Definition
Return on investment is the gain from an investment minus its cost, divided by the cost. The gain is profit, not revenue: for marketing that means the gross margin on the sales attributed to the activity, after the cost of goods, returns, discounts and fulfilment. The cost is everything spent to produce that gain: media, agency and platform fees, creative production, tools, and the time of the people involved. A positive ROI means the activity returned more than it consumed; zero means it broke even; a negative figure means it lost money.
The formula is simple and the inputs are where the work is. Which margin, which attribution, which costs and which time window all have to be fixed before the number means anything, and two teams using different choices will report different ROI for the same campaign.
(Gain − Cost) ÷ Cost(%)
How to interpret it
Direction is "higher is better", with the caveat that a very high ROI on a tiny spend is less valuable to a business than a moderate ROI on a large one; ROI is a rate, and the absolute profit matters too. Comparing activities by ROI is fair only when they are measured the same way, over the same horizon and with the same attribution. A retention email programme and a prospecting campaign will never show comparable ROI because one works on customers already acquired and the other on strangers.
The main confusion is with ROASRevenue generated per unit of ad spend., which divides attributed revenue by media spend and ignores margin and non-media costs. ROAS can be high while ROI is negative whenever margins are thin or overheads are large, which the ROAS page explains in its own terms. The second trap is time. Most marketing spend produces some of its return after the reporting period ends, through repeat purchases and referrals, so an ROI measured on first orders alone understates it, while an ROI that counts projected lifetime value is only as good as the forecast. The marketing ROI route covers how to set up the measurement so the window is explicit.
What moves it
- Gross margin. The single largest lever; the same revenue at a higher margin is a higher gain.
- Cost completeness. Adding fees, salaries and tooling to media cost lowers ROI and makes it truthful.
- Attribution model. Platform-reported revenue, last-click revenue and incrementally measured revenue produce different gains.
- Time horizon. First-order, annual and lifetime windows give rising ROI for the same spend.
- Customer mix. New customers carry acquisition cost; returning customers mostly carry margin.
- Order value and purchase frequency, which raise the gain without raising cost.
- Returns, discounts and fulfilment, which quietly eat into the gain if left out.
The first four are choices made by whoever builds the measurement, which is why two ROI figures for one campaign can both be right; the last three are facts about the business that the measurement should capture.
Reading it in context
| Situation | What ROI shows | What to read beside it |
|---|---|---|
| High ROAS, low or negative ROI | Thin margin or heavy non-media cost | Product-level margin, cost breakdown |
| ROI positive on first orders | Activity pays back immediately | Repeat rate; it may be better than it looks |
| ROI negative on first orders, positive at lifetime | A payback period that needs cash | Acquisition cost payback months, cash position |
| ROI rising as spend falls | Cutting the least efficient spend | Whether absolute profit also fell |
| ROI incomparable across channels | Different attribution or windows | Standardise before comparing |
Same campaign, four ROI figures
A campaign spends £20,000 on media and £5,000 on agency fees and creative. Platform-attributed revenue is £100,000 at a 40% gross margin, so the gain is £40,000 and ROAS reads 5.0. Using media cost only, ROI is (40,000 − 20,000) ÷ 20,000 = 100%. Using full cost, ROI is (40,000 − 25,000) ÷ 25,000 = 60%. An incrementality test then shows only 70% of that revenue was caused by the ads, cutting the gain to £28,000 and full-cost ROI to 12%. Adding expected repeat margin of £9,000 from the new customers over the following year lifts it to 48%. Each figure is correct for its definition; only the last two inform the budget decision.
Related metrics
ROASRevenue generated per unit of ad spend. is the quick, media-only cousin of ROI and the figure platforms report; it is a useful efficiency signal and a poor profitability one. CACCost to acquire one new customer. and LTVTotal value a customer brings over time. are the customer-level version of the same question, comparing the cost of acquiring a customer with the margin that customer returns over time, which the LTV and CAC economics guide develops. AOVAverage revenue per order. moves the gain side of ROI directly, since a larger order at the same acquisition cost is more margin for the same investment. The ROAS calculator covers the revenue-based figure; ROI needs the margin and cost inputs added by hand, which is a feature rather than a limitation. Each of the four is a different view of the same question: whether the spend returned more than it cost.
Common mistakes
- Using revenue as the gain. That is ROAS, not ROI, and it flatters every low-margin product.
- Counting media cost only and reporting a figure that overstates what the finance team will recognise.
- Mixing windows across channels, so that lifetime-value ROI competes with first-order ROI for budget and wins unfairly.
- Reporting ROI without the attribution basis, which makes the number impossible to audit or reproduce.
- Maximising ROI rather than profit. The highest-ROI budget is often a small one that leaves profitable spend on the table.
- Forgetting the costs that arrive later, such as returns and customer service, which land after the campaign has been declared a success.
- Reading a single period's ROI as the campaign's verdict when the return arrives over several, especially for subscription and repeat-purchase businesses.
