You have the products, the store is open, and the next question seems simple: how much money should you put into ads? You look for a concrete answer, but you meet amounts that differ widely and recommendations that don't say what they include. Some refer only to the money paid to the platforms. Others include management, but leave out photos, measurement or fixing pages.
An online store's promotion budget is set starting from the contribution per order, the goal of the test and the resources needed for execution. Separate media spend from materials, management and preparing the site. Build scenarios, set the acceptable loss limit and check commercial results before expanding. There is no universal minimum amount that guarantees orders or profit.
Start with three decisions: which products you promote, what you want to find out and how much you can sustain without compromising stock or the running of the business. The budget thus becomes a controlled choice, with visible assumptions, not a number copied from another store.
Contents
- What goes into the budget
- The economics of an order
- The goal of the test
- Three hypothetical calculations
- Products and channels
- Cash and the calendar
- Small budget and limited data
- The budget sheet
- The 30-day plan
- Mistakes to avoid
- Frequently asked questions
- Checklist and next step
What goes into the promotion budget
The money for distributing ads
The media budget is the amount paid to the platforms for displaying and distributing ads. It is the part you see most easily in the advertising account, but it doesn't represent the whole cost of the activity. If you have several platforms, decide who consolidates the spend and which currency the shared report uses.
Monthly planning must be tied to how the platform manages spend. In Google Ads, an average daily budget shouldn't be read as an identical ceiling for each day. Consult the documentation on budgets and spend and check the rules that apply to your campaign. Don't promise in the plan a perfectly even daily distribution if the mechanism used doesn't work that way.
Materials and their preparation
Photos, video demonstrations, copy, format adaptations and product updates take time or money. Some materials can be reused, but production must be funded before you find out how much they will contribute. Separate the initial cost from the cost of recurring updates.
You can start with a limited set of good materials, tied to a few real arguments. You don't automatically need expensive production, but not dozens of superficial variants either. For a technical product, a precise explanation of compatibility can be more useful than a spectacular animation that doesn't answer the buyer's question.
Management and infrastructure
Include setup, management, analysis, any tools and the technical interventions needed. State explicitly whether the collaboration's price covers the feed, measurement, landing pages and reporting. A seemingly lower offer can have a different scope, without necessarily being more efficient.
If the Shopify store needs adjustments, treat them separately in the plan. A checkout fix can serve all traffic sources, not just the new campaign. Allocating the cost is a management convention you must explain, not change to make a channel look better.
Team time and the intervention reserve
Who answers customers' questions, approves materials and checks stock? If all these tasks depend on the same person, their capacity can limit the test. A financial budget without operational availability produces delays that end up being mistaken for advertising problems.
Keep a reserve for foreseeable situations, such as replacing a creative or fixing an integration. The reserve doesn't have to be spent. Its role is to let you make a justified intervention, without consuming the money needed to deliver orders already received.
Start with the economics of an order
Revenue is not the amount available for ads
The order value must be defined consistently: with or without taxes, with or without shipping, before or after discounts and returns. For the economic decision, talk to the person responsible for finance about the relevant net basis. Don't combine gross revenue from the platform with net costs from accounting and then draw a conclusion about profit.
Contribution before advertising is the relevant revenue minus the variable costs associated with orders. These can include the cost of products, payment processing, packaging, shipping paid and the estimated effect of returns. The exact list depends on the business; the aim is to avoid omitting expenses that grow with sales.
From the contribution, money must still remain for fixed costs and the desired result. That is why the threshold at which advertising consumes the entire contribution is not a healthy default target. It is only a simplified limit, useful for seeing how quickly the economic room can be lost.
Calculate by category when differences are large
A store average can hide products in completely different situations. Accessories, bulky products and items with many returns shouldn't automatically be treated identically. If you promote mostly a category with low contribution, the general average can overestimate how much you can afford to pay for those orders.
Choose a level of detail you can maintain. It isn't useful to build a very sophisticated model on incomplete data. You can start with three commercial groups and refine them once you have enough orders, updated costs and an owner for the records.
Separate the order from the new customer
Cost per order divides the spend by the number of orders. The acquisition cost of a new customer uses new customers, identified consistently, and a declared cost scope. A customer can have several orders, and some orders may come from people already won earlier.
For the budget, both perspectives are useful. The first shows the pressure of advertising on transactions. The second helps analyse expanding the customer base. Don't rename cost per order as “new customer cost” just because the platform reports purchases and you haven't yet done the reconciliation.
