How to set a proving period for a new Google Ads campaign

Every new engagement begins with three numbers written down before the first dollar moves: the spend cap, the proving period length, and the verdict date.

A campaign without these is not a test. It is a donation.

This guide is the full version of how those three numbers get set, and what happens when the date arrives.

The spend cap

This is the all-in maximum: my fee plus the ad spend, stated as one number. If the total crosses the cap, the campaign stops, full stop. No extension conversation, no “one more week.”

The cap is set from the owner’s side, not from mine. The question is: what is the most you can afford to lose while learning whether this works? For most local service businesses that lands between $1,500.00 and $5,000.00 all-in for a first proving period, depending on the trade and the ticket size. The number is only right if the owner can say it out loud without flinching, because the whole point is that the downside is defined before the hoping starts.

A cap that requires a great month to survive is not a cap. It is a bet dressed up as discipline.

How to calculate your cap

Step 1: Determine your break-even acquisition cost

Break-even = Average ticket × Gross margin

Example: $800.00 ticket, 35% margin Break-even = $800.00 × 0.35 = $280.00

If you pay more than $280.00 to acquire a customer, you lose money.

Step 2: Set your target acquisition cost

You don’t want to break even. You want to profit.

Rule of thumb: Target = 60-70% of break-even.

Target = $280.00 × 0.70 = $196.00

Step 3: Calculate jobs needed for a fair test

10 booked jobs is the minimum for statistical significance.

Proving Budget = 10 booked jobs × target cost per booked job

At $196.00 target: $1,960.00 ad spend

Step 4: Add management fee

My flat fee is quoted separately. The cap includes both.

Example total cap: $1,960.00 ad spend + $500.00 fee = $2,460.00 all-in

Step 5: Can you serve 10 jobs?

If yes, your budget is correct. If no, your goal isn’t more budget. It’s better tracking.

How long the window runs, by trade

The proving period matches how the trade actually books work. A drain-cleaning call and an insurance roof replacement do not pay back on the same calendar, and pretending otherwise breaks the promise in exactly the verticals that most need it.

Trade	Proving window	Why
Emergency plumbing, drain cleaning	30 days	Calls close same-day
Cleaning, pest control	45 days	Quotes convert within a week
HVAC replacement	60 days	Multiple estimates, financing decisions
Roofing, insurance-covered	75 days	Claims adjusters lengthen the cycle
Auto detailing, mobile services	30 days	Immediate booking, high close rate
Electrical, garage door	45 days	Scheduled work, some same-day

The window is the minimum time needed for the money to complete its round trip: click to call, call to booking, booking to paid job. Shortening it produces a verdict on incomplete data. Lengthening it past the trade’s real cycle just delays the answer.

Why the window varies

Emergency trades (plumbing, locksmith, water damage): The decision is instant. A burst pipe at 2am means the homeowner calls the first number that answers. 30 days is enough to produce a number worth judging.

Scheduled trades (HVAC replacement, roofing, painting): Multiple estimates, financing, spousal discussion. The click-to-close cycle is 3-8 weeks. 45-75 days lets the cycle complete.

Recurring trades (cleaning, pest, landscaping): The first booking is the cheapest customer you’ll ever buy if the service keeps them. The proving period must be long enough to see the first renewal cycle.

The verdict date

On this date, the Ledger answers one question: continue, adjust, or stop.

The verdict requires confirmed evidence. Booked jobs, not clicks. If the proof stops at platform metrics because the owner never confirmed which jobs booked, the report says exactly that, and the verdict is downgraded to match the data.

Three honest outcomes, and all three are acceptable:

• Continue. The cost per booked job sits under the target, and the spend earns another period.
• Adjust. The numbers are close but not there. One named change, one more bounded window, same cap discipline.
• Stop. The numbers do not justify the spend. I say it out loud, the account pauses, and the owner keeps everything learned, including the audit findings.

A stop verdict is not a failure of the engagement. It is the engagement working. The owner paid a bounded amount to learn something expensive, and will never pay it again on faith.

What the Ledger shows on the verdict date

Metric	What it tells you
Total spend	Fee + ad spend, actual dollars out
Booked jobs confirmed	Client-verified count
Cost per booked job	Spend ÷ confirmed jobs
Revenue confirmed	Booked jobs × ticket values
Return per ad dollar	Revenue ÷ ad spend
Payback status	Has estimated profit crossed spend?

