In short
- You calculate ROAS (return on ad spend) by dividing the conversion value of your ads by what they cost. €5,000 in revenue on €1,000 of ad spend is a ROAS of 5, or 500%.
- ROAS ignores your margin, and it cannot tell a new customer from one who would have bought anyway. A healthy-looking ROAS can hide a loss.
- Steer on margin after ad spend, cost per new customer against customer value, and one figure across all channels as well. Give every number in your dashboard a decision it supports.
The ROAS formula is simple: revenue attributed to your ads, divided by the cost of those ads. Whether you can manage by it is another matter. That only works if you know your margin and how much of the revenue the ads actually caused. Without those two, a campaign with an impressive ROAS can still lose money.
So ROAS is a useful number for tuning campaigns within a single channel. To split budget across channels, or to see whether marketing makes a profit, you need a few more.
What is ROAS and how do you calculate it?
ROAS stands for return on ad spend: what you get back for each euro spent on advertising. Google Ads uses conversion value for this, for example the revenue from an order. Its help article on Target ROAS bidding gives a worked example: £5 in sales for every £1 of ad spend is a target ROAS of 500%.
ROAS = conversion value / ad spend
ROAS shows up either as a ratio (5) or a percentage (500%). Agree once what goes into conversion value, such as VAT included or excluded, before or after returns, and stick to it.
A worked example: ROAS and break-even ROAS
This example is fictional. An online shop sells homeware. In one month:
- ad spend: €10,000
- conversion value reported by the ad platforms: €50,000 (400 orders at an average of €125)
- gross margin: 40% of revenue
ROAS is €50,000 divided by €10,000, so 5, or 500%. At a 40% margin, though, €50,000 of revenue leaves €20,000 of gross profit. Take off the ad spend and €10,000 remains for everything else: staff, warehouse, software.
Your break-even ROAS is 1 divided by your margin: 1 / 0.40 = 2.5. Below 2.5, every euro this shop spends on ads brings back less than a euro of gross margin. At a 20% margin the break-even point is 5, and the same ROAS of 5 means you only just cover your ad costs.
Why does ROAS mislead when it is your only measure?
Margins differ by product. ROAS treats €100 of revenue on a 10% margin product the same as €100 on a 60% margin product. A campaign that mostly sells keenly priced bestsellers can post a high ROAS and earn very little.
New and returning customers get mixed up. A loyal customer who searches for your brand name and clicks an ad counts the same as someone buying for the first time. You would probably have had that loyal customer without the ad. Google Ads recognises the difference itself: customer lifecycle goals let you bid higher for new customers, provided you tell it who your existing customers are through a customer list or website tag.
Retargeting harvests demand that already existed. Retargeting shows ads to people who have already visited your site, close to the moment of purchase, so these campaigns often post the best ROAS. Suppose that in the example retargeting costs €2,000 and is credited with €24,000 of revenue (ROAS 12), while campaigns aimed at new audiences cost €8,000 for €26,000 (ROAS 3.25). Move budget towards the highest ROAS and you end up paying more and more for people who were about to buy anyway.
Attribution assigns credit, not cause. An attribution model decides how much credit each ad interaction gets for a conversion. Google Ads now offers data-driven and last-click attribution (Google Ads Help). Google and Meta can also both claim the same order. To find out whether a campaign actually adds sales, you need an incrementality test: an experiment comparing a group that sees the campaign with a group that does not. Think with Google describes attribution as helpful for day-to-day optimisation, but harder to rely on alone for budget decisions.
Which marketing KPIs are better for steering?
Four measures fill the gaps. Here they are, using the same fictional shop.
POAS (profit on ad spend). Gross profit divided by ad spend: €20,000 / €10,000 = 2. Above 1, sales cover their own ad costs before any other expenses. Google Ads can report gross profit if you send cart data with your conversions and add the cost of goods sold to your Merchant Center feed (Google Ads Help).
Cost per new customer. Ad spend divided by the number of new customers. If 160 of the 400 orders come from new customers, each one cost €10,000 / 160 = €62.50. For a sharper figure, count only the campaigns aimed at new audiences.
CAC against customer value. CAC (customer acquisition cost) is what it costs in marketing and sales to win a new customer. Customer value, or customer lifetime value (CLV), is the margin a customer brings in over a fixed period. In this example we only count ad spend, so CAC is €62.50. The first order earns €50 of margin (40% of €125), which is less than the customer cost. If the average new customer orders 2.5 times within two years, their value is €125 of margin, twice the CAC.
MER or blended ROAS. MER (marketing efficiency ratio) is total revenue divided by total marketing spend, across every channel. Blended ROAS is the same idea, usually with ad spend only. Neither depends on attribution. If the shop took €120,000 that month, including organic and direct traffic, and spent €15,000 on marketing, its MER is 8. If you raise the budget and total revenue stays flat, the extra spend is probably not doing much, however good the platform ROAS looks.
How do you build a marketing dashboard that supports decisions?
A report tells you what happened. A dashboard you can steer by shows, for each number, what you would do about it.
- Start with the decisions. Which choices do you make every week or month? Shifting budget, pausing a campaign, giving a product range more room. Only include numbers that feed one of those choices.
- Take revenue and margin from your own systems. Orders, returns and cost prices come from your shop platform or ERP, ad spend from the platforms. Your dashboard revenue then matches your accounts.
- Separate new from returning. Define a new customer, for instance a first order on an email address or customer number, and report new customers and retargeting separately.
- Give every number a threshold and an owner. If a campaign's POAS stays below 1 for two weeks, the campaign owner decides whether to scale it back.
| Measure | Question it answers |
|---|---|
| Margin after ad spend | Do we make money on what we advertise? |
| Cost per new customer | What does growth cost? |
| CAC against customer value | Can we pay more per customer? |
| MER | Does extra budget bring extra revenue? |
| ROAS per campaign | Which campaign needs adjusting? |
If you use AI to summarise reports, a clear KPI model helps there too: a language model can only summarise well what has been defined clearly. That is part of what you practise in our AI for marketers workshop.
What does this look like in practice?
At an e-commerce company, data from several channels never came together. We built a KPI model around the cost per new customer, separated new customers from retargeting, and restructured the collaboration so briefing and execution stay strictly apart.
At a direct marketing organisation active in several countries, all the customer and sales data was in house, but there were no figures to steer on. We built the ETL process (extracting, cleaning and loading the data), the data models and a secured BI portal covering campaigns, customer journeys and returns. Most of the work went into that foundation. Within weeks the first employees were working in the portal, and staff now use it every day.
When is ROAS good enough?
ROAS works reasonably well with one channel, similar margins across products and almost only new customers. Even then, work out your break-even ROAS and steer on that.
If you sell B2B with a long sales cycle, the value only shows up later in your CRM. Look at cost per qualified lead and the revenue those leads produce. With a small budget, an incrementality test often yields too little data. Follow MER over a longer period instead, and avoid changing everything at once.
Want a KPI model or dashboard that steers on margin and new customers? Get in touch and we will look at what data you already have.
Sources
- Google Ads Help, About Target ROAS bidding
- Google Ads Help, About customer lifecycle goals
- Google Ads Help, About attribution models
- Google Ads Help, About conversions with cart data
- Think with Google, Use incrementality testing for effective marketing measurement (October 2023)