Define what you need to learn from the first test
A clear question makes the amount easier to justify
“I want more sales” is an intention, but it doesn't describe the test sufficiently. A more useful question is whether a category can attract retained orders under an acceptable media cost, using a well-defined offer and channel. That way you can identify which conditions must be prepared and which information you will track.
Write the hypothesis before launch. Mention the products, the relevant audience, the commercial argument and the interval in which you will evaluate. You don't need a sure prediction. You need a formulation that makes a conclusion possible even when the result is weak or the data remains insufficient.
Distinguish exploration from expansion
A store without history also buys information: it discovers costs, objections and problems in the journey. A store with stable data can test whether results hold when it invests more. The two situations justify different plans and shouldn't be compared only by immediate revenue.
Still, the label “learning” doesn't justify unlimited spending. Decide what you will concretely learn and how it can inform the next decision. If you can't say what question the test clarified, it is worth reviewing how it was built, not just extending its duration.
Decide what would stop the test
There are technical reasons to stop: a non-functioning checkout, a wrong price, unavailable products or duplicate conversions. There are also economic reasons: a loss above the accepted limit, insufficient contribution or the inability to fund delivery. These situations shouldn't be treated the same as normal day-to-day variation.
Note who can suspend spend and who confirms the fix. A simple rule prevents discussions in which each person assumes the other is tracking the problem. Don't turn every day without orders into an automatic alarm, though, especially when the volume is small and purchase has a delay.
Three hypothetical examples for the budget
Hypothetical example 1: the media-per-order ceiling
You analyse a category in a completed month. The average net value of a retained order is 250 RON. The cost of products is 140 RON, and the other average variable costs are 25 RON. Contribution before advertising is 250 − 140 − 25 = 85 RON per order.
You want to keep 35 RON for fixed costs and result. The working ceiling for media becomes 85 − 35 = 50 RON per order. This is a planning choice for our example, not a cost the market is obliged to offer.
If you aim for 80 orders at this cost, the corresponding media amount is 80 × 50 = 4,000 RON. The formula shows the relationship between the goal and the acceptable cost. It doesn't prove you will get 80 orders with that amount, because demand, competition and page performance remain unknown.
Hypothetical example 2: scenarios for traffic and conversion
For a 30-day test you plan 3,000 RON in media. You assume, for the calculation only, an average cost of 1.50 RON per click, so 3,000 / 1.50 = 2,000 clicks. In the scenario, these correspond to 1,700 measured sessions. Clicks and sessions are different units; the relationship isn't fixed in practice.
If 34 of the sessions include an order, the rate of sessions with an order is 34 / 1,700 × 100 = 2%. Assuming a single retained order for each such session, the media cost per order is 3,000 / 34 ≈ 88.24 RON.
In another scenario, 51 of the same 1,700 sessions include a retained order each. The rate becomes 3%, and the media cost is about 58.82 RON per order. The difference illustrates the budget's sensitivity to conversion. Neither 2% nor 3% are recommended standards for your store.
Hypothetical example 3: the full cost of the test
You allocate 4,000 RON for media, 1,200 RON for materials and 800 RON for setup and management. The total cost is 6,000 RON. At the end you have 60 retained orders, of which 45 come from new customers and 15 from existing customers, one for each person in this example.
The media cost per order is about 66.67 RON. The total cost of the test per order is 100 RON. If you allocate the whole test spend to the 45 new customers, it comes to about 133.33 RON per new customer. This last split is a declared convention, not a perfect causal attribution.
The materials can produce value after the test too. For the economic analysis of a future period you can discuss a different allocation, but the cash has been spent now. Keep in the report both the actual spend and the allocation convention. Don't hide the initial cost to make the launch look cheaper.
Choose products and channels you can support
Don't fund the whole catalogue at the start
Select products by stock, contribution, commercial advantage and clarity of presentation. A popular product isn't automatically the most suitable for the first test. It can have strong competition, high logistics costs or a differentiation that is hard to explain in the space of an ad.
Build a short list and note the reason for including each product. For example, one item can have stable stock and clear instructions, while another may first need photos that show its size. This selection is a commercial and operational decision, not just a campaign setting.
Tie the channel to the buying context
For products searched for actively, evaluating Google Ads campaigns can start from demand and the relevance of pages. For products explained visually, Facebook and Instagram campaigns can test demonstrations and arguments. In both cases you need an offer, a working site and measurement.
Don't split the budget equally just for symmetry. An initial distribution must reflect your questions and resources. If you don't have the capacity to prepare two channels properly, a focused test can be easier to interpret than two activities launched incompletely.
