Two returns, two audiences
Return on ad spend (ROASRevenue generated per unit of ad spend.) is revenue attributed to advertising divided by the cost of that advertising. It is fast, available in every platform, calculable per campaign, ad group or product, and directly usable as a bidding target. Its audience is the people who run channels: they need a number that moves with their decisions, daily or weekly, and ROAS does.
Return on investment (ROIProfit relative to cost.) is the profit an investment produced, less its cost, divided by its cost. Applied to marketing, the profit is contribution after the variable costs of the sales, and the cost is everything marketing spent: media, agency fees, production, tools and, depending on convention, people. Its audience is finance and leadership: they need to know whether marketing, as a use of the company's money, made or lost money, and whether it did better than the alternatives.
The two numbers are linked but differ in almost every input. ROAS uses revenue; ROI uses profit. ROAS counts media only; ROI counts all costs. ROAS is usually read over days or weeks; ROI over months, quarters or longer. ROAS almost always uses a platform's attribution; ROI should use the business's own. The glossary entry for ROI and the glossary entry for ROAS give the short definitions; this page is about using them in reports. A platform's ROAS is its own version of the metric: ROAS as Google Ads reports it shows how much of that figure is set by the platform's conversion settings rather than by the business's revenue.
Compare table
ROI vs ROAS
| Row label | ROI | ROAS |
|---|---|---|
| Formula in words | Profit from marketing less marketing cost, divided by marketing cost | Attributed revenue divided by media cost |
| Numerator | Contribution after variable costs of the sales | Revenue, often platform conversion value |
| Costs included | All marketing costs: media, fees, production, tools, sometimes people | Media only |
| Margin and overhead | Built in through contribution margin; overhead by convention | Ignored; must be compared with a margin-based target |
| Time horizon | Monthly, quarterly or longer; can include lifetime effects | Daily to weekly; usually short attribution windows |
| Attribution | Business's agreed standard, ideally tested for incrementality | Usually the platform's own model |
| Granularity | Programme, channel or total marketing | Campaign, ad group, product, even keyword |
| Primary audience | Finance, leadership, board | Channel managers, agencies, bidding algorithms |
| Main use | Judging whether marketing makes money and comparing investments | Optimising and bidding within a channel |
| Where it misleads | When long-term or brand effects fall outside the horizon; when attribution is weak | When margin varies, costs outside media are large, or platforms over-credit |
| Reconciles to the ledger | Yes, if built from ledger costs and agreed margin | Rarely, since platform revenue and spend differ from finance's |
Where ROAS is the right number
ROAS is the right number inside a channel. Bidding systems in the major ad platforms optimise to a return target expressed as conversion value over spend, and the only way to steer them is in that language. Channel managers deciding between campaigns, products or audiences within one platform need a metric available at that level of detail and updated as quickly as their decisions. ROAS meets both needs.
It is also right for comparing similar things on the same basis: two campaigns selling the same products, under the same attribution model, with the same margin. Within those limits, a higher ROAS genuinely means a better use of the media.
What makes ROAS safe is the target it is compared with. A target derived from contribution marginRevenue left after every cost that varies with each sale., the break-even return on ad spend (Break-even ROASMinimum ROAS to avoid a loss.) plus a profit buffer, turns a revenue ratio into a proxy for profit. The derivation lives in the contribution margin and break-even guide, and the break-even calculator does the arithmetic. A ROAS reported against a generic or historical target, by contrast, says nothing reliable about profit.
ROAS fails as soon as it leaves the channel. Summed across platforms, it double-counts conversions that several channels touched. Sent to finance, it is read as a profit ratio, which it is not. Compared across products with different margins, it ranks the wrong things first. The ROAS calculator is useful for checking a figure, but the figure is only meaningful against the right target.
Where ROI is the right number
ROI is the right number whenever the question is whether marketing, or a major part of it, makes money for the business. That is the question finance asks at every review, leadership asks at budget time and boards ask when growth slows.
ROI is also right for comparing marketing with other uses of money. Finance compares investments across the business: a new hire, a system, a warehouse, a marketing programme. Only a return on the full cost, measured in profit, can be set beside those alternatives.
