Calculating the ROI of customized software: without fantasy figures
A guide for DACH B2B service providers who want to build a resilient investment case before a partner writes a line of code
Table of Contents
The ROI of customized software can be calculated. The problem is that most calculations work with fictitious inputs. Not always dishonest, but structural: a development partner prepares an offer. The offer must show a positive return. So the increase in productivity is estimated at 20 percent, the reduction in errors at 15 percent, the increase in turnover at 10 percent. The figures are plausible. However, they are not derived from anything in your real business.
The result is an investment case that looks rigorous and is not. It won't stand up to scrutiny by a CFO