Check the estimates without turning them into promises
Planning tools can help build scenarios, but they don't remove uncertainty. Google explains the limits and the factors that influence Keyword Planner forecasts. Use the estimates to ask better questions, not to guarantee future revenue.
In an existing store, your own history can be more relevant than a generic average. But check whether the period compared had the same prices, products, stock and conditions. A campaign from a seasonal peak is not a neutral basis for all the following months.
The budget must also fit the collection calendar
Orders are not immediately available cash
A confirmed order can be cancelled, refused or returned. Collection can arrive later than the payment for advertising or the purchase of stock. If you plan only from the revenue shown in the platform, you can end up with a cash shortage even in a period with many orders.
Draw a simple calendar of the money: when you pay for goods, when you pay for ads, when you deliver and when you collect. Add the interval in which returns are clarified. You don't need a complicated financial model to notice that fast growth can require additional cash before it generates any available.
Don't fund repurchase from assumptions
Some stores accept a higher initial cost because customers come back. This approach makes sense only if there is data on repurchase, the contribution of later orders and the time until them. Estimated future revenue can't pay today's obligations.
Use cohorts, meaning groups of customers won in a shared period, to observe behaviour over time. Separate very old customers from recently attracted ones and don't assume all will follow the same path. A historical average can hide differences between products, offers and acquisition sources.
Keep a limit for the experiment
The test budget must be an amount you can sustain even if the commercial result is below expectations. That doesn't mean you plan for failure, but that you keep the freedom to decide based on data. If every day of advertising threatens payment to suppliers, the pressure can produce contradictory interventions.
Set a total limit and review points, without confusing review with an obligation to change things. At each point ask what you've learned, what has changed and whether the initial hypothesis is still viable. Continuing must be justified as clearly as stopping.
What to do when the budget is small and the data is limited
Reduce complexity before you reduce quality
A small budget can become hard to interpret if you distribute it across too many campaigns, products and variants. Choose fewer questions and prepare them well. You don't need to test all audiences, all messages and all pages at once to learn something useful.
You can prioritise one category and two distinct commercial arguments. Track whether visitors find the right product, whether they understand the offer and whether they can complete the order. If you discover a clear error, fix it; you don't need a huge volume to prove that a button doesn't work.
Avoid a verdict after a daily variation
With few orders, a single transaction can change the average cost a lot. Compare periods and volumes relevant to the buying cycle. Google describes conversion delay, which can make recent results look weaker before all purchases are recorded.
Waiting isn't the only option, though. Check intermediate signals such as the relevance of the products viewed, checkout errors and customers' questions. These observations don't replace orders, but they can explain why the test hasn't yet reached the desired result.
Don't confuse the volume of data with truth
A big report can be wrong if events are duplicated or the purchase value is sent incorrectly. Before discussing whether the volume is sufficient, check a test order and reconcile with the store. Transaction identifiers and consistent definitions are more important than the look of a dashboard.
Note separately what you know, what you assume and what you can't yet measure. This discipline prevents turning a correlation into a certain explanation. You can take a practical decision under uncertainty, but it is useful to keep its limits visible, together with the conditions that would prompt a review.
The budget sheet you can use for every test
| Component | The question to clarify | The evidence or document |
|---|---|---|
| Products | What selection do we promote and why? | Stock, contribution, pages |
| Media | What amount and what total limit do we have? | Approved plan and accounts |
| Creative materials | What arguments and formats do we prepare? | Brief and deliverables |
| Management | Who sets up and analyses? | Explicit responsibilities |
| Measurement | How do we count retained orders? | Test order and reconciliation |
| Cash | When do we pay and when do we collect? | Calendar of payments and collections |
| Review | What conditions justify the next step? | The test log |
Fill in the sheet together with the people who know the goods, the site and the finances. Marketing can't deduce on its own the cost of a return or a supplier's availability. When each piece of information has an owner, the budget conversation becomes faster and less dependent on assumptions.
Keep the initial version and the later changes. If the offer changes in the middle of the test, record the date and the reason. At the end you will be able to separate the evaluation of the initial hypothesis from the evaluation of the intervention, without retrospectively rewriting the goal to fit the result.
A work plan for the first 30 days
Week one: economics and preparation
Gather data on products, costs and stock. Choose the category, define the contribution and set the total budget. Check the site on mobile and completing an order, then prepare the materials. If essential information is missing, prioritise completing it before launch.
End the week with a short test sheet that everyone involved can explain. Don't turn the document into a list of technical terms. It must say what you are trying, how much you are risking, who tracks the results and what counts as a problem that calls for immediate intervention.
