How We Structure Google Ads Conversion Actions (The Right Way) After Importing From ServiceTitan

If your Google Ads account is optimizing for phone calls from the same customer five
times in one day… you’re not optimizing for revenue, you’re optimizing for Karen.

That’s the problem with most Google Ads setups: too many conversions, too much noise,
and not enough clarity on what actually drives ROAS.

Most PPC managers miss the crucial post-import step even if they are using
ServiceTitan Marketing Pro Ads: properly editing each conversion action for count method, attribution window, and
optimization priority. The result? Duplicate conversions, inflated performance reporting, and Smart Bidding stuck
chasing the wrong signals.

We fix that, every time.

Here’s how we structure conversion actions from ServiceTitan Marketing Pro Ads to
give Smart Bidding the cleanest, highest-fidelity signals possible, and avoid the common traps that break your ROAS
before it ever has a shot.

What It Enables

ServiceTitan Marketing Pro Ads doesn’t just pass form fills and phone calls into
Google Ads, it gives you something most trades advertisers have never had: the ability to optimize for actual job
value.

Whether you’re passing through:

• Actual revenue from closed
  jobs, or

• Job Value Prediction (JVP)
  from ServiceTitan’s AI model (when jobs aren’t complete yet),

…you’re no longer guessing what a conversion is worth. You’re telling Google:

“This lead is projected to bring in $12,748,
go find more like it.”

That changes the game entirely.

It means Smart Bidding is trained on real downstream value, not just lead signals
like call duration or form fills. It means you can shift from Cost Per
Lead to Return on Ad Spend (ROAS)… and you can do it with
confidence.

But only if you configure those imported conversions correctly.

What JVP Is & Why It Matters (a
Lot)

Job Value Prediction (JVP) is ServiceTitan’s
AI-powered estimation of how much revenue a booked job will generate. It uses historical performance, job type,
technician, geography, and more to give Google Ads a real-time revenue signal, even if the job hasn’t closed yet.

That’s a big deal. Because in trades, there’s often a huge lag between booking a job
and collecting actual revenue.

Why That Lag Breaks Hurts ROAS Optimization

• For HVAC, the job might run same day, but the invoice could close days later.

• For high-ticket installs or custom work, jobs might not be complete for
  weeks.

• In cases like garage doors or remodels, it could take 12+ weeks to close
  the loop.

If you’re relying solely on completed-job revenue for bidding signals, you’re
always optimizing on stale data, sometimes 2–12 weeks behind the current funnel.

Now imagine trying to scale during summer with performance decisions based on May
bookings. That’s a losing play.

Why JVP Solves It

JVP gives you the predicted value at the time of booking, and updates it
retroactively once the job completes and revenue is finalized. That means:

• You get immediate performance data for Smart Bidding

• You still get corrected revenue later for reporting accuracy

• You’re no longer flying blind or waiting weeks to make strategic shifts

It’s the best of both worlds: fast signals for optimization, and accurate data
for ROAS tracking.

Real-World Example: Garage Doors

Take a garage door company selling custom doors:

• The lead books a tech consult

• The door is sold and ordered, but install won’t happen for 8 to 12 weeks

With normal revenue tracking, that lead looks worthless until Q4.

With JVP? You immediately see a projected $8,400 in job value tied to the ad click
that generated it, and you can double down on what worked while it’s still working.

Why We Use JVP (And Stopped Trying To Build A Cheaper One)

We abandoned any effort to build our own value model years ago. There’s just no
competing with the volume of data ServiceTitan has:

• Tens of millions of historical jobs

• Hundreds of variables per transaction

• Constant model refinement and revenue correction baked in

We focus on performance strategy, not reverse-engineering machine learning models
that have already won.

Our Cheat Sheet: The Free Agency Conversion Settings Framework

Once those conversion actions land in Google Ads from ServiceTitan Marketing Pro Ads, we don’t leave anything to chance. Here’s how we
structure everything to ensure bidding trains on real value and attribution stays clean.

Primary Conversions (Train Smart Bidding)

Only these should be set to Primary,
included in account goals, and factored into ROAS optimization.

Why “One” vs. “Every”?

• Contact actions (calls,
  bookings) use Count: One to avoid duplicates from
  frequent or indecisive contacts.

