The only common ground is a phone number or email and the order things happened in.

What they'd already tried

The agency had Zapier connections in place, and they worked some of the time. A zap would fail with a "lead ID missing" error even though every test showed the lead ID coming through. The lookups behaved inconsistently, with no clear pattern to when they worked.
The bigger problem was cost. Adding up multiple deals per lead meant searching back through 60 to 90 days of past deals on every update. A Connex consultant flagged on that first review that building those searches as Zapier steps would make task costs skyrocket as volume grew. Zapier was the right tool for sending data out. It was the wrong place to do the matching.

What we built: a lead-to-revenue matching layer

The Connex team moved the heavy lifting into a dedicated matching layer: Airtable as the working database, Make for the scenarios, and Zapier kept only where a light hand-off made sense.

One pattern, adapted per CRM

For each client, jobs, deals, and estimates flow into Airtable from that client's CRM, by webhook where the CRM supports one and by scheduled pulls (as often as hourly) where it doesn't. New and updated leads come in from WhatConverts by webhook. Make then matches each job to its lead and writes the quote and sales values back to the right lead in WhatConverts.
The same structure was built for Pipedrive, then ServiceTitan, then Jobber, then Workiz. Each new client gets a copy of the base and scenarios, organized in its own Make folder.

Matching rules that hold up against real field-service data

  • Phone and email matching with time logic. A job only matches a lead that came in before it. Timestamps are converted to UTC so time zones don't create false matches.
  • Sales value only when the job is done. The sales value is set when a job is marked complete. Until then, the quote value carries the number.
  • Clear estimate rules. If several estimates connect to the same lead and one is sold, the sold value syncs. If none are sold, the highest estimate syncs. Sold and unsold estimates are never added together.
  • Estimates without a job. Some CRMs let an estimate exist without a job attached. Those are matched by searching the jobs table by phone and email around the creation time, with a fallback search directly against the leads.
  • Updates, not just new records. Jobs updated after they were created are re-checked for a lead match, so a lead doesn't stay unmatched just because the job changed later.
  • New lead sources. Text-message leads and online booking-request forms were brought into the same matching flow as calls and web forms.

Staying under the database ceiling

By late 2024 the agency's main base was approaching the 125,000-record-per-base limit on Airtable's Business plan. Before anything was deleted, the Connex team confirmed that removing old Airtable records would not reset values already synced into WhatConverts. A daily cleanup now removes connected leads older than 540 days (about 18 months), plus leads and jobs that never matched anything.

The results

These are real figures from the engagement, not estimates:
  • Four CRMs, one revenue loop. Clients on Pipedrive, ServiceTitan, Jobber, and Workiz all send quote and sales values back to the same lead-tracking platform, and from there to Google Ads.
  • About 162,000 records across five bases. The largest base holds about 113,000. Some data refreshes every 30 minutes.
  • About $50K in quote value surfaced from one fix. When estimates without an attached job started matching by phone and email, one client's tracked quote value rose by roughly $50K. That revenue had been invisible to the ad platform.
  • Under the cap without manual cleanup. The 18-month retention runs daily, so the bases stay under Airtable's record limit without anyone deleting rows by hand.
The outcome the agency sells is visibility. Its clients' ad budgets now optimize toward booked and completed jobs, not call counts.
There's no time-saved dollar figure here because the agency never measured how long the old spreadsheet and CSV process took. We'd rather leave the number out than invent one.

What took longer than it should have

Scaling the pattern to new CRMs showed where the logic lived only in scenarios and in one consultant's head. During one handoff, a filter removed by mistake quietly stopped a client's matches from updating until it was caught and fixed. The fix was documentation: a written and recorded walkthrough of how the matching works, so the next build doesn't start from scratch.

Client voice

The agency now speaks about the system as its own. When booking a later build, its lead wrote: "We've built several similar integrations using Make.com and Airtable for platforms like ServiceTitan, Jobber, and Pipedrive, so we're familiar with the general structure and matching logic." The work has continued for more than three years, across more than 30 build sessions. The next phase is already scoped: moving the storage layer from Airtable to a Postgres database (Supabase), with Make writing to both during the cutover so nothing goes down.

FAQ

Will this work with our CRM?
If the CRM can send webhooks or exposes an API for jobs and estimates, the same pattern applies. One client's Workiz plan didn't include webhooks, so the build waited until the plan was upgraded. Check that before you scope anything.
Why not just do this in Zapier?
Zapier works well for sending a finished value somewhere. Matching one lead to many jobs means searching and summing past records on every update. In a task-priced tool, that cost grows with every lead.
What happens when we hit the Airtable record limit?
Set a retention rule early. Here it's 18 months, which covers the comparison window the agency reports on. Confirm first that deleting old records won't change values already sent downstream.
How long does adding a new client take?
Because the logic is reused, the Workiz build was scoped at about five hours. That's an estimate from the scoping call. The first versions took longer while the matching rules were still being worked out.

Stuck connecting leads to revenue?

If your ad reporting stops at "calls received" because the revenue lives in a CRM nothing talks to, book a call with the Connex team. We'll look at your stack and tell you what's worth automating.