Stop Optimizing Cost Per Lead. Run a Payback Window.
If you run a service business, your media buyer is probably reporting cost per lead to two decimal places. It is the wrong number. A 40 TL lead that never books is worse than a 400 TL lead that becomes a 38.000 TL job. The metric that actually governs whether you can scale is CAC measured against gross margin, lifetime value and the number of weeks it takes to get your cash back.
Why cost per lead lies to you
Cost per lead (CPL) treats every form fill and phone call as identical inventory. They are not. Between the lead and the bank deposit sit three multipliers that vary wildly by campaign: lead-to-booked rate, booked-to-completed rate, and average ticket. A channel that produces cheap leads at a 12% close rate and a low ticket can quietly lose money while a channel with double the CPL prints cash.
Here is the trap in numbers. Campaign A delivers leads at 150 TL and closes 10% into jobs worth 4.000 TL gross profit. Campaign B delivers leads at 350 TL and closes 25% into jobs worth 9.000 TL gross profit. CPL says A wins by a mile. But A's cost per acquired customer is 1.500 TL against 4.000 TL of margin, while B's is 1.400 TL against 9.000 TL. B is not just cheaper per customer, it is roughly three times more profitable. CPL pointed you at the loser.
The fix is not a better CPL target. It is to stop optimizing the top of the funnel and start optimizing the unit you actually bank: a paying, completed customer. Everything downstream of the lead is where the money hides.
Derive a target CAC, don't guess one
Target CAC is not a feeling, it is arithmetic. Start with lifetime gross profit, not revenue. LTV here means the gross margin a customer throws off across their relationship with you: first job plus realistic repeat and referral value, multiplied by your gross margin, discounted for the ones who never come back. Revenue-based LTV will flatter you into overspending.
Then pick a payback window: how many months of margin you are willing to spend to acquire a customer, and how fast you need that cash returned. A common healthy posture for SMB service firms is CAC ≤ first-job gross profit, so you are cash-positive on job one. A more aggressive growth posture lets CAC run to 2-3x first-job margin if repeat revenue is dependable and you have the cash to float it.
The formula is simple: Target CAC = (LTV gross profit) × (acceptable payback fraction). If a dental patient is worth 24.000 TL in lifetime gross profit and you want full payback inside the first treatment plan that yields 8.000 TL of margin, your hard CAC ceiling is 8.000 TL — not a kuruş more — even though lifetime economics could justify 24.000 TL. The payback window, not LTV, sets the budget you can actually survive.
Cash is the real constraint, not ROAS
A lifetime-value model can be perfectly profitable and still bankrupt you. The reason is timing. You pay the ad platform today, you book the job next week, you complete it three weeks later, and for many service businesses the invoice settles 30-60 days after that. Your margin is real but it arrives after the next ad invoice is already due.
This is why payback period beats LTV:CAC ratios for cash-constrained SMBs. A 5:1 LTV:CAC looks gorgeous on a slide, but if that 5x lands over 18 months and you are funding spend from a thin operating account, you will run out of cash long before the model pays off. The question is not how profitable a customer eventually is, but how many weeks your money is locked up before it returns to fund the next customer.
Practically, set a monthly acquisition budget you can replenish from collected cash, not projected revenue. If you collect 200.000 TL of margin this month and your payback window is under 30 days, you can responsibly recycle a large share of it into spend. If your payback is 90 days, your sustainable monthly budget is a fraction of that, no matter how attractive the lifetime math looks.
A lead, a booked job, and collected revenue are three different things
Most agencies optimize to the conversion the platform can see — the lead. But there are at least three gates after it, and money only exists at the last one. Gate one: lead to booked appointment. Gate two: booked to completed and invoiced. Gate three: invoiced to collected. A 30% leak at each gate compounds into roughly a third of your apparent pipeline reaching the bank.
This matters because Google and Meta optimize toward whatever signal you feed them. If you only send back the lead event, the algorithm hunts for cheap leads and will happily flood you with tyre-kickers. Feed it the booked-job or collected-revenue event with a value, and it starts optimizing for the customers who actually pay. Same budget, radically different customer mix.
So the instrumentation goal is to push the deepest reliable signal you have back into the ad platforms. For a clinic that might be a confirmed, attended appointment with an estimated value; for an HVAC firm it might be a completed work order. The closer your optimization event is to collected cash, the less CPL theatre you have to manage.
A worked example: an HVAC firm in İstanbul
Take an HVAC company running installs and service contracts. Average install revenue is 45.000 TL at a 35% gross margin, so 15.750 TL of first-job gross profit. About 40% sign an annual maintenance contract worth 6.000 TL revenue at 50% margin, and stay an average of three years — that adds roughly 3.600 TL of expected lifetime margin per acquired customer. Lifetime gross profit is therefore about 19.350 TL.
Now the funnel. Google Ads delivers leads at 300 TL. 35% become quoted site visits, 45% of those become signed jobs — so each customer costs 300 ÷ (0.35 × 0.45) ≈ 1.900 TL in CAC. Against 15.750 TL of first-job margin, payback happens on the very first completed install. That is a sub-one-job payback window: healthy, and it means you can reinvest collected margin into more spend within the same month.
Because CAC of 1.900 TL sits so far under the first-job margin, this business is not spending too much — it is almost certainly spending too little. The discipline flips: instead of pushing CPL down from 300 TL, you should be willing to pay 600 or 800 TL per lead if it buys higher-intent customers, because the payback window still closes inside one job. The ceiling is set by cash and capacity to deliver, not by an arbitrary CPL target.
What to change Monday morning
First, instrument the funnel past the lead. Wire your CRM or booking system so that booked, completed and collected each become a tracked event, and pipe the deepest one you trust back to Google and Meta with a value. Until you do this, every CAC number is an estimate built on the platform's most flattering metric.
Second, compute one number for each service line: target CAC = first-job gross profit × your payback fraction. Write the ceiling down. Then pull your last 90 days of spend and reconstruct true CAC per channel using close rate and ticket, not CPL. You will usually find one channel quietly subsidising another.
Third, set your monthly budget off collected cash and your measured payback window, then move money toward the channels where CAC clears the ceiling with the fastest payback — even if their CPL is the highest on the report. Stop rewarding the cheapest lead. Start rewarding the fastest, cheapest paying customer.
- CPL ignores close rate, ticket size and cash timing — three multipliers that decide whether a channel actually makes money.
- Derive target CAC from lifetime gross profit and a payback window: Target CAC = LTV margin × acceptable payback fraction.
- For cash-constrained SMBs, payback period beats LTV:CAC ratio — your money must return before the next ad invoice is due.
- A lead, a booked job and collected revenue are three different things; optimize ad platforms toward the deepest signal you can track.
- Set budget off collected cash, write down a hard CAC ceiling per service line, and reward the fastest-paying customer, not the cheapest lead.
Ready to scale?
Let's identify what's holding your campaigns back — and what it would take to fix it.
30-minute audit · no retainers to start