How to determine ROI on a business?

The pressure point is simple. Most teams pitch momentum while the bank account tells a different story. ROI is supposed to bridge that gap. Instead, it often becomes a slide with pretty ratios that hide burn, custom work, and slow payback. If you want ROI to guide real decisions, you need to rebuild it from the operator’s side. That means cash first, time boxed, and tied to a counterfactual. Where ROI breaks in startups is not the formula. It is the framing. People talk about return like it is a single number. In early companies, ROI has layers. Capital ROI. Customer ROI. Project ROI. Founders who blend these or cherry pick the best one drift into false confidence. The antidote is a clear stack, consistent definitions, and a habit of reconciling slides with bank data.

Start with the numerator. If your return is revenue, you will over invest. If your return is gross profit, you will still miss hidden labor and tooling. The real numerator is net cash contribution after fully loaded delivery costs and the incremental people hours required to earn that cash. That includes support tickets, success calls, one off integrations, and discounts. If you do not have the tracking, you estimate with simple guardrails and tighten the estimate over two or three cycles. The goal is not a perfect model. The goal is to stop lying to yourself with a clean deck that ignores time.

Now fix the denominator. Capital in is obvious for a discrete project or asset. For a whole business, you are mapping invested capital that carries the work. That includes cash, vendor credit you rely on, founder time valued at what you would pay to replace it, and any equity dilution you would accept to fund the same work. If you leave time and dilution out, you will green light projects that only work because you are subsidizing them with your life and future cap table.

Set the clock. ROI without a period is just theater. Pick a period that matches your cash reality. For subscription models, 90 days is a clean diagnostic window and 12 months is a strategic one. For transactional models, work in cycles tied to your inventory or marketing recovery curve. If cash is tight, prioritize diagnostics that answer one question. Will this spend return net cash before I need to raise or borrow again.

Every good ROI frame also states the counterfactual. What would happen if you did nothing or deployed the same capital to the next best use. Without that, any return looks good in isolation. With it, most vanity projects die on contact.

With those basics set, use a three layer ROI stack that matches how a real company functions.

Layer one is project ROI. This is the return on a specific initiative like a paid channel test, a new sales hire, a feature priced as an add on, or a warehouse setup. Numerator is net cash contribution attributed to that initiative within the chosen period. Denominator is the fully loaded cost to run it. If you cannot attribute profits cleanly, attribute by lift over baseline. If baseline conversion was 2 percent and the test lifts it to 2.6 percent, the lift is the project effect. Tie that lift to cash, not just revenue, and you get a clean project ROI that beats story time.

Layer two is customer ROI. This is where many teams hallucinate. Deck math shows CAC and LTV. In practice, LTV is not a number. It is an assumption about retention and contribution that shifts with pricing, discounting, and support intensity. Use net cash contribution per customer by day 90 as your working metric. Then map how cohorts pay back by channel. If day 90 is negative for a channel, you do not have a growth engine. You have a financing need. This forces the hard choice. Either fix onboarding and pricing or cut the channel. When customer ROI improves, you will see it in bank reconciliations within a quarter.

Layer three is capital ROI. This is return on the capital base that supports your operating system. Think ROCE for a startup. Numerator is operating profit before founder comp and before interest, because you are testing business productivity not your financing choice. Denominator is invested capital in working assets. Hardware, prepaid tools, net working capital, build cost for core software, and any capitalized leases you cannot exit without pain. This frame tells you if the business creates value when it is not surfing promotions or short term tricks. When capital ROI is weak while customer ROI looks fine, you have system debt. Tools, people structure, or fulfillment design is swallowing your margin.

Watch for false positive metrics that routinely trick smart teams. One is revenue growth with soft gross margin. It feels like traction because charts go up. It is actually burn in disguise when discounts and custom work expand to win deals. Another is blended CAC that hides channel drift. If organic is covering paid sins, your blended view will look safe while your paid lines are eroding. A third is cohort LTV modeled on early adopter behavior. Early adopters are loyal and cheap to serve. Mainstream users are not. If your LTV depends on support you cannot scale, you are buying fake ROI with founder heroics.

