Set the budget from contribution and affordable loss. Keep purchase CPA, new-customer CAC and ROAS distinct, allow conversion lag and scale only after the economics hold.
You will create: A three-scenario budget and a written review decision.
See the full advertising learning pathA useful ad budget pays for a decision your store can afford. No universal spend level guarantees sales. Model your economics first, then replace assumptions with observed customer outcomes.
Build the first-order model
Illustrative order: $100 net revenue and $40 contribution before ads. At $25 media CPA, $15 remains before fixed overhead and creative costs. ROAS is attributed net revenue divided by media spend: 100 / 25 = 4x. First-order break-even ROAS is 1 / 0.40 = 2.5x before overhead. Use expected returns, variable fulfillment and payment fees in the contribution calculation.
Turn assumptions into a capped test
If assumed CPC is $1 and purchase conversion rate is 4%, implied CPA is $25. A $750 media plan would imply 750 clicks and 30 purchases only if the assumptions hold. Actual outcomes can differ substantially. Thirty purchases do not guarantee an adequately powered comparison. Add production, tools and creator fees separately and choose a maximum loss you can sustain.
Define the experiment
Write one hypothesis, a primary outcome, comparison, review window and minimum worthwhile improvement. Change one major factor such as the proof angle, rather than audience, offer and page together. Prefer randomized experiments where supported. Ordinary delivery may distribute exposure unevenly and confound creative comparisons.
Read the right measures
Track CTR and video attention to diagnose the message. Track purchase conversion to diagnose the journey. Track new-customer acquisition cost and contribution after ads to evaluate the business. Reconcile refunds and orders, align currencies and windows, and avoid adding the same order across platform dashboards. Attributed ROAS is not proof of incrementality.
Pause, revise or scale
Pause broken links, wrong prices, unavailable products and invalid tracking promptly. At the planned review point, account for conversion lag and uncertainty. If the sample is small, label the test inconclusive. A spend multiple of target CPA is a review trigger you can choose, not a universal statistical rule.
Scale when profitable outcomes repeat, marginal acquisition cost remains acceptable, and inventory, support and cash can absorb demand. Small budget increases are an operating choice, not a guaranteed platform rule. Review new versus existing buyers, returns and contribution after each change. Protect the economics rather than maximizing a dashboard score.
See how sensitive your plan is
Use three scenarios before committing to spend. The following model holds media spend at $750, CPC at $1, net revenue per order at $100 and pre-ad contribution at $40. All conversions are assumed to be purchases; these numbers are illustrations, not forecasts.
| Purchase rate | Expected orders | Media CPA | Revenue ROAS | Contribution after media |
|---|---|---|---|---|
| 1% | 7.5 | $100 | 1x | -$450 |
| 2% | 15 | $50 | 2x | -$150 |
| 4% | 30 | $25 | 4x | $450 |
Fractional orders are mathematical expected values. Production, licensing and overhead are excluded here and must be deducted separately. The table shows why a small conversion-rate assumption can change the decision dramatically. Use the existing order economics calculator to check your own margin inputs.
Use precise denominators
Purchase CPA equals media spend divided by purchases. Media CAC equals media spend divided by new customers attributed under your chosen method. Fully loaded acquisition cost includes the additional acquisition expenses you define and allocate. ROAS uses attributed revenue, while contribution after ads uses revenue less the relevant variable costs and advertising. Do not switch denominators halfway through a comparison.
For example, 30 purchases may include 10 repeat buyers. At $750 media spend, purchase CPA is $25, but media CAC for 20 new customers is $37.50. Keep returning-customer revenue visible without treating those buyers as newly acquired.
Translate the budget into platform controls
Google's average daily budget is not necessarily a same-day ceiling. For most campaigns, its published limits allow up to twice that daily amount on a day and use 30.4 times the average daily budget for the monthly limit, with special handling when budgets change. Review the budget report and the applicable setup before assuming a $25 daily setting guarantees exactly $25 of daily spend.
Choose the available scheduling and spending controls that fit your loss ceiling, and monitor actual cumulative spend. A spreadsheet ceiling does not itself stop a campaign. Assign an owner and a review cadence that can catch overspend before it matters.
Prewrite the review decision
| Finding | Action | Evidence to record |
|---|---|---|
| Wrong price, broken checkout or invalid tracking | Pause and repair | Successful end-to-end verification |
| Relevant traffic but too few mature purchases | Continue within the cap or call inconclusive | Spend, conversion lag and remaining uncertainty |
| Repeatable orders but poor contribution | Revise offer, cost or acquisition approach | Economics by product and customer type |
| Healthy contribution across completed review windows | Test a measured increase | Marginal cost, stock, returns and cash |
Choose the window before seeing results and avoid changing several major variables together. An early favorable result can reverse. Compare absolute customer counts and costs as well as percentage changes, and explain the uncertainty when sample sizes are small.
Put this guide to work
Open the advertising campaign workbook, complete the section for this lesson, and record one decision you will make next. Keep your assumptions beside the evidence so you can update the plan when results arrive.
Questions store owners ask
What ROAS is good for ecommerce?
The required ROAS depends on contribution, acquisition costs, refunds and the business objective. In the illustrative model, a 40% pre-ad contribution margin implies 2.5x break-even revenue ROAS before overhead.
When should I stop an ad with no sales?
Stop technical failures promptly. For a functioning campaign, use your pre-agreed loss ceiling, relevance, conversion lag and evidence. A small number of clicks without purchases is not enough by itself to prove failure.
Put it into practice
Sources
Sources checked October 14, 2026. Platform screens and fees change; confirm current details in your own account.
Related guides
- The Ecommerce Profit Calculator: What Do You Keep From Each Order?
Calculate order contribution, allowable acquisition cost, and break-even ROAS with a transparent worked example.
3 min read - My Store Has a 4× ROAS. Why Am I Still Losing Money?
See how product costs, fulfillment, returns, overhead, and attribution can turn attractive ROAS into weak profit.
3 min read - Why Your Ad Platform, Email Platform, and Store Report Different Revenue
Understand attribution windows, overlapping credit, revenue definitions, and a practical process for reconciling reports.
3 min read
Want help applying this to your store?
Take the free store assessment for a suggested next step, or book a 15-minute call to talk it through.