How much budget an online shop needs for Google Ads

The budget buys information, and that information is counted in orders

A budget in euros says little until someone adds how many orders it is meant to buy. Guides suggest around €300 to €500 to get started. At €20 per order, €400 still buys 20 orders; at €150, it does not buy even three. The same amount is a test in one case and chance in the other.

So I work backwards: what question should the month answer, how many orders does that answer need, and what does one order cost. That gives you the requirement. Set a second number against it, the loss limit: the amount you can afford to lose if the answer turns out badly.

Cost per order = cost per click ÷ conversion rate Monthly budget = orders needed × cost per order

Before launch, you estimate the cost per click with Keyword Planner, which shows the average cost for a keyword. It remains an estimate, and for Shopping I only use it as a rough approximation. The conversion rate comes from your own shop, not from someone else's benchmarks. It does, however, include repeat customers and direct visitors, so I assume a lower rate for cold ad traffic.

How many orders you need depends on the question

The figures in circulation answer different questions. Mix them up and you plan either too tight or too expensively. Google's figures are as of November 2026.

Statistics show why 15 orders are an entry ticket and not yet an answer. For count data, the mean and the variance are equal, so the spread is the square root of the count. For a 95% range, my model calculation uses twice the spread: around 52% at 15 orders and around 28% at 50. A measured ROAS of 4 then means roughly 1.9 to 6.1 at 15 orders and roughly 2.9 to 5.1 at 50 orders. If your break-even ROAS is 3, the range reaches deep into loss territory in the first case and just below it in the second. Only at around 60 orders does it reach break-even. Fluctuating order values add to this, so the real uncertainty is larger. How many orders are enough in a given case is covered in the article on conversions for target ROAS.

Worked example: the same revenue, a completely different requirement

Suppose two shops each turn over €30,000 a month. All figures are invented and rounded.

The requirement follows the cost per order, not the size of the shop. A small shop with expensive orders needs more budget than a large one with cheap orders. That is why no table that sorts budgets by revenue band will help you.

A percentage of revenue is an outcome

The often-quoted ratios are no use for this question. Gartner's figure of 7.7% of revenue covers the entire marketing budget and comes mostly from companies with more than USD 1 billion in revenue. I know of no sound Google Ads ratio for online shops.

As a check at the end, the percentage is still useful. In the example, the 50 orders cost Shop A 3.3% of its monthly revenue and bring in €3,000, a tenth of revenue. For Shop B it is 25%, and the €20,000 from ads would be two thirds of its revenue. That figure tells you the plan for B is a big bet. It does not tell you how much budget it needs.

The second number: what you can afford to lose

The risk is not the whole budget, because even a disappointing test brings orders and therefore contribution margin. You calculate the possible loss for a bad case:

Loss = budget × (1 − contribution margin per order ÷ actual cost per order)

Suppose both shops have a contribution margin of 40% before ad spend, which is €160 per order for Shop B. The plan assumes €150 per order, a surplus of €10. If the actual cost turns out 50% higher, at €225, the €7,500 budget buys only around 33 orders, and around €2,200 is missing. The same bad case costs Shop A around €200.

You set the loss limit before launch, as an amount, based on the contribution margin you actually have to spare, not on the plan. The calculation counts only the first order. If your new customers reliably buy again, you can be more generous, as long as you can back that up with your shop data. If the contribution margin is already below the cost per order in the plan, the budget no longer buys information, and the question belongs with the cases in which Google Ads does not pay off for the shop.

Requirement above the limit: make the question smaller

The obvious reflex is the most expensive one: spreading a tight budget across several campaigns so that everything is included once. Afterwards, no campaign has enough orders to say anything. According to Google, low daily budgets can also slow down the ramp-up. Instead, I would check the following in this order:

1. One campaign instead of several. Google itself advises consolidating campaigns where possible so that conversion data is pooled.
2. A smaller range. You start only with the products where contribution margin and purchase intent fit together best. That can lower the cost per order, but the answer then only applies to that slice.
3. A longer measurement period. 50 orders over two months cost no less, but the money goes out more slowly and you can stop earlier. In exchange, seasonal effects mix in more strongly.
4. A rougher question. With 15 to 20 orders, you cannot tell whether the ROAS is 3.5 or 4. You can see, though, whether the cost per order is roughly in line with the plan or twice as high.
5. Not starting yet. If even the rougher question cannot be answered, that is a commercial decision.

How to split the budget

The number of campaigns follows from the budget, not the other way round: campaigns = monthly budget ÷ (orders per campaign × cost per order). At €2,000, €20 per order and 50 orders, that makes two. At €40 per order, it makes one.

I do not give fixed ratios for Search, Shopping and Performance Max, because none of them can be backed up. In my view, an additional campaign is only justified when it pursues a different commercial goal and gathers enough orders of its own; that is the logic of an account structure built on profitability. I always mentally set brand queries aside, because they flatter the cost per order and otherwise make the requirement for everything else look too small.

For the daily budget, divide the monthly budget by the average number of days in a month, around 30.4. In Performance Max, daily budget × 30.4 is the monthly charging limit, and on individual days spend can reach up to twice the daily budget. A few expensive days are therefore not a budget error. You also do not have to get it right the first time: according to Google, budget changes do not trigger a "Learning" status and do not remove historical data. Google advises evaluating performance after the ramp-up over two conversion cycles or two weeks, whichever is longer, and Performance Max only after at least six weeks. That is the minimum period; how much the result tells you depends on the number of orders.

Without demand, no budget helps

Before launch, check the clicks you need: orders divided by conversion rate, in the example 2,500 for Shop A and 5,000 for Shop B. If there are not that many relevant search queries, more budget will not help either.

Miscounted orders undermine the whole calculation, so reconcile them with your shop system first. For very expensive products with few orders, 50 a month may simply be out of reach; then the loss limit sets the budget, and you read the result only roughly over a longer period. Finally, an account that is already running no longer needs this derivation: there you have real costs per order, and the question is what the next euro brings in. That is covered in the article on limited by budget.