When ads start going well, the first move that comes to mind is raising the budget: it takes a single setting and brings more clicks the next day. But if the measurement is incomplete, what grows is not sales but only spend, and the difference often becomes clear only at the end of the month, when someone looks at the till. Before adding money to the account, we suggest writing down an answer to four questions.
Which conversions are counted
The first question looks simple, but in many accounts the answer is unclear: what does the ad platform count as a conversion, and what does it leave out? Answering it calls for a measurement audit.
- Make the list. Write down the key events in GA4, the conversion actions in Google Ads and the events in the Meta pixel side by side. Note which conversion the bidding strategy is aiming for, that is, which ones are marked as primary.
- Look for double counting. The same purchase may be imported from GA4 and also counted by the Google Ads tag. A new sale may be recorded every time the thank-you page is refreshed.
- Look for the wrong goal. If add-to-basket, a page view or a click on the phone number has been made a primary conversion, the platform will try to multiply those instead of sales.
- Look for gaps. When cookie consent is refused, with ad blockers and under some browser restrictions, the pixel in the browser misses the event. Server-side tagging, the Meta Conversions API and enhanced conversions in Google Ads narrow this gap. If the same event is sent from both the browser and the server, it must be deduplicated with a shared event ID.
- Connect offline sales. If the sale closes in a quote meeting for a B2B manufacturer, at the examination for a clinic or on the phone for an education provider, the platform only sees the form. Matching the won deals in your CRM to the click ID (such as the GCLID for Google Ads) and uploading them back as offline conversions makes the platform look for the people who become customers, not the people who fill in forms.
The last step of the audit is a manual test: trigger each conversion once yourself and write down how many times it appears in GA4, in the ad platforms and in the CRM.
Platform sales against the till
Ad platforms credit sales to themselves under their own attribution rules. For example, a platform may count the seven days after a click and the one day after the ad was viewed. If the same customer saw ads on two platforms, the sale shows up in both reports. That is why the platform reports usually add up to more than the sales in the till.
The way to see this is a weekly reconciliation table. Each row shows, for the same week, the sales each platform reports, the sales GA4 reports and the net sales in the e-commerce panel or ERP. Returns and cancellations are deducted from net sales; all columns are kept on the same basis, for example excluding VAT and shipping. On the B2B side, the same table is built with the value of won quotes in the CRM.
The aim is not an exact match; that is rarely possible. The aim is for the gap to be explainable and steady from week to week. Once the table has been kept for a few weeks, the usual ratio between platform and till emerges, and platform figures can be read with that ratio in mind. If the ratio suddenly changes, the first place to look is the measurement setup. If the sales the platform reports are rising while the sales in the till stand still, you are reading a false signal for a budget increase.
The cost of the next order
The number people look at most in ad accounts is the average cost per order: total spend divided by total orders. For a budget decision, though, what really matters is the cost of the next order the extra budget will bring. As the budget grows, the platform moves into more expensive auctions, less relevant audiences and lower-intent searches; the added orders usually cost more than the earlier ones.
Say you spend 50,000 TL a week and get 250 orders: the average cost per order is 200 TL. If raising the budget to 70,000 TL takes orders up to 300, the extra 20,000 TL has brought in the extra 50 orders, and each of those orders has cost 400 TL. The new average is about 233 TL and looks reasonable in the dashboard; yet the decision should be based on 400 TL.
The budget decision is set by the cost of the next order, not the average cost.
To see the marginal cost, it is enough to log budget changes with their dates and compare the periods before and after each change, with everything else held constant. If the account has several campaigns, the comparison is made campaign by campaign: in one, extra budget may still bring in orders at a reasonable cost while another has hit its limit.
Searches for your brand name also distort this calculation. Brand campaigns bring in orders at low cost, but some of those orders would have come through the organic results even without the ad. You need to separate out the brand orders that flatter the average and look at the cost of non-brand campaigns on its own.
What the margin can carry
The last question is how much an order can set aside for advertising. This means calculating the contribution margin: subtract the product cost, shipping and packaging, the payment fee, the marketplace commission if there is one and the expected cost of returns from the sale price, then divide what remains by the sale price. Everything is calculated on the same basis, excluding VAT. Break-even ROAS follows from a single formula:
break-even ROAS = 1 / contribution margin
For example, take an order of 1,000 TL excluding VAT, with a product cost of 450 TL, shipping and packaging of 100 TL, and a payment fee and returns allowance of 50 TL. The contribution before advertising is 400 TL, and the contribution margin is 0.40. Break-even ROAS = 1 / 0.40 = 2.5. In other words, every 1 TL spent on ads for this product must bring in at least 2.5 TL of sales excluding VAT; the same calculation means an ad cost of no more than 400 TL per order. Once ROAS drops below this limit, every sale that comes in grows turnover but loses money.
Back to the example in the previous section: if the average order is 1,000 TL, the 400 TL cost of the added orders sits exactly at break-even. The extra 20,000 TL has grown turnover, not profit. If the sale value the pixel sends includes VAT, the ROAS the platform shows must be brought onto the same basis too. If you have repeat customers, you can accept a higher cost on the first order; but work this out from the real repeat purchase behaviour in your own data, not from an assumption.
Raising it step by step
Once the answers to the four questions are in writing, a budget increase is no longer a guess but an experiment. This is how we set it up:
- Write the stop rule in advance. For example: “If the marginal cost per order goes above 350 TL for two weeks in a row, we go back to the previous budget.” This rule goes into the decision log with its owner and review date.
- Increase in small steps. For example, by a fifth of the budget each time. Big jumps can push the platform back into a learning period and make the result harder to read.
- Hold each step for a week or two. Don’t change the creative, the targeting and the landing page at the same time; otherwise you can’t separate the effect of the increase from the other changes.
- Look at the till. At the end of each step, decide on the net sales in the reconciliation table and the marginal cost, not on the platform report.
With the four answers in writing
When you know which conversions are counted, the gap between platform and till, the cost of the next order and the limit your margin can carry, a budget increase stops being a move to fear. Seeing these four answers in the same report every week is a simple way to make sure profit grows as the account grows. If you would like to set up the measurement audit and the budget plan together with us, take a look at our advertising service.