Built well, ROI reconciles to the ledger. Its cost comes from the profit and loss account, including fees, production and the costs that sit outside the marketing line such as commissions and discounts; the marketing P&L guide explains where those sit. Its profit comes from contribution margin agreed with finance. Its attribution uses the business's own standard, ideally calibrated by incrementality tests. Customer acquisition cost (CACCost to acquire one new customer.) on the same full-cost basis is its natural companion.
ROI misleads in two ways. If the horizon is too short, it penalises spend whose effects arrive later, such as brand advertising and content; the honest response is to report those effects over a longer horizon, not to exclude them. And if attribution is weak, ROI inherits the weakness while looking more authoritative than ROAS. The analytics root's marketing ROI page covers the measurement side.
Reconciling the two in one report
The two metrics do not need to compete. A single monthly report can carry both, provided it shows how they connect.
Start with media spend and attributed revenue, giving ROAS by channel against each channel's margin-based target. Then add a bridge. Convert attributed revenue to contribution using the agreed margin. Reconcile attributed revenue to the company's recognised revenue and show the gap. Add non-media marketing costs from the ledger. The result is ROI on contribution with all costs, and every step between the two numbers is visible. For the Google Ads line, the tutorial on reconciling conversions with sales records builds that revenue bridge from the order system, net of returns and cancellations. If the same report splits ROI by channel, the non-media costs have to be shared out first, and the guide to allocating shared marketing costs explains why that split should be shown beside the direct costs and never used to decide where spend goes.
A KPI tree is a useful way to lay this out: ROAS and channel measures at the branches, contribution and ROI at the trunk. Publish both definitions in one metrics dictionary, with the formula, the source of each input and the attribution model, so neither team can quietly change them. Agreeing which return figure each audience sees is one line in the written split of ownership described in the guide to working with a finance business partner.
Send the right section to the right meeting. Channel reviews read the ROAS section; finance reviews read the bridge and the ROI. Everyone can see both, and nobody mistakes one for the other. The finance reporting hub covers the wider reporting relationship.
One quarter reported both ways
Kestrel Garden, an online garden supplies retailer, spent £120,000 on media in a quarter, with £540,000 of attributed revenue: a ROAS of 4.5 against a margin-based target of 3.8, so channel teams were on target. The bridge for finance applied the agreed 32% contribution margin, giving £172,800 of contribution. Reconciling attributed revenue to the ledger removed £60,000 of double-counted revenue, reducing contribution to £153,600. Non-media marketing costs from the ledger, agency fees, production and tools, came to £38,000, so total marketing cost was £158,000. ROI on contribution was (£153,600 − £158,000) ÷ £158,000, slightly negative. Media alone was profitable; the fees and production pushed the programme below break-even, which became the focus of the next budget review.
For a plan whose objective is online orders, the online sales objective hub sets contribution after ad cost as the outcome and reads ROAS against break-even ROAS for each channel.
When channel ratios beat their targets while profit falls, the diagnosis of high ROAS but not profitable rebuilds the month from the ledger and finds whether the target, returns, over-crediting, capture spend or missing costs opened the gap.
If the dispute is less about which return to use than about whether finance believes any of marketing's figures, the page on finance distrusting marketing numbers works through ledger reconciliation, overlapping attribution, incrementality evidence and stable definitions.
Common mistakes
- Sending ROAS to finance as a measure of profit. Finance reads any return as profit; a revenue ratio over media alone will be misunderstood or dismissed.
- Setting ROAS targets without margin. A target copied from elsewhere may approve loss-making spend or block profitable spend.
- Summing ROAS across platforms. Overlapping attribution double-counts conversions and inflates the total.
- Calculating ROI on revenue. Revenue less cost divided by cost is not a return on investment; profit must come from contribution.
- Leaving fees and production out of ROI. Media-only cost makes ROI look like an inflated ROAS.
- Changing definitions between reports. Without one dictionary, the same name means different things and both teams stop trusting the numbers.
- Judging brand spend on short-horizon ROI. Effects that arrive over many months need a longer reporting window, not exclusion from the calculation.
