ROAS (return on ad spend) tells you how many times over your ad spend came back as revenue; ROI tells you whether anything was left as profit once every cost is accounted for. CTR measures how compelling the ad is, and conversion rate measures how persuasive the page and the offer are. As a business owner, no single one of these is enough: always compare ROAS with your own margin and your own order data, or a great-looking number can hide a loss-making campaign.
As an agency, we see the same problem in monthly reports over and over: lots of numbers, but no clear answer to whether the business is doing better. Ad platforms measure their own performance by their own rules. That isn't malice, it's simply how they work. Your job is to ask for a report you can actually make business decisions from.
- ROAS = revenue attributed to ads ÷ ad spend. It measures revenue, not profit.
- Your break-even ROAS comes from your margin: 1 ÷ contribution margin. Anything below it loses money, however good it sounds.
- ROI shows the full picture: cost of goods, shipping, returns and the agency's work are all included.
- CTR and conversion rate are diagnostic metrics, not business goals.
- Each month, don't ask for platform reports; ask for a summary reconciled with your own order data, split by branded vs. non-branded and new vs. returning customers.
What exactly does ROAS measure, and how is it different from ROI?
ROAS sees the world through the ad platform's eyes: it divides the revenue attributed to ads by ad spend. A ROAS of 4 means four times the spend came back as revenue. Google Ads also expresses this as a percentage: according to its help page on Target ROAS bidding, the system tries to keep conversion value per cost at the target you set, and in their example 500% means a fivefold return. Agree with your agency on which format they use, because "400" and "4" mean the same thing.
The catch is that revenue isn't your money. It has to cover the product, shipping, packaging and payment processing, and returned orders disappear from it too. That's why we calculate a break-even ROAS for every client: 1 divided by the contribution margin. A simple example: if a quarter of the revenue on an average order is left as contribution margin, break-even ROAS is 4. A campaign at ROAS 3 burns money on every sale; one at 6 makes money. The same ROAS of 4 is excellent for a high-margin service and barely breaks even for low-margin electronics.
ROI, by contrast, counts every cost: beyond ad spend, the agency's work, creative production and software. This is the number owners care about, but it's rarely possible to calculate precisely at campaign level, because every cost has to be allocated.
| Metric | What it measures | When it misleads |
|---|---|---|
| ROAS | Ad-attributed revenue relative to spend | When you don't compare it with margin, or when it mostly comes from branded search and returning customers |
| ROI | Profit relative to total investment | When some costs (agency, returns) are left out |
| CTR | Clicks relative to impressions | When clicks don't lead to purchases or enquiries |
| Conversion rate | Conversions relative to visits or clicks | When you don't know what counts as a conversion, or what the denominator is |
When is a good ROAS misleading?
We run into four situations again and again.
When it "wins" customers who would have bought anyway
People searching for your brand name already know you, and returning customers often click an ad purely out of convenience. These campaigns naturally show high ROAS, but the real incremental gain is small. That's why we always ask for branded and non-branded campaigns separately, and for new and returning customers separately.
When several platforms claim the same purchase
Google Ads, Meta and your email tool each apply their own attribution rules, and each tends to credit the order to its own ad. Add up the revenue the platforms report and you'll often get more than what actually came into your shop. We cover this in more detail in our article on attribution.
When tracking has gaps, or conversions arrive late
Without consent, part of the data is lost from the start (our piece on Consent Mode v2 explains the effect), and purchases often happen days after the click. Google accounts for this itself: according to the same help page, its Target ROAS recommendation excludes the last few days, because some conversions come in later. A report pulled at the start of the month will therefore undercount the last week of the previous one.
When revenue isn't the same as paid, kept orders
The ad platform sees the order being placed. It doesn't see whether it was later paid, cancelled or returned. In fashion, or with cash on delivery, this gap can change the picture substantially.
What do CTR and conversion rate tell you, and what don't they?
By Google Ads' definition, CTR is clicks divided by impressions: 5 clicks from 100 impressions is a 5% CTR. Google also stresses on the same page that a good CTR depends on what you're advertising and on which networks. So there's no universal "good CTR", and a high CTR isn't a goal either. An attention-grabbing ad that overpromises brings lots of clicks and few sales. We use CTR to check whether the ad and the search intent actually match.
For conversion rate, ask two questions:
- What counts as a conversion? If newsletter sign-ups and add-to-carts count too, the rate looks great but says little about sales. Ask for the primary conversion to be the real business event: a successful payment or a qualified enquiry.
- What's the denominator? The ad account counts per click, analytics per session or user. The two numbers never match, and they don't need to; you just need to know which one you're looking at.
If CTR is fine but conversion rate is weak, the problem usually isn't the ad, it's the site: slow loading, a muddled offer, a long checkout. In that case, start with site speed and the purchase flow rather than fine-tuning campaigns.
What should you ask your agency for every month?
We consider this one-page summary the minimum:
- Total ad spend and revenue calculated from your own system. Their ratio is your blended return, which can't be inflated by platforms double-counting.
- ROAS per campaign compared with break-even, split into branded and non-branded.
- The share of new vs. returning customers among ad-driven orders.
- Tracking health: how big the gap is between the platforms and your order database, and whether it changed since last month.
- What they changed, and why. One sentence of reasoning next to every significant change.
- Next month's experiment, and how you'll know whether it worked.
- What they can't measure. An agency that states this openly usually takes the other numbers seriously too.
And one question worth asking after every report: "If we switched this campaign off next month, how much revenue would we actually lose?" If the only answer is the platform's number, they don't know yet.
How do we handle it?
For us, tracking gets fixed before campaigns start. We begin with a technical audit, check that conversions come from the real business event, and reconcile reporting with the webshop or CRM data. We treat platform numbers as directional signals and calculate blended return from your own database. We often automate the report too, so the start of the month isn't spent copying spreadsheets.
Frequently asked questions
What counts as a good ROAS?
There's no universal good value. Your own break-even ROAS (1 ÷ contribution margin) is the benchmark: above it you make money, below it you lose. When acquiring new customers, a lower figure can be a deliberate choice if those customers come back later.
Should I look at ROAS or ROI as a business owner?
Both, for different purposes. ROAS is good for day-to-day and weekly decisions between campaigns; ROI tells you whether marketing pays off overall. Calculate full ROI at least once a quarter.
Why is revenue different in Google Ads, Meta and my webshop?
Because each attributes purchases differently, with different time windows, and missing consent or blockers mean none of them sees everything. The reliable number is your own order database. Platform numbers are for trends and for comparing campaigns.
How often should I look at the numbers?
As a business owner, monthly, together with your agency. Daily swings are often just noise caused by late conversions, and intervening too often also disrupts automated bidding.
What is blended ROAS or MER?
Total revenue divided by total marketing spend, regardless of channel. It won't tell you which campaign works, but it can't be inflated either, which makes it a good sanity check next to the platform reports.
If your monthly reports don't make it clear whether advertising pays off, we're happy to review your tracking and campaigns. You can read more on our marketing services page.