ROI (Return on Investment): from the 1919 DuPont formula to its real limits in marketing 2026

ROI —Return on Investment— measures how much profit an investment generates relative to its cost. The basic formula is simple:

ROI = (Profit obtained - Investment) / Investment × 100

If you invest €10,000 and recover €15,000 in attributable profit, the ROI is 50%. If you lose the investment, it is negative. If you only recover what you invested, it is 0%.

That simplicity is why it is one of the most widely used financial concepts in any company. And it is also why it is one of the most misused. The formula hides several implicit decisions —what counts as investment, what counts as profit, over what time horizon, with what attribution— that, when made poorly, produce numbers that look rigorous and are fiction. It pays to know what ROI really says and what it does not before making big decisions based on it.

The origin: DuPont, 1919, and Donaldson Brown

The modern ROI formula was popularized by the work of Donaldson Brown at DuPont around 1919-1920. Brown was an electrical engineer turned finance executive, and he developed what is now known as DuPont analysis or the DuPont identity: a decomposition of ROI into diagnostic components.

Brown's intellectual contribution was not inventing the idea of a return/cost ratio —which was already known— but showing that ROI can be decomposed to understand why it is worth what it is worth:

ROI = Net margin × Asset turnover

Or, in more detail:

ROI = (Net profit / Sales) × (Sales / Assets)

The operational usefulness of this decomposition is that it lets you see where the return originates: is it because you sell with a high margin on each unit? Or because you turn over assets quickly? Two companies can have the same ROI with completely different profiles —one with high margin and low turnover (luxury), another with low margin and high turnover (mass consumer goods). The conclusion about what to do in each case is radically different.

Brown brought this logic to General Motors when he joined as CFO in 1921, where his measurement methods became an industry standard. The DuPont formula is still taught in business schools a century later and appears in the financial reports of publicly traded companies even today.

Knowing this matters because most trivial uses of ROI —"this campaign had a 200% ROI"— ignore the diagnostic dimension Brown introduced. ROI used well is not just a grade; it is a tool for understanding the origin of the return.

The five real limitations of ROI

ROI has structural problems as a decision metric that any serious user must keep in mind:

It is blind to time. A 50% ROI achieved in one year is not comparable to a 50% ROI achieved over five years. The metric does not distinguish them. For serious financial decisions, it pays to move to metrics that do include time: IRR (Internal Rate of Return), NPV (Net Present Value), payback period.

It is blind to risk. An investment with an expected ROI of 30% but high risk (a significant probability of total loss) is not equivalent to one with a 30% ROI and low risk. The formula does not integrate probabilities. Deciding on ROI alone overlooks risk dimensions that can be decisive.

It is blind to intangibles. A campaign that builds brand, retention, or authority generates value that does not show up in the numerator in the short term. If you only measure a campaign's ROI by directly attributable sales, you under-record the real value of brand or relationship investments.

It is susceptible to attribution problems. Especially in marketing. If a customer discovers the brand through SEO, sees an ad, receives an email, reads a case study, and buys weeks later after a sales call —which action does the revenue get attributed to? Platforms and attribution models give very different answers, and they are all approximations.

It is manipulable depending on what you count as investment. If you exclude indirect costs (overhead, proportional salaries, tools, opportunity cost), ROI looks better than it is. If you include them generously, it looks worse. The definition of "total investment" is not objective in many situations.

Those five limitations do not disqualify ROI; they place it. It is a useful metric when you understand what it measures and what it does not.

ROAS vs. ROI: the distinction that confuses many

In digital marketing it is common to confuse ROI with ROAS (Return on Ad Spend). They are not the same:

ROAS = Attributable revenue / Advertising cost
ROI = (Profit - Total investment) / Total investment

The practical differences are several:

Revenue vs. profit. ROAS uses gross revenue; ROI uses profit (after costs). A campaign with a 4x ROAS looks great —but if cost of goods sold is 60%, operating costs are 25%, and the additional advertising cost is 15%, the real profit may be zero or negative.

Advertising cost vs. total investment. ROAS only counts ad spend; ROI counts all the associated investment (creative, management, tools, proportional salaries). For a brand that invests heavily in creative production, the difference between ROAS and full ROI can be dramatic.

Horizon. ROAS typically measures the short term (the specific campaign); ROI can measure longer horizons.

