Planning your Google Ads budget for the year: by margin and demand

The budget comes at the end of the calculation

You do not set an annual Google Ads budget, you calculate it. For each month and product group:

Ceiling = allowed ad spend per unit × number of units that can be sold through ads

The sum of these values is your annual budget. That is more work than a percentage. In return, the calculation answers what a percentage leaves open: whether the budget pays for itself on average, and whether there is actually stock on the shelf for the sales it triggers.

A percentage of revenue is the most widely used method among small businesses, because it is simple. Critics consider it too cautious because it is tied to past sales. My objection is a different one: it treats every product group the same. Suppose two groups each bring in €100 net per unit. One leaves €40 before advertising, the other €15. At a flat 10%, the first keeps €30 and the second €5. The first group could carry more advertising, the second can barely carry what it already gets.

The usual formula built from target conversions, CPC and conversion rate contains neither stock nor variable costs. It works as a plausibility check.

From margin to allowed ad spend per unit

First work the product group through to contribution margin before advertising, then deduct what should at least remain after advertising. A worked example with round, invented figures:

The €15 is your business decision: it has to cover fixed costs and profit. Google cannot set it for you. Which items belong in the calculation is covered in the article on break-even ROAS. From the €25, this example gives a ROAS of 4 on net revenue, and with it the value for your target ROAS. Bid target and budget planning therefore draw on the same line.

The €25 also shows whether the plan is feasible. At an assumed 2% conversion rate and one unit per order, a click may cost 25 × 0.02 = €0.50 on average. If you pay considerably more for the relevant searches, the calculation cannot hold at this price. A bigger budget will not help then. You need a better offer, a better margin or a lower plan.

Each month, whatever runs out first sets the limit: stock or demand

For the number of units you need two columns. The first is the stock available in the month, minus what sells anyway through regular customers, newsletters or other channels. The second is the demand that Google Ads is likely to bring at the allowed €25 per unit. The smaller number counts. The €25 is an average, and the last sales of a month usually cost more.

December is where percentages and seasonal uplifts get it wrong. Demand is high, but the sensible budget is low, because November has sold most of the stock. A common objection to a flat monthly budget is that it runs out when demand is strongest. That is only true as long as there is enough stock. So reduce budgets actively at the turn of the month, or earlier if the planned stock has already sold.

The best estimate for the demand column comes from your own months in the previous year. There is a trap here: months with an empty warehouse or an exhausted budget show too little, because they measure your limit rather than the market. Adjust such months upwards before you use them as a baseline.

Google Trends helps with the shape of the curve. Each data point there is divided by all searches in the region and time period, so you see relative popularity, not volumes. A month with an index of 25 and 400 units sold gives 16 units per index point, which puts a peak month with an index of 100 at 1,600. That is an estimate, not a forecast: it assumes the cost per sale stays the same at the peak. I would therefore check it against last year's figure and take the more cautious value if they differ substantially. The curve says nothing about the size of a market.

Purchasing sets the peak season budget

The ceiling for November and December is fixed on the day the order goes out to the supplier. If you only discuss the Christmas budget in October, you are merely re-planning what purchasing decided long ago. That is why the budget round belongs before the order deadline. If the main selling season starts on 1 November and you need 12 weeks of delivery time plus 2 weeks for warehousing, the order has to go out by around 25 July. The planning round is therefore in July, not in autumn, and in it purchasing and marketing discuss the same number.

More budget requires more stock. In the example, December is 800 units short. If you order them, the ceiling rises by €20,000, and after advertising 800 × 15 = €12,000 of additional contribution margin remains. To get it, you have to pre-finance €36,000 of cost of goods plus the ad spend, and the demand figure has to materialise. This trade-off needs both sides at the same table, because neither marketing nor purchasing alone sees both sides. It also reveals the third limit alongside stock and demand: liquidity.