If the client’s margin is on file, the Ledger shows estimated profit at their margin. Without margin, it stops at revenue and says so. Revenue crossing spend is not payback; at 40% margin it’s still a loss.

The owner’s side of the window

The proving period has one requirement from the client: ten minutes a month, confirming which jobs booked and what they were worth. No confirmation, no Ledger, and the verdict date will say so.

This is stated in the offer, not discovered after signing. Reconciliation takes two signatures: my tracking and the client’s confirmation.

What “ten minutes” looks like

1. I send a simple list: tracked enquiry → keyword → booked? → ticket value?
2. Owner marks Y/N and fills in the ticket value
3. I update the Ledger, send it back
4. Next month repeats

If a client refuses this step, the Ledger stops at platform metrics and the report says: “Booked jobs not confirmed. Cost per booked job cannot be computed. Verdict based on platform leads only.”

What the written agreement contains

Before launch, the owner gets one page with:

1. The spend cap, all-in.
2. The proving period length and the verdict date, as calendar dates.
3. The target cost per booked job, and the arithmetic behind it.
4. What counts as an attributed job, and what deliberately does not.
5. The three possible verdicts, stated in advance.

If anyone proposing to run your ads will not put those five things in writing before the first dollar, you are not buying a test. You are buying a subscription.

Common objections, answered

“What if I don’t know my numbers yet?”

Then the first month’s job is finding them. The cap covers the discovery. The proving period starts when tracking is live and the negative list is loaded. The clock starts on real data, not on launch day.

“What if the season changes mid-window?”

The budget schedule follows the season. If you launch HVAC in May, the proving period is placed inside cooling season. If you launch in October, the budget rests until heating season. The verdict date lands while demand is real.

“Can I extend the window if it’s close?”

No. The cap is the discipline. If the numbers are close, the verdict is “adjust”: one named change, one more bounded window, the same cap. Extending the window without a named change is just hoping.

“What if Google’s automated bidding needs more time?”

Smart Bidding optimizes to what you told it counts as a conversion. If your conversion definition is clean, 10 booked jobs is enough data. If it’s not clean, more time just optimizes toward better junk.

The scaling rule that prevents catastrophe

Never scale more than 20% per month.

Month 1: $2,000.00 budget, 10 leads, 3 jobs, $667.00 CPBJ (break-even, so you pause)

• Lesson: You didn’t have enough volume.

Month 1 (corrected): $3,000.00 budget, 15 leads, 4.5 jobs, $667.00 CPBJ (same)

• Lesson: Same math, but now you can survive a bad month.

Month 2 scale incorrectly: $5,000.00 budget (+150%). Leads don’t scale linearly. You get 18 leads (+20%), not 45. Now CPBJ is $1,111.00. You just burned $2,000.00 learning what you already knew.

The correct path:

• $2,000.00 → $2,400.00 (+20%) → $2,880.00 (+20%) → $3,456.00 (+20%)

It takes 4 months to double your budget responsibly. This is intentional. Each month of data at each spend level makes the next month’s decision safer.

The mobile detailer’s lesson

One detailing account ran at $20.00/day with the formula:

Ticket: $350.00 average Margin: 45% Break-even: $157.50 Target CPA: $110.00 (70% of break-even)

$20.00/day × 30 days = $600.00/month $600.00 ÷ $110.00 target = 5.5 leads needed

But at $54.34 actual per lead (from the PPF/tint case study), $600.00 buys 11 leads. 11 leads × 30% close rate = 3.3 booked jobs = $1,155.00 in new ticket sales for $600.00 in spend.

That’s a workable business.

The opposite example: $20.00/day for a $50.00/lead cleaning company. 12 leads at 20% close rate = 2.4 jobs. On a $200.00 avg ticket, that’s $480.00 in sales for $600.00 spend. That’s a loss.

Budget fit to ticket fit, or you fail before you start.

Bottom line: Your budget is not a “suggestion.” It’s the minimum amount needed to prove whether a channel fits your business economics. Below that threshold, you’re not running ads. You’re burning money to learn nothing.

I count conservatively. Every untracked job is your upside, not my credit.