How to measure the real return on your ad campaigns

The basic formula is: conversion value or revenue attributed to the ads ÷ ad spend. You can present the result as a multiple or a percentage. 75,000 on a spend of 25,000 is 3x, or 300%. Make sure the currency, the period, and the data source match, and know whether the value includes tax, shipping, or orders that were later cancelled. Comparing a total sales figure against a partial spend produces a flattering indicator that is useless for a decision.

• ROAS = revenue attributed to the ad ÷ cost of the ad.
• A ROAS of 1x means revenue equals spend, before any other cost.
• A ROAS of 4x means four pounds of revenue per pound of advertising — not four pounds of profit.
• Compare channels on a consistent conversion definition, attribution window, and value.

Suppose a store spent 20,000 EGP in a month and the platform attributed 80,000 EGP of sales to the campaign. Platform ROAS is 4x. If cost of goods is 40% of revenue, and shipping, payment, and returns are 15%, a gross contribution of 45% remains before advertising — that is 36,000 EGP. After deducting the 20,000 spend, 16,000 is left before fixed costs. The campaign looks healthy. But had the gross margin been only 20%, the contribution would have been 16,000 — less than the spend — even with ROAS still sitting at 4x.

Do not ask whether ROAS is high. Ask whether ROAS is above the break-even point for your profit model, after cancellations and returns.

Start from contribution margin before advertising: revenue minus cost of goods, payment fees, subsidised shipping, packaging, returns, and the other variable costs. If 25% of every pound of sales is left to cover advertising and fixed costs, your approximate break-even ROAS is 1 ÷ 0.25 = 4x. If 40% is left, it is 2.5x. Use a realistic average from your own data rather than your best product alone, and recalculate when prices, discounts, or the return rate change.

1. 1Work out the average net order value after discount.
2. 2Subtract cost of goods and the variable costs per order.
3. 3Divide the remaining contribution by revenue to get the margin rate.
4. 4Divide 1 by the margin rate to estimate break-even ROAS.
5. 5Add a safety margin, management cost, and profit targets when you set the actual goal.

ROAS compares the revenue attributed to the advertising with the cost of that advertising. ROI compares net profit, or total return, with the total investment as your accounting defines it. CAC measures the cost of acquiring a new customer and may include advertising, sales, tooling, and the agency. A campaign with a lower ROAS can be the more useful one if it acquires customers who buy repeatedly, while a retargeting campaign looks strong precisely because it claims credit for sales from customers who were already close to buying.

Use the indicators together: ROAS to manage the channel and the campaign, CAC to understand the cost of growth, LTV for customer value over time, and ROI or profit contribution for the investment decision. Do not blend revenue with lifetime value on undocumented assumptions; start from actual purchases, then build the LTV model from repeat rate, retention, and margin.

Every platform uses its own attribution window and its own rules for deciding which ad gets the credit. A customer may see a Meta ad, then search the brand name, click a Google ad, and buy; depending on the setup, both platforms may claim the conversion. A return or cancellation may never reach the platform, a purchase event may fire twice when someone refreshes the thank-you page, or a value may be sent in the wrong currency. So compare the platform against your order system and GA4, and understand the difference rather than forcing the numbers into exact agreement.

• Review the click and view windows for each platform.
• Exclude cancelled orders and send back return values where the setup supports it.
• Use transaction_id to prevent duplicate purchases.
• Verify currency, tax, shipping, and discounts.
• Separate new customers from returning ones when the goal is acquisition.

Conventional ROAS measures attributed revenue according to the attribution model. Incremental ROAS tries to measure the additional revenue that would not have happened without the advertising. Google explains in its Conversion Lift measurement that iROAS equals incremental revenue divided by total spend. It needs an experiment or a control group and sufficient volume, but it answers a deeper question: did the campaign create new sales, or did it capture an order that was going to happen anyway?

The Target ROAS strategy relies on the conversion values you send to Google Ads, and it tries to maximise value while reaching the target average. Do not start it on incomplete values or unstable events, and do not set a target far above historical performance and then expect steady volume. Make sure the purchase or the lead carries a value that reflects the business, that the campaign receives enough data, and watch the average over a period rather than a single day.

Start with tracking, the offer, and the landing page before you narrow the audience. Improve data quality and values, separate products with different margins, and review search terms, messaging, speed, and checkout. Test one meaningful variable at a time, and give the test an appropriate volume and duration. Lifting ROAS by switching off every exploratory campaign may improve the report temporarily, but it chokes off new customers and increases your dependence on retargeting.

1. 1Fix tracking, values, and returns.
2. 2Set the break-even ROAS for each category or margin band.
3. 3Improve the offer, the page, and the conversion rate.
4. 4Move budget towards groups with proven value.
5. 5Keep a portion for testing new audiences and messages.
6. 6Scale gradually and watch CAC and profit alongside ROAS.

Bring together in one table the spend, the attributed revenue, the orders, the net orders, the new customers, the margin, the platform ROAS, the adjusted ROAS, and CAC. Compare against the previous week and a four-week average, and log any change to budget, offer, or site. A daily decision off a small fluctuation disrupts learning, while waiting a full month can waste budget. Pick a cadence that suits your data volume, and give every decision a clear condition for review.