Google Ads ROAS: Calculate Break-Even and Improve It

ROAS is a ratio, not profit

Return on ad spend is conversion value attributed to ads ÷ advertising cost. If Google Ads reports £24,000 conversion value from £6,000 spend, reported ROAS is 4.0, or 400%. That says nothing by itself about product cost, fulfilment, returns, agency fees, overhead or whether the sales were incremental.

Use three views together: platform ROAS for campaign optimisation; contribution after advertising for unit economics; and blended business performance to detect channel attribution or demand-capture effects.

Calculate your break-even ROAS

For a hypothetical £100 order with £45 cost of goods, £8 fulfilment and payment cost, and £7 expected return/discount allowance, contribution before ads is £40. Break-even revenue ROAS is £100 ÷ £40 = 250%. If the business requires £10 contribution after ads, allowable ad cost falls to £30 and required ROAS rises to 333%.

Use weighted contribution across the actual product mix, or pass margin-informed values when possible. A single account target can misallocate spend if one category contributes 60% of revenue and another 15%.

For lead generation, build return from the CRM

ROAS based on arbitrary lead values is decorative. Suppose 100 leads cost £5,000. Forty become qualified, ten close, and each new customer contributes £1,200. Attributed contribution is £12,000, so contribution ROAS is 240%. Raw lead CPA is £50, but customer acquisition cost is £500. Both are useful; neither replaces the other.

Import qualified or won outcomes so Google can learn from quality. Keep the assumptions visible: close rate, lag, cancellation and value horizon.

Audit the numerator before optimising it

• Reconcile order IDs and values with the commerce or CRM system.
• Check tax, shipping, discounts, currency, refunds and duplicate actions.
• Separate brand from non-brand and new from returning customers where observable.
• Compare attribution windows and conversion dates before reconciling Ads with GA4.
• Inspect conversion lag before judging recent periods.

GA4’s attribution paths report can show touchpoint sequences and time to key event, but attributed credit is still a model. It does not prove that every credited sale was caused by advertising.

Diagnose a ROAS change as a decomposition

For ecommerce, ROAS can be expressed as conversion rate × average order value ÷ CPC. If conversion rate falls 20% while CPC and order value are unchanged, ROAS falls roughly 20%. Break the change into those components before touching bids.

A defensible improvement sequence

1. Repair conversion count and value errors.
2. Remove clearly irrelevant demand and unavailable products.
3. Split reporting by brand, acquisition, margin and customer type.
4. Improve the constraint with the largest evidenced loss: query, feed, offer, page or follow-up.
5. Change bidding only after its optimisation signal reflects the desired outcome.
6. Validate uncertain changes with an experiment where traffic permits.

Google distinguishes Maximise Conversion Value from Target ROAS in its bidding guidance. A higher target is not a free efficiency improvement; it can reduce reach. Start from representative realised performance and commercial requirements, then monitor both return and volume.

Maintain four clearly labelled return views

Do not switch between these without labels. Platform revenue ROAS of 500% and contribution return of 180% can both be correct. A board cash decision and a bid strategy need different views.

Include repeat value without fantasy

Build cohorts by first purchase month and source. Measure realised contribution at 30, 90, 180 and 365 days, net of refunds and service cost. Use mature cohorts to estimate a conservative forward value and state the horizon.

Hypothetically, a first order produces £20 contribution and 40% of customers make one later order worth £18. Expected 12-month contribution is £27.20 before discounting and overhead. If repeat behaviour varies by product or offer, one global multiplier misprices acquisition.

Attributed return is not incremental return

Brand Search, remarketing and returning-customer campaigns can capture demand that may have converted anyway. Attribution assigns credit; incrementality asks what would not have happened without ads. The latter requires an experiment or credible counterfactual.

Where a suitable experiment is available, define the incremental outcome and guardrails in advance. Do not subtract all brand revenue, but do not call every attributed order incremental either.

Segment return by economics

Join item IDs to margin, return probability and customer type. A high-revenue campaign may buy discounted, frequently returned products. For leads, join campaigns to won contribution and sales cycle. The conversion tracking guide covers importing those outcomes.

Review distribution, not only average. Median contribution, product concentration and share from brand or repeat customers reveal fragility. A 400% ROAS driven by one bulk order should not automatically set next month’s target.

Monthly ROAS review

1. Reconcile spend, outcomes and values.
2. Mark incomplete cohorts.
3. Report platform, contribution, new-customer and blended return.
4. Decompose movement into CPC, conversion rate, value and mix.
5. List campaign, site, price, stock and attribution changes.
6. State actions, mechanisms and validation dates.

“ROAS was 4.2” is an observation. A useful conclusion is to scale a segment, repair a signal, test a hypothesis or accept lower short-term return for a defined growth objective.

Set a Target ROAS from the value definition

If Google receives gross revenue, the target must absorb variable costs. If it receives contribution value, the number changes. Suppose revenue is £150, non-ad variable cost £90 and required post-ad contribution £20. Allowable ad cost is £40; required revenue ROAS is 375%. If Google instead receives £60 contribution value, the equivalent target is 150%. Never transfer a target between those definitions.

Common failure modes

• Raising a target to express ambition and unintentionally suppressing volume.
• Sharing a portfolio target across incompatible value definitions.
• Judging recent periods before delayed conversions arrive.
• Reporting attribution-window changes as media improvement.
• Ignoring margin, returns and brand or repeat-customer mix.

Run a sensitivity table

Model low, base and high gross margin, return allowance and repeat contribution. Calculate break-even and required ROAS in each case. If the campaign is profitable only under optimistic assumptions, the plan is fragile. Finance should own the assumptions and refresh them when price, mix or fulfilment changes.

ROAS decision record

Whenever a target changes, record the value definition, economic calculation, historical mature return, expected volume trade-off, affected campaigns and review date. Include the finance owner who approved assumptions. This prevents a target surviving after margin, pricing or customer mix changes.

Review both achieved return and absolute contribution. A campaign can hit a higher ROAS while creating less total contribution because spend and revenue collapsed. Optimisation requires the ratio and the scale.

Build a platform-to-finance bridge

Reconcile monthly with finance. The bridge need not make platform and accounts identical; it must explain material differences.

Lead-value weighting example

Assume a raw lead has a 50% chance of qualifying, a qualified lead has a 30% chance of closing, and a customer contributes £2,000. Expected raw-lead value is £300. If campaign segments have different qualification or close rates, import or assign segment-specific downstream values rather than a universal £300 placeholder.

Refresh weights from mature cohorts. If the sales team changes qualification or pricing, old values can steer bidding toward yesterday’s funnel.

Scenario planning before scale

Forecast the next budget tranche under lower conversion rate, higher CPC and weaker order mix. Suppose current £20,000 spend returns £80,000 revenue at 400% ROAS. If the next £10,000 produces £30,000, marginal ROAS is 300% and blended ROAS becomes 366.7%. Whether scale is correct depends on the marginal contribution requirement, not the fact that the original campaign achieved 400%.

Also model operational constraints. Extra volume can increase fulfilment cost, returns or sales response time, lowering conversion value. Give finance and operations the assumptions before changing targets or budget.

ROAS audit questions

• Which conversion value definition is used?
• Does it reconcile with orders or CRM?
• What variable costs and refunds are excluded?
• How much value is brand, repeat or delayed?
• What is the marginal return at current scale?
• Which assumption would change the decision?

Keep the calculation sheet with the monthly report. When assumptions change, restate historical comparisons or clearly mark the break rather than presenting unlike ROAS definitions as one trend.

Sources

• Google Ads Help: choose a bid strategy
• Google Analytics Help: attribution paths