• Revenue-based actions
  (JVP) use Count: Every to capture all job value, even from repeat customers.

Pro Tip: To improve match rates for contact-based
conversions, enable Enhanced Conversions for Web under Goals → Conversions → Settings in Google Ads (not inside individual
conversion actions). Choose the Google tag setup method and allow
automatic detection or provide a code snippet to capture hashed user data (like name, email, phone).

Secondary Conversions: Still Useful, Just Not Optimization Drivers

These should still be imported, but set to Secondary so they don’t influence bidding
or ROAS reporting. They help you see funnel behavior, but shouldn’t train Google’s Smart Bidding model.

Translation: track everything, but only optimize for the clearest,
value-tied signals. That’s how you give Smart Bidding the right inputs and avoid chasing inflated or
redundant outcomes.

The Exception (Specifically For The First ~30 Days)

Here’s where we contradict ourselves, temporarily.

The two call-based primary actions above, ServiceTitan Calls Conversion Action and Calls from Ads, are filtered. One requires a 60-second minimum call duration plus a “Lead” status in ServiceTitan. The other uses a >90s threshold to weed out junk. Long-term, those filters are exactly what you want feeding Smart Bidding.

But on day one? They’ll starve it.

Here’s why. Your CSRs are human. Early on, they won’t be used to the inevitable competitor calls that come in through as your PPC team is pruning search terms. A homeowner calls asking for another company, your CSR politely says “wrong number”, and excuses the call so it doesn’t count against their booking rate. That caller is calling for a service you provide, in your service area, but never gets the “Lead” tag. Google never sees a conversion. And you may think “good, I don’t want to be optimizing toward my competitors calls, they’re harder to book”, but you’d miss the point.

Multiply that across dozens of calls in your first two weeks. Google’s spending your money, seeing almost no conversions come back, and concluding your campaigns aren’t working. Smart Bidding starts pulling back. Clicks drop, impression share tanks, and you’re in a death spiral before the account ever had a chance. So, volume first, refine second.

The fix: Flip the priority for the first 14–30 days.

Temporary primary actions (launch phase):

During this phase, move ServiceTitan Calls Conversion Action and Calls from Ads to Secondary. They’ll still track, you just won’t optimize on them yet.

This gives Google a realistic volume of call conversions to learn from while your CSRs build the habit of tagging dispositions correctly. You’re trading signal precision for signal volume, which is the right trade when the algorithm has nothing to work with.

When to switch back: Once you’ve hit roughly 30 conversions in 30 days on your primary call actions, the threshold where Smart Bidding has enough data to optimize reliably, swap them. Move the filtered actions (Calls Conversion Action and Calls from Ads) back to Primary. Move the unfiltered inbound call actions back to Secondary.

Think of it like breaking in a new dispatcher. You don’t hand them the most complex routing rules on day one. You let them take every call, learn the patterns, then tighten the criteria once they know what good looks like. Same logic applies to Google’s algorithm.

Bottom line: The framework above is where you want to end up. But if you set it on day one, you’ll choke the algorithm before it learns anything. Give it volume first. Give it precision second.

Final Thoughts

ServiceTitan Marketing Pro Ads gives trades businesses something most
platforms can’t: direct access to both actual revenue and predicted revenue through Job Value Prediction
(JVP), seamlessly pushed into Google Ads. That means you’re not just optimizing for contact, you’re
optimizing for what actually drives the business.

But when those conversions land in Google Ads, they don’t come perfectly
pre-configured. And that’s where most setups fall short, not out of bad intent, but because no one teaches
you how to clean it up.

Here’s what we’ve learned after rebuilding 100+ accounts:

• JVP is your MVP,
  revenue signals > contact signals

• Count contact actions
  once, not every time Karen calls

• Use secondary
  conversions for context, not bidding

• Give Smart Bidding clean
  signals so your budget chases profit, not noise

These aren’t hacks, they’re fundamentals. But fundamentals most people
overlook. If you want Smart Bidding to act like a pro, you have to feed it like one.

Need a Second Set of Eyes?

We’ve helped trades businesses go from default setups to dialed-in ROAS
machines, often with nothing more than better signal prioritization.

Let’s chat conversion structure
and make sure your campaigns are chasing the right wins.

Bri Ski

 bri@freeagency.ai