Now apply a clean decision framework whenever you ask how to determine ROI on a business. Step one is define the unit. Are you deciding at the project level, the customer level, or the capital base. Step two is choose the period. If cash is tight, begin with day 90. If runway is healthy, add a 12 month lens. Step three is fix the numerator to net cash contribution after delivery and support. Step four is price time honestly. If the work needs founder hours to sustain, include a replacement wage or risk premium. Step five is state the counterfactual. What would the next best use of capital and hours return within the same window. Step six is apply a risk haircut. If inputs are noisy, shave 20 to 30 percent off the modeled return and see if it still clears your bar. Step seven is create an operational trigger. If the project is not tracking to payback by week six or milestone two, stop. ROI is a living test, not a tattoo.

Examples make this real. Imagine a sales led feature worth 20 dollars of monthly add on per account. You spend 15 thousand dollars to build it and 5 thousand dollars in launch collateral. Your success team expects 10 minutes per call for upsell. You have two success reps at 80 thousand dollars fully loaded each, handling 600 accounts. Ten minutes per call across 600 accounts is 100 hours. At 40 dollars per hour fully loaded, that is 4 thousand dollars of time cost. In the first 90 days, 300 accounts convert. Cash receipts are 18 thousand dollars. Net cash contribution is 18 thousand minus 4 thousand minus any incremental infra, say 1 thousand. That is 13 thousand. Denominator is 20 thousand. Project ROI at day 90 is negative 35 percent. Before you kill it, look at leading indicators. If conversion continues at the current slope and churn impact is neutral, you hit break even by month six. The trigger then becomes simple. If conversion rate does not hold for the next two months, pause upsell calls and rework pricing or packaging. This is how ROI becomes a throttle, not a post mortem.

Another example is a paid channel that brings new customers at 60 dollars CAC. Contribution per customer after delivery is 30 dollars by day 30 and 65 dollars by day 90. Day 90 customer ROI is positive 8 percent after adjusting for refunds and chargebacks. The deck will celebrate. The bank will not. Why. Because the working capital cycle is starved by the lag between spend and payback. If monthly spend is 100 thousand dollars, you are carrying about 200 thousand dollars of float across two cycles before cash returns. If you do not have a facility or runway to fund that float, the right ROI decision is to slow the channel until payback time is pulled forward or contribution improves. You can improve contribution by raising price on first purchase, bundling to lift order value, or cutting over servicing. You can improve payback time by fixing onboarding and pushing for earlier value in the user journey. ROI and cash timing are the same conversation. Treat them that way.

The most underrated ROI lens is time ROI. Founders pretend it is free because it feels noble to grind. Value your time at the replacement rate for a strong operator. If a project needs your continued presence to keep working, that time belongs in the denominator. A sales sprint that only closes when you join every call is not ROI positive unless you plan to be the sales rep forever. A feature that sells itself with no added support might be ROI positive even with modest revenue because it frees hours across the team. When you measure time, portfolios shift toward compounding systems and away from heroics.

Tie ROI to operating rhythms so it stays honest. Monthly, reconcile project ROI against actual cash movement. Quarterly, reconcile customer ROI against cohort behavior and support costs. Twice a year, look at capital ROI and compare it to the bar your investors would expect for similar risk. Publish the definitions so your team learns the same language. If marketing counts return as revenue and finance counts return as contribution, you will fight the wrong fights.

A useful rule is to keep ROI within your circle of control. If the result depends on a partner launching a feature or a policy change you cannot time, label it strategic and do not count it in your core ROI. Strategic bets can be worth it. They just should not crowd out bread and butter projects that keep the lights on. There is also a point where ROI loses the plot. Tiny optimizations can show great percentages while contributing little in absolute cash. Do not let small percent gains distract you from large cash wins with lower headline ROI. A pricing change that lifts contribution by two percent on a big base can beat a funnel tweak that lifts a micro metric by twenty percent. Run ROI in both percentage and absolute cash so you can rank projects by true impact.

The last step is governance. ROI without a stop rule becomes sunk cost worship. Set a maximum loss limit in time and money before a project must be re justified. Use pre agreed milestones like first value time, week two retention, or cost to serve per user. If a project misses two milestones in a row, you do not need another meeting. You need to stop or redesign. Most founders do not need more optimism. They need stricter exit criteria.

When someone asks you how to determine ROI on a business, the real answer is that you do not. You determine ROI on the decisions that create the business. You stack those decisions into a system that compounds cash and trust over time. You measure with the right numerator, the right denominator, the right clock, and a clear next best alternative. Then you act when the numbers tell you to act. Most teams do not need another deck. They need to fire their fake ROI and let cash and time make the call.