Too much stock raises the allowed ad spend. For goods that would otherwise only sell at a discount, deduct only what the unit would still earn without an ad, instead of the €15. Suppose 200 units would be left over after the season and would sell in January at €30 less, without ads, with €2 of costs for storage and tied-up capital. The alternative is then 40 − 30 − 2 = €8, and up to €32 of ad spend is justifiable for these 200 units instead of €25. This only applies to stock that really is left over and loses value. A best-seller that can be reordered does not get this uplift.

Scarce stock lowers the budget per unit. In December, 300 units are not enough for 1,100 possible sales anyway. If you raise the target ROAS for this group from 4 to 5, a sale may cost €20 instead of €25. If demand is still sufficient for the 300 units at this price, spend falls from €7,500 to €6,000, and contribution margin rises by exactly those €1,500. How to put this into practice in the account is described in the article on Google Ads by stock level.

Stock levels also have to be right in the feed. If a product is not available but is marked in_stock, Google disapproves it. If you want to hide available products briefly, there is the pause attribute for up to 14 days. Marking available products as out_of_stock is not a way to steer visibility.

Markets and channels share the same stock

Each market gets its own column with its own contribution margin, because shipping, returns, payment methods and prices differ. Suppose the same product leaves €40 before advertising in Germany and €28 in Switzerland after higher shipping and handling costs, with the CHF revenue converted. Then the ceiling per unit is lower there as well, even if ROAS looks similar at first glance.

Markets also compete with each other when they are supplied from the same warehouse. If stock is the bottleneck, allocate it first to the market with the highest contribution margin after advertising, and only then plan the budgets. Trends is of little help when comparing markets: the same search interest does not mean the same search volume. You can read the seasonal curve for each country, but not the size of the market.

The same logic applies to channels. Each unit is sold only once, so newsletter, marketplace, Google and Meta all draw on the same inventory. I would allocate stock in this order: first sales without media costs, then existing demand via Google, then demand-creating channels, as long as stock and margin remain. If you want to build reach for next season during the peak, you can deliberately set a different order. How to run the calculation between the two big channels is covered under budget between Google and Meta.

What Google's planning tools can contribute

Google's planning tools check whether your plan can be spent in the coming period. They do not replace annual planning. Performance Planner works with the last 7 to 10 days, adjusted for seasonality, and shows projections at monthly and quarterly level. The same data window applies to Keyword Planner forecasts. Neither knows your margin or your stock.

This also explains why Performance Planner can propose a budget of 0 for some campaigns: according to Google, they then do not contribute to the most efficient distribution of spend. In your own calculation, that very campaign may be the one selling high-margin or overstocked goods. I would therefore read such proposals as a hint, never as a plan.

To implement the plan, divide the monthly ceiling by 30.4, because that is how Google converts a month into days. €30,000 becomes roughly €987 a day. According to Google, up to twice that amount can be spent on a single day for most campaigns, so just under €1,974, while the monthly limit is 30.4 times the daily budget. A single daily budget is therefore a poor fit for a month with one strong week: there you can set a separate value for each week and check it against the monthly ceiling.

Plan once a year, recalculate on clear triggers

The annual plan is drawn up once, before the most important order deadlines. After that you compare actual against plan each month and recalculate when needed. A fixed date alone is too little in my view, because changes rarely arrive at the end of a quarter. These are the triggers I would set. The 20% is my rule of thumb:

In peak season or during market disruption, this becomes a weekly rhythm. Google also recommends weekly planning when markets change, until they settle.

No previous year, no demand figure

A new shop with no figures from the previous year has no reliable demand column. There, a deliberately fixed test budget is the more honest approach, and you treat it as the cost of learning until the data is in. The question of scale is covered in how much budget a shop needs.

With dropshipping and made-to-order production, the stock column becomes a capacity column: delivery times and production set the limit instead of the warehouse. With short delivery times and best-sellers that can be reordered, demand is almost always the bottleneck, and stock planning plays hardly any role in the budget.

If you deliberately buy new customers and reliably know their repeat purchases, you can spend more on the first sale than the table here allows. That requires solid repeat purchase data. With a broad range of thousands of items, you also plan by product group rather than by product, and only plan individually the items that carry the month. The calculation then only gives you a frame.