What does ROI mean? Platform notes and checklist

What is ROI? A ratio that is only as trustworthy as the return you attribute

For advertising, Google Ads describes ROI using profit relative to costs and notes that the exact calculation depends on the business goal. Its physical-product example subtracts product and advertising costs, then divides the resulting net profit by the total cost basis shown in that example. A media-only calculation can instead define advertising spend as the investment, but that is a different scope.

The formula is easy. The difficult part is the word return. If the numerator includes revenue or profit that the investment did not actually cause, the ratio can look precise while answering the wrong question.

ROI versus ROAS

Return on ad spend (ROAS) commonly compares attributed revenue with advertising spend: ROAS = attributed revenue / ad spend With $8,000 attributed revenue and $2,000 spend: ROAS = 4.0x (400%) That does not tell you product cost, fulfillment cost, discounts, or whether the ads caused the sales. ROI is broader and usually requires a profit/return definition. Do not label a revenue multiple "ROI" merely because both are ratios.

The attribution problem

Imagine a repeat customer receives an email, sees a social ad, searches the brand on Google, and buys. Which investment gets the return? Analytics systems use attribution rules to assign credit. Google Analytics explicitly treats attribution models and lookback windows as configuration that affects how credit is distributed across eligible touchpoints. Changing the model can change reported channel return without changing any customer behavior.

This creates a crucial distinction:

• Attributed ROI asks how a measurement model assigns observed outcomes
• Incremental ROI asks how much additional outcome the investment actually caused compared with what would have happened without it

The second is a causal question and generally requires a stronger experimental or quasi-experimental design.

The time-window problem

ROI can be made to look better or worse by changing the measurement horizon. Suppose a retention program costs $10,000 to launch. In the first 30 days it generates only $6,000 of incremental gross profit, but over 12 months it produces $30,000. A 30-day calculation and a 12-month calculation answer different questions. Neither is automatically wrong if the time window is declared. Always state:

• investment period
• return observation period
• whether future cash flows are included
• whether repeat purchases are included
• how attribution expires

The cost-boundary problem

What counts as investment? For an advertising campaign, possibilities include:

• media spend
• agency fees
• creative production
• discounts or incentives
• software/tooling
• internal labor
• incremental fulfillment/service cost

A "campaign ROI" using only media spend may be perfectly useful for media optimization, but it is not the same as a fully loaded business-case ROI. Name the cost boundary.

The return-boundary problem

Likewise, return can mean:

• revenue
• gross profit
• contribution profit
• operating profit
• lifetime gross profit
• cost savings
• a monetized nonfinancial outcome

For ecommerce marketing, gross profit or contribution profit is often more informative than raw revenue when margins differ significantly across products.

A campaign comparison where revenue lies

Campaign A:

• spend: $5,000
• attributed revenue: $20,000
• gross margin on sold mix: 25%
• gross profit before ads: $5,000

Campaign B:

• spend: $5,000
• attributed revenue: $14,000
• gross margin: 60%
• gross profit before ads: $8,400

Campaign A has higher ROAS (4.0x vs 2.8x) but, under this simplified gross-profit view, contributes no profit after ad spend while Campaign B leaves $3,400. The example shows why product economics belong in ROI conversations.

ROI cannot prove causality by itself

A high ratio can reflect:

• strong targeting
• existing customer demand
• brand search that would have converted anyway
• seasonality
• inventory changes
• organic word of mouth
• a measurement model that gives the channel generous credit

The ratio summarizes the inputs you give it. It does not validate those inputs. Where the stakes justify it, use holdouts, randomized experiments, geo tests, conversion-lift methods, or other incrementality approaches to estimate what changed because of the investment.

ROI for non-marketing investments

The same discipline applies beyond ads. For an email automation project, investment may include implementation labor and software cost. Return may include incremental gross profit, support-cost savings, or both. If you mix one-time implementation cost with monthly return, specify the evaluation horizon.

For inventory equipment, return may include labor savings and reduced errors over several years. A simple ROI percentage ignores the timing of cash flows; capital budgeting may require measures such as payback period, NPV, or IRR instead. ROI is useful, but not universal.

A practical ROI specification

Before reporting the percentage, write 5 lines:

If those 5 lines are unclear, the decimal places in the ROI are false precision.

The formula

**ROI = (return - investment cost) / investment cost × 100%**

Worked example

Suppose an ecommerce campaign spends $2,000 and generates $8,000 of tracked revenue.

• Calling this "300% ROI" by doing (8,000 - 2,000) / 2,000 treats revenue as if it were profit before the investment. But suppose the sold products have $4,000 of COGS

• Under a simplified model:

• revenue = $8,000
• COGS = $4,000
• campaign cost = $2,000
• remaining profit before other expenses = $2,000

• If the comparison treats the $2,000 ad spend as the investment, ROI on this simplified profit basis is:

($4,000 gross profit - $2,000 ad spend) / $2,000 = 100% The correct numerator depends on what decision the business is making, but revenue divided by spend is not the same metric as ROI.

Common mistakes

1. Revenue over ad spend is not the same as profit-based return on investment
2. 2 campaigns with identical revenue can have very different profit
3. Attribution allocates credit; it does not automatically establish causality
4. Longer horizons can capture more repeat revenue and change the result
5. A narrow media ROI can be useful, but label it honestly

Questions we get asked