An advertising account can be created quickly. Money can leave it even faster. The worst campaign is not the one that obviously fails. It is the campaign with a nice click count and a green report, where the company discovers only later that it subsidized every order.
For this article I prepared an Excel spreadsheet and my first narrated video guide in 2015. The video had its flaws and, according to a contemporary home review, was rather boring in places. The basic question has not aged: how much can you afford to pay for an advertising result so that something is still left for the company?
An advertising metric is not yet a business result
CPC says what a visit costs. CTR says how many people clicked the ad. The platform may show purchases or inquiries. None of these alone says whether the campaign is profitable.
First I need to know the result the advertising should produce. For an e-shop it may be a paid order. For a B2B company, a qualified inquiry, only some of which become contracts. With a long buying cycle, the first measurable step may be a meeting booking. Each result has a different value.
So I do not begin with “What CPC should we set?” I begin with:
- What exactly counts as a campaign result?
- What value does it have for the company after direct costs?
- How many such results can sales and operations actually process?
- Where and how will we measure the result?
The maximum order cost comes from contribution, not revenue
Revenue looks good in a presentation. But advertising is paid from the money left after the costs of selling. For a product, I count at least purchase cost, company-paid shipping, packaging, payment fees, returns, and the work created by the order. For a service, I count the people’s time needed for delivery.
The simplified process is:
- Take revenue excluding amounts that do not belong to the company.
- Subtract variable costs of delivering the product or service.
- Decide what share of the remaining contribution customer acquisition may consume.
- Leave room for fixed costs, measurement errors, and profit.
The result is the maximum acceptable cost per order or customer. If the actual cost remains higher over time, changing a few words in the ad is not enough. The offer, website, price, margin, or repeat-purchase work must change — or the campaign should be stopped.
ROAS is a useful indicator, not universal truth
ROAS is the ratio of advertising cost to revenue. It works well for a quick check of the same assortment. But when one category has a high margin and another almost none, the same ROAS can mean completely different profit.
I am similarly careful with future customer value. It makes sense to include it when the company’s own data shows how many customers return, after how long, and with what contribution. The assumption “they will definitely buy again” is not customer value. It is an Excel wish.
Measurement must come before the budget
Before launching a campaign I verify that a submitted inquiry, order, or other real result is recorded. I use consistent UTM parameters and check that the data reaches company analytics. The advertising platform and analytics tool may differ because of attribution. What matters is agreeing in advance which source drives decisions, rather than choosing the best-looking number afterwards.
Google Analytics 4 can connect website visits and events, but it does not know margin or inquiry quality by itself. Those data must come from the shop, CRM, or accounting view. Marketing analytics without business data is only half a map.
A test needs a limit, not endless patience
A new campaign needs room to collect data. That does not mean it should run without conditions. I set the test budget, period, and possible decisions in advance: continue, adjust, or stop.
I check the whole chain:
- Advertising: does it attract people for whom the offer makes sense?
- Page: does it match the advertising promise and lead to one clear step?
- Sales: does it respond promptly and recognize inquiry quality?
- Economics: is money left after acquiring and serving the customer?
When measurement is missing, the page does not work on mobile, or sales cannot respond, another budget usually makes the problem larger. In that situation I recommend slowing the advertising and first reviewing the online marketing checkpoints.
Four numbers I want before increasing the budget
- The real cost of a business result, not just the cost of a click.
- The contribution left for the company after delivering the product or service.
- The share of inquiries that become business.
- The company’s capacity to serve new customers well.
Advertising is not a money machine. It is an amplifier. It can accelerate a good offer and functioning business. It can make unclear economics more expensive.
If you need to connect an advertising report with what really happens in sales and operations, see how I work with companies. The first step is not a larger budget. It is deciding what the budget should earn.