Say you spend $100 on an ad and the report credits it with $240 in sales. A 2.4x ROAS looks like $2.40 back for every dollar spent, which makes adding budget tempting. But the product was not free, and neither was getting it to the buyer. What is actually left once those costs are paid? I would work that out before congratulating the creative.
Start with where that $100 went
These are hypothetical numbers to work through, not customer results: 10,000 impressions, 200 clicks of the selected type and $100 in spend. The report credits the ad with 4 purchases worth $240. The period, currency and calculation basis are consistent throughout.
| Metric | Calculation | Result |
|---|---|---|
| CTR | 200 / 10,000 x 100% | 2% |
| CPC | $100 / 200 | $0.50 |
| Purchase CPA | $100 / 4 | $25 |
| ROAS | $240 / $100 | 2.4x, or 240% |
| Purchases per selected click (exercise) | 4 / 200 x 100% | 2% |
You do not need to memorise the abbreviations yet. People saw the ad, some clicked, those clicks cost money, and each order ends up carrying $25 in advertising costs. The 2.4x ROAS describes revenue, not profit. The final row is the click-to-purchase ratio in this exercise, not necessarily the conversion rate your account calculates.
If you spend money but get no orders, do not enter zero for cost per order: there are no orders to divide by. With no clicks, this CPC cannot be calculated either. ROAS is zero when spend is positive and attributed revenue is zero.
After product and delivery costs, four orders leave $8
Continue the exercise. Assume $60 revenue per order after discounts, refunds and collected taxes. Product, payment fees, fulfilment and expected return-handling costs total $33. Use one currency throughout. Do not subtract the same refunded revenue twice if it has already been removed from sales.
| Where the money goes | This example |
|---|---|
| Left per order for ads and other bills | $60 - $33 = $27 |
| Left across four orders | $27 x 4 = $108 |
| After advertising | $108 - $100 = $8 |
That changes how 2.4x ROAS feels. All four orders leave just $8 before fixed staffing, software and other overheads. It is not $2.40 in profit per advertising dollar. It does not even establish a final profit.
To work out a reference point for your own product, take one more step: $27 is 45% of $60, so 45% of revenue is available to pay for advertising first. In this example, 1 / 45% is about 2.22x ROAS, where ads use all of that money. This assumes equal revenue and costs per order, with the report using the same net revenue basis.
Do not copy 2.22x into a bidding target. Fixed overheads remain unpaid, and crediting these orders to an ad does not prove they were all additional sales. Understand your costs first. 'They will buy again later' is not a reason to ignore a loss unless you have evidence of repeat purchases.
Back in your own report, read the column names
The numbers in an exercise arrive neatly arranged. Yours may not. Open the column descriptions first: what kind of click is being counted, and does a conversion actually mean an order?
CTR divides clicks by impressions: how often an ad display produces a click. An impression is not a unique person; someone may see the ad several times. Google Ads: Clickthrough rate (CTR)
CPC is the average cost of a click. Google Ads divides actual click charges by clicks, rather than using your maximum bid. When comparing ads, take costs and clicks from the same measurement basis. Google Ads: Average cost-per-click
With CPA, check what the action is. A purchase, a cart addition or an email signup may be configured as a conversion. Google Ads' Conversions are not necessarily purchases. Mixing carts with orders does not give you cost per order. Google Ads: Understand conversion tracking data
Purchase CPA is not CAC, the cost to acquire a new customer. If those four orders include existing customers or one person ordering twice, four purchases do not mean four new customers.
Think of attributed revenue as sales the platform credits to advertising, not the whole store's turnover. The same care applies to Meta reports: not every click means a store visit, and not every credited order is an additional order caused by the ad.
Change the creative, or fix the store first?
Replacing the image every time orders disappoint can send you in the wrong direction. I would follow the shopper: see the ad, open the page, decide to buy, complete payment. Find where that journey stalls before making changes.
| What you see | Check here first |
|---|---|
| CTR falls using the same measurement basis | Audience, placement and repeated exposure, then whether the creative explains the product |
| Clicks but few product-page visits | Whether clicks are website clicks, whether the page loads, and whether redirects and visit tracking work |
| Visits but few purchases | Price, shipping, whether the item matches the ad, plus payment and purchase tracking |
| Good ROAS but little money in the account | Valid orders, refunds, costs and when payments actually arrive |
Back to those four orders: an $8 remainder is thin, and there are few orders. I would not increase budget on that basis. If payments fail, pause the affected ads. If nothing is clearly broken but reporting is incomplete, observe within the spending boundary agreed beforehand and review at the planned time.
If repeated, comparable observations point to the message, change one main thing. Perhaps the shoe looks beautiful but gives no sense of how to wear it; try an everyday styling image. Changing the image, price and product page on the same day makes it much harder to know what helped.
You can stop at the agreed spending boundary without a clear result. Record it as inconclusive instead of forcing a winner. Note when you repaired checkout or changed the image, so a working store is not mistaken for a winning new creative.
Make sure the comparison itself isn't misleading
Someone who clicks today may buy a few days later. Google Ads notes that this delay can make recent cost per conversion look higher. Do not compare today's unfinished record directly with a month whose purchases have had time to arrive. Google Ads: Conversion delay
- Align dates, currency, time zone, click type, purchase event and attribution settings before comparing.
- An ad for new customers and a promotion for existing customers have different jobs. ROAS alone does not rank them fairly.
- If one or two orders can reverse the result, hold off declaring a winner. Ordinary ad sets can receive different spend and impressions; they are not controlled A/B experiments.
- Meta and Google may both credit themselves for the same order. Adding their revenue figures does not give you store revenue.
One more calculation to watch: do not average two click-through rates. Suppose one ad has 100 impressions and 10 clicks; another has 9,900 impressions and 99 clicks. Together, that is 109 clicks from 10,000 impressions, or 1.09% CTR, not the 5.5% average of 10% and 1%. For CPA and ROAS too, add up comparable underlying figures before dividing.
Answer these three questions next time
What did you spend? What does an order leave after costs? Where might shoppers be getting stuck? Answering these is more useful than memorising abbreviations. Use the record below alongside your report, and leave anything you cannot confirm marked as unknown.
Date range / time zone / currency: Platform / ad or creative ID: Goal and conversion event: Full click-column name: Impressions / clicks / spend: Purchases / attributed purchase revenue: Revenue treatment of tax, discounts and refunds: Attribution settings / reporting-delay notes: Pre-ad contribution and cost assumptions: Observation / conclusions not yet supported: One next check or change: Spending boundary / next review time:
Write a specific next step: not 'optimise ads', but 'check why shipping does not appear on mobile' or 'keep other conditions steady and try a clearer everyday-styling image'. That is when the report has helped you make a decision.
Key takeaways
- Check the definitions of clicks and conversions before comparing ads.
- Revenue-based ROAS is not a profit margin or proof of incremental sales.
- Diagnose the full journey; one attractive metric or a few orders cannot establish a winner.
Pick a product you sell and write down revenue per order, product costs, delivery and other order costs. Work out what is left to pay for advertising, then return to that ROAS.
Further reading
Sources checked:
