How to Calculate the ROI of a Software Purchase

Software ROI is the net return from a software purchase relative to its total cost of ownership over a stated period. Software is one of the investments where ROI is most often overstated, because vendors quote time savings at full value, licenses are the visible cost while implementation and internal time are not, and the benefits depend on adoption by people. A fair calculation takes the full cost, applies a realization factor to benefits and tests the result under pessimistic assumptions. This guide shows how to calculate software ROI, with a three-year worked example and checked arithmetic, covers cost categories and benefit types, explains realization and ramp-up, runs a sensitivity test and lists the errors that make software cases fail after purchase. It is part of business analytics for finance and operations, follows marketing ROI and ROI vs. ROAS, and leads to the ROI of analytics investments and a project ROI template. Software costs flow into budgeting and forecasting.

Total cost of ownership

The license price is only part of the cost. Include the following.

Ask the vendor for a full quote, including optional modules you will need, and check with a customer in a similar situation what the total spend was compared with the quote.

Types of benefit

1. Time savings. Hours saved times loaded cost. Valid only if the time is removed, avoided or redeployed to something valuable.
2. Avoided costs. A legacy system retired, a contractor not hired, a service no longer outsourced.
3. Revenue or margin improvement. Better pricing, higher conversion, fewer stockouts. Count at contribution margin and be careful with attribution.
4. Error and rework reduction. Fewer corrections, returns or write-offs.
5. Risk reduction. Fewer compliance breaches or outages. Estimate expected loss avoided, with explicit assumptions.
6. Strategic benefits. Scalability, data access, speed. Describe them, and leave them out of the number unless you can estimate them.

A worked three-year example

A company evaluates planning software (hypothetical). Costs:

• Year 1: license $30,000; implementation $25,000; internal time of 200 hours at $60, which is $12,000; training $5,000. Total $72,000.
• Years 2 and 3: license $30,000 plus administration of $6,000 each year. Total $36,000 per year.
• Three-year cost: $72,000 + 2 × $36,000 = $144,000.

Benefits at full run rate:

• Time saved: 1,500 hours × $55 loaded cost = $82,500 of potential saving. Management agrees that only 60 percent will convert to real value (less overtime, reduced temporary help and time redeployed to analysis), so the benefit is 0.60 × $82,500 = $49,500.
• Legacy tool retired: $14,000 a year.
• Fewer errors: $8,000 a year.
• Total at run rate: $49,500 + $14,000 + $8,000 = $71,500 a year.

The benefits take time to build as users adopt the system, so assume 50 percent in year 1 and 100 percent in years 2 and 3: year 1 is $35,750; years 2 and 3 are $71,500 each. Total benefits are $35,750 + $143,000 = $178,750.

• Net gain: $178,750 - $144,000 = $34,750
• ROI: $34,750 / $144,000 = 24.1 percent over three years
• Payback: the cumulative net position is -$750 at the end of year 2 and turns positive early in year 3, so payback is about 2.0 years.

A return of 24 percent over three years, equivalent to about 7.5 percent a year compounded, is not exciting for a project with execution risk. That is the kind of answer a good business case should reveal before the purchase, not after.

Sensitivity: what if realization is lower?

The weakest assumption is the 60 percent realization of time savings. Suppose it is 40 percent: the time benefit becomes 0.40 × $82,500 = $33,000, and the run-rate benefit falls to $33,000 + $14,000 + $8,000 = $55,000. Year 1 at 50 percent is $27,500, and years 2 and 3 are $55,000 each. Total benefits are $27,500 + $110,000 = $137,500. The net gain is $137,500 - $144,000 = -$6,500, and the ROI is -6,500 / 144,000 = -4.5 percent. The investment now loses money. This result tells the team where to focus: the project succeeds or fails on whether it can convert saved hours into real value, so the plan should say how, such as reducing overtime or taking over analysis work now outsourced.

Realization and ramp-up

Two adjustments separate good software cases from bad ones. The realization factor reflects the share of theoretical benefit that becomes real. For time savings, typical sources of leakage include savings spread thinly across many people, work that expands to fill the time and tasks that change in nature. The ramp-up factor reflects that adoption takes months. Both should be explicit in the model, with a rationale, and reviewed after launch.

Discounting

For multi-year software cases, a net present value calculation at the company's cost of capital gives a fairer comparison with other uses of funds. With a discount rate of 10 percent, the net flows in the example (-36,250, 35,500, 35,500) discount to -36,250 + 35,500 / 1.10 + 35,500 / 1.21 = -36,250 + 32,273 + 29,339 = $25,362. The NPV is lower than the undiscounted net gain of $34,750 because later benefits are worth less. NPV is still positive, so the base case clears a 10 percent hurdle, but only just in comparison to the risk.

Questions to ask the vendor

• What did a typical customer of my size pay in total, including services, in the first year?
• What is the typical time to full adoption?
• Which benefits come from the software and which from the changes in process that go with it?
• What are the renewal terms, and how much can prices increase?
• What does it cost to leave?

Common mistakes

1. Using the license cost as the whole cost.
2. Valuing every hour saved at full cost.
3. Assuming immediate adoption.
4. Omitting the administrator.
5. Counting benefits that come from process change as benefits of the tool, or vice versa.
6. Not measuring after go-live.
7. Ignoring renewal increases.

Renewal increases and contract terms

A three-year model should include price increases at renewal. Suppose the annual license of $30,000 rises by 7 percent each year. Year 2 becomes 30,000 × 1.07 = $32,100 and year 3 becomes 32,100 × 1.07 = $34,347. The extra cost over the flat assumption is (32,100 - 30,000) + (34,347 - 30,000) = 2,100 + 4,347 = $6,447. Total cost rises to $144,000 + $6,447 = $150,447, the net gain falls to $178,750 - $150,447 = $28,303, and the ROI falls to 28,303 / 150,447 = 18.8 percent. A modest increase takes more than five points off the return. Negotiate caps on increases at signing, and check the terms for user growth, extra modules and minimum commitments.

Compare against the alternatives

The right comparison is not software against nothing but software against the real alternatives. List them: do nothing, improve the current process with the tools you already have, buy a different product, or build something in-house. For each, estimate cost and benefit in the same way. Often the best option is a smaller purchase, or a better use of existing software. In many cases the benefit comes mostly from the process change that accompanies the software, and the software only enables it. If you can capture most of the benefit by redesigning the process in a spreadsheet, the case for the purchase rests on the remainder. See build vs. buy for analytics.

After go-live: protect the return

Most of the return from software is earned after the project team has moved on. Assign an owner for the benefits, not only for the implementation. Set adoption measures, such as active users, share of reports produced from the system and share of transactions processed through it. Check at three, six and twelve months whether the benefits assumed in the case have appeared: hours removed, legacy tool retired, errors reduced. If the retired tool is still running, the $14,000 benefit is not being earned. If adoption is low, find out why before blaming the software, since the cause is often training, unclear process or the absence of a decision that forces the change. Report actual ROI to the sponsor after a year, and keep the results for the next purchase.

FAQs

How do I calculate the ROI of software?

Add all costs over the period, estimate benefits with realization and ramp-up factors, then divide net benefit by total cost. Show payback and a pessimistic case.

What costs should I include?

Licenses, implementation, data migration, internal time, training, ongoing administration, infrastructure and exit costs.

How do I value time savings?

Hours saved times loaded hourly cost times the share that will convert into real savings or redeployed output.

What period should I use?

Three years is common for subscription software, matching typical contract terms and the period over which benefits can be forecast.

Is a positive ROI enough to buy?

Not alone. Consider payback, risk, strategic fit and what else the money could do.