Online advertising budget: how much to invest and how to calculate it

Why “a percentage of turnover” is not the full answer

When you ask how much to invest in online advertising, you are often given a percentage of turnover. That benchmark can help with general marketing planning, but it says nothing about what the money will actually achieve. For a business with a high margin and returning customers, the same percentage may be too low. For one with a thin margin and one-off sales, it can lead straight to a loss.

A good budget answers a different question: how much can you pay to win a customer and still make a profit? Everything else follows from that: how high you can bid, which channels make sense, how much to test and when to scale. A budget set by instinct (“what I can afford this month”, “what competitors spend”) leads to one of two opposite mistakes: you either stop too early, with no data, or you carry on for months in a channel that cannot be profitable.

Another reason a percentage is not enough is the gap between budget and goal. If you want 100 orders a month and a customer costs you 25 EUR, you need 2,500 EUR for ads alone; if your budget is 500 EUR, the realistic result is around 20 orders, however well you optimise. The numbers have to fit together.

For a first picture, the ads budget simulator lets you play with cost per click, conversion rate and budget to see an indicative outcome. The result depends on your assumptions, so treat it as a sketch, not a forecast.

Step 1: start from margin, not revenue

Sales are not profit. To know how much you can spend on ads you need your contribution margin: what is left of an order after product cost, packaging, shipping, payment fees, returns and other variable costs, but before advertising. Details of the calculation for shops are in our article on the profit margin of an online shop.

Two key figures follow from the margin:

• Break-even cost per acquisition (CPA): the contribution margin per order in euros. If you pay more than that for a new customer, you lose money on the first order.
• Break-even ROAS: revenue divided by contribution margin, which is 1 divided by the percentage margin. At a 35% margin, break-even ROAS is about 2.86: every euro spent must bring in at least 2.86 euros in sales for you not to lose.

Mind the VAT: calculate revenue without VAT, otherwise the margin looks larger than it is. A “good” ROAS does not exist in the abstract, only one that is good for your margin. A shop with a 60% margin may be profitable at a ROAS of 2; one with 20% needs 5 just to break even.

A worked example, step by step

Imagine a hypothetical online shop, with figures chosen purely for the example. They are not market averages and should not be copied as they stand.

Notice what the example shows: on these assumptions the estimated cost per customer (27 EUR) is very close to the break-even level (28 EUR). There is no profit on the first order. You have three levers: raise the conversion rate (going from 1.5% to 2% would bring CPA down to 20 EUR), lower the cost per click, or raise the order value. Each of these is improved outside the ad budget; for the first, see how conversion rate is calculated.

For services the logic is similar, with one extra step between the click and the customer. If a lead costs you 15 EUR and one in five leads becomes a customer, a customer costs you 75 EUR in advertising. Is that good or bad? It depends on the profit a customer brings, not on the figure itself.

Step 2: the long-term value of a customer

If your customers come back or pay on a recurring basis, you can accept a higher cost on the first purchase, because the profit arrives over time. Two concepts help: LTV (the average lifetime value of a customer, as profit, not revenue) and CAC (the total cost of acquiring a customer). The ratio between them shows whether your business model can support paid advertising.

Work them out with the LTV and CAC calculator. Caution matters here: do not use an optimistic LTV that counts repeat purchases that have not happened yet. For new businesses with no history, start with the value of the first order and treat repeat business as a bonus, not a guarantee.

Include in CAC not only the money paid to platforms but also creative costs, agency fees and sales team time. If you ignore these, advertising looks cheaper than it is and the budget swells as you go.

How big should the test budget be?

A test that is too small tells you nothing: it does not produce enough data to tell a good campaign from a lucky one. A test that is too large burns money on unproven assumptions. The balance depends on a minimum number of conversions, not on a round sum.

Ad platforms need a minimum volume of events for their automated systems to learn. Platform documentation has mentioned thresholds of roughly tens of conversions a month for automated bidding in Google Ads and, at Meta, around 50 optimisation events per ad set a week to exit the learning phase. These figures can change, so check the current values in the platforms’ interfaces and official help pages.