Week two: launch and operational checks
Track whether the ads lead to the right pages, whether the products are available and whether events are recorded. Compare spend with the plan and check any differences. Don't assume everything works just because the platform shows impressions and clicks.
Keep a log of errors and corrections. A stock problem can explain the drop in orders better than a bidding change. Such a log also helps the collaboration with the agency, because commercial information isn't always visible in the advertising account.
Week three: interpretation and a justified correction
Analyse results by product or category, within the volume available. Separate unsuitable traffic, a weak offer, page problems and a cost that is too high. Choose the intervention that answers the evidence, without changing all the components at once out of frustration.
If the data is too limited, formulate exactly what remains unknown. You can continue within a limited scope, modify the hypothesis or stop the test. None of these options is a failure of management if the decision is supported by the information and resources available.
Week four: the decision about the next budget
Pull together total spend, retained orders, contribution and new customers. Take into account orders still being clarified and the purchase delay. Present both the observed result and its limits, especially if the period includes promotions or important changes.
The next budget can be larger, equal, smaller or zero. The question is what decision makes sense now, not how you necessarily justify an increase. If you expand, also check stock, delivery and the capacity to produce new materials; advertising doesn't work separately from the rest of the store.
Mistakes that make the budget look simpler than it is
The first mistake is using a percentage of revenue without checking contribution. The percentage can be an internal benchmark, but two categories with the same turnover can have different economic room. A convenient rule becomes dangerous when it hides a change in the product mix.
The second is comparing management offers with different scopes. Ask for deliverables, responsibilities, access and the frequency of analysis. Don't evaluate the collaboration only by the number of campaigns or materials, because the volume of activity is not the same as the value of decisions.
The third is adding up conversions attributed by several platforms. The same order can appear in multiple reports. For the company's budget you need a shared commercial register and explanations of each channel's attribution, not a total that counts the same transaction twice.
The fourth is treating unspent money as an automatic lost opportunity. If there are no suitable products, clear materials or cash for delivery, keeping the reserve can be the right decision. The goal isn't to consume the plan, but to use resources in a way you can sustain.
The fifth is moving the budget at every fluctuation. Frequent interventions can make the test hard to interpret and can hide the real problem. Define analysis intervals, but keep exceptions for obvious errors. Discipline doesn't mean passivity; it means being able to explain every change.
Frequently asked questions about the promotion budget
Can I start with a very small amount?
You can build a limited test, but you must accept that the volume may be insufficient for conclusions about profitability. Concentrate the products and the questions, keep the preparation costs visible and don't interpret a few visits as the definitive evaluation of the channel. The amount must be related to your costs and commercial cycle.
Do I have to invest on all platforms?
No. Choose the channel that fits the hypothesis and the resources available. Being on many platforms at once can fragment the budget and the team's time. Add a channel when it can clarify a question or bring a distinct opportunity, with materials and measurement prepared.
How long must the test last?
The duration depends on the volume obtained, the purchase delay and the financial limits. A 30-day calendar is a practical way to organise, not a guarantee of sufficiency. Set reviews and stopping criteria, and keep conclusions proportional to the data observed.
Does the agency's fee go into the acquisition cost calculation?
It depends on the report's definition, which must be specified. For the complete perspective on the test, include the relevant management and production costs. You can track media cost and total cost separately, but don't compare metrics with different scopes as if they meant the same thing.
Can I use a competitor's budget as a benchmark?
Only as contextual information, if it is real and comparable. You don't automatically know the competitor's margin, retention, stock or goals. A large budget can support a different business model or can produce losses. Build the decision from your own store's data.
When do I increase the amount?
When the observed results, contribution and operational capacity support expansion, and measurement is credible. Check the effect of the additional money, not just the historical average. There is no universal growth percentage that automatically keeps the same efficiency in any account.
What do I do if no orders appear?
First check measurement and whether purchasing works, then the relevance of traffic and the offer. Evaluate the volume and the decision time before a verdict. If the test consumes the accepted limit without a viable explanation, stopping and reformulating it can be more useful than automatically adding more.
Checklist before you approve the budget
- I know which products and which hypothesis I'm funding.
- I have separated media, materials and management.
- I know the contribution and the relevant variable costs.
- I have tested purchasing and measurement.
- I have set the total limit and the review moments.
- I can sustain payments until orders are collected.
- I have an owner for data, stock and interventions.
- I know what result would change the initial decision.
The right budget isn't a spectacular amount, but one tied to the store's economics and a clear question. If you want to build this plan together with a specialist team, talk to BlueAds about products, costs and goals. The useful starting point is a test you can explain, fund and evaluate.