The practical rule: ROAS is a useful tactical metric for optimizing campaigns; ROI is a strategic metric for deciding whether an entire operation or channel is profitable. Confusing the two leads to companies with seemingly excellent ROAS and questionable real profitability.

Measuring marketing ROI honestly in 2026

Measuring marketing ROI has entered a difficult phase. The two main reasons, already covered in other articles but relevant here:

Attribution has become opaque. The degradation of third-party tracking, iOS App Tracking Transparency since 2021, and progressive browser restrictions have drastically reduced the accuracy of cross-channel attribution. Multi-touch attribution models that were standard in 2018 today produce results with high uncertainty.

The classic models are coming back. Marketing Mix Modeling (MMM) —which compares investment across channels with aggregate results over time— has resurfaced. Incrementality testing (turning off a channel and measuring what changes) has become standard for serious advertisers. More statistical, more expensive to maintain, more honest.

For credible marketing ROI measurements in 2026, you need to combine:

• Platform attribution (with its known bias) for tactical optimization.
• Marketing Mix Modeling to understand the real weight of each channel at an aggregate level.
• Incrementality tests to validate that the channel really contributes.
• Cohort analysis to understand retention and LTV by source.

An ROI measurement based only on what Meta or Google reports is an approximation that tends to be optimistic. Combining it with MMM and incrementality is what distinguishes serious growth teams from those who confuse vanity metrics with results.

How to use ROI without falling into its traps

Three practices that distinguish useful use from decorative use:

Be explicit about numerator and denominator. Document what counts as investment and what counts as return before calculating. What seems obvious rarely is when examined in detail.

Compare ROI only between comparable things. Same time horizon, same definition of investment, same attribution methodology. Comparing the ROI of SEO vs. paid social as if they were equivalent numbers is often a mistake because their horizons and attributions are very different.

Combine ROI with complementary metrics. Especially CAC and LTV, payback period, margins, and cohort analysis. A decision based on ROI alone is usually incomplete.

Common mistakes in using ROI

Confusing ROI with margin. A 30% margin is the proportion of each unit of revenue that remains as profit. A 30% ROI is the proportion of return on investment. They are different metrics, and mixing them produces wrong conclusions.

Calculating ROI with inappropriate horizons. SEO takes 6-18 months to produce significant traffic (covered in how long a blog takes to rank). Calculating SEO ROI at month 3 produces terrible numbers that do not reflect the structural reality.

Not including all costs. The "ROI of the last campaign" usually excludes internal team time, opportunity cost, tools, management. When these are included, ROI drops significantly.

Optimistic attribution without a control. Accepting what the platform reports as attributable conversions without independent validation systematically overestimates ROI.

Using ROI to decide everything. Some investments (legal, security, initial branding, infrastructure) have ROI that is hard to measure directly but are necessary. Applying the metric where it does not fit produces dangerous under-investment decisions.

ROI and creative operations

For an agency or in-house team, a substantial part of the total investment that goes into the ROI calculation is creative production: ads, landing pages, videos, case studies, content, sales materials. When that production is managed in silos with no reuse between campaigns, the cost of each new piece is high and aggregate ROI erodes structurally.

That is why the connection with creative operations is direct: the editorial calendar lets you sequence production to reuse prior work, content production scales variants with a consistent brand system, and creative KPIs measure not just direct ROI but compound ROI across several campaigns.

At Polimake, that logic lives across three surfaces: Studio to coordinate campaigns that take advantage of prior investment, Studio to produce variants from reusable masters, Media as a repository where case studies, templates, and prior assets are accessible —so the tenth campaign has a better ROI than the first through reuse, not luck.

---

If you manage a marketing budget, finance, or any kind of investment and you got here looking for an answer about ROI, the most useful thing you can take away is probably the most sober: ROI is a useful but partial metric. It works well when you understand its five limitations (time, risk, intangibles, attribution, scope) and combine it with complementary metrics. It works poorly when treated as the single answer that decides whether something is worthwhile.

To complement, CAC as a diagnostic covers how acquisition cost is interpreted, LTV covers the necessary counterweight, and the conversion funnel covers how metrics connect to the sales operation.

Quick references

• CAC as a diagnostic — the most relevant complementary metric.
• LTV — the counterweight to CAC.
• Conversion funnel — how ROI connects with the sales operation.
• Cohort analysis — for ROI analysis by customer group.
• How long a blog takes to rank — to understand why measuring SEO ROI in the short term is misleading.
• Direct advertising — where the difference between ROAS and ROI is operationally most visible.