A practical calculation: test budget = desired conversions x estimated CPA x 1.3 to 1.5 (a cushion for wrong assumptions). In the example above, 30 orders at 27 EUR come to about 800 EUR, to which you add the cushion. If your budget is below what the minimum volume needs, you have two options: concentrate the money on one campaign in one channel, or pick an easier goal (for instance leads rather than direct sales) so that you collect data faster.

Think about seasonality too: a test in a quiet month may understate the potential, and one at the peak may overstate it.

Budget phases: test, optimise, scale

A budget is not a fixed sum for the whole year; it evolves in phases. Each phase has a clear purpose and a condition for moving to the next.

1. Preparation (before the first ad): conversion tracking works correctly, the landing page has been checked, and you have calculated the maximum CPA and a decision threshold. Without correct measurement you cannot judge anything; see conversion tracking with GA4 and Meta Pixel.
2. Testing (generally four to eight weeks): a limited budget, a few message and audience variants; the aim is to find out what works, not to maximise sales.
3. Optimisation: remove what does not work (negative keywords, weak audiences, tired creatives), consolidate what does and track CPA against your threshold.
4. Gradual scaling: raise the budget in small steps (a common rule of thumb is 15 to 25% at a time) and wait a few days to see the effect, so that the system does not lose stability.
5. Maintenance and expansion: when a channel reaches its limit (cost per customer starts to rise as you grow), move extra budget to another channel or audience.

You move to scaling only if CPA stays below your threshold for long enough, not after one good day. It is also common for CPA to rise as you increase the budget: the first conversions are the cheapest, and the following ones gradually cost more.

How to split the budget between channels

There is no universal split. A useful way of thinking: search channels (such as Google Ads) capture existing demand, meaning people already looking for what you sell, while social channels create demand and reactivate it through remarketing. The detailed comparison is in Google Ads or Facebook Ads.

To begin with, you can give most of the budget to the channel with the strongest buying intent and a smaller share to a test on a new channel. A 70/30 split, for example, is a reasonable place to start, not a rule. After the first data, move the money to where cost per customer is lower and volume allows.

In Google Ads the budget is usually set as an average daily amount; on a particular day Google may spend up to about twice the daily amount, but over a month you are not charged more than roughly the daily budget multiplied by 30.4 (check the current rules in the official help). If the budget is too small for the available demand, campaigns can lose impressions; if it is too large for what you can manage, wasted spend creeps in.

What belongs in the budget besides ad spend

A realistic budget covers more than the amount paid to the platform. If you leave these out, your profitability calculation comes out too optimistic.

• Creative: photos, video, copy, variants for testing.
• Landing page and website: if the page is weak, money spent on traffic is wasted.
• Measurement: setting up tracking, reports, any paid tools.
• Campaign management: an agency fee or internal time.
• VAT and platform fees, where they apply. If you cannot recover VAT, the standard Romanian rate is 21%.
• A 10 to 20% reserve for adjustments and opportunities.

Common mistakes when setting the budget

• Starting without knowing your margin and maximum CPA.
• A budget too small for the minimum data volume, followed by the conclusion that “ads do not work”.
• Stopping a campaign after a few days, before it has left the learning phase.
• Doubling the budget at once after one good week.
• Judging profitability on revenue instead of margin.
• Ignoring the landing page and conversion rate.
• Comparing channels on different attribution models without checking against your own data.

How we work

Before we talk about sums, we ask for the basic figures: order value, margin, repeat-purchase rate, delivery capacity and the monthly goal. On that basis we propose a test budget with a decision threshold, not a “starter” sum picked at random. If the data shows that a channel cannot be profitable under the given conditions, we say so before the money is spent.

When running campaigns we track cost per customer against the agreed threshold each month and propose scaling only where results hold. For details of how we manage accounts, read about our Google Ads service or write to us through the contact page.

In short

• Start from contribution margin and maximum cost per customer, not from a fixed percentage.
• Break-even ROAS is 1 divided by the percentage margin.
• The test budget must produce enough conversions to draw conclusions.
• Phases: preparation, testing, optimisation, gradual scaling, expansion.
• Include creative, landing page, measurement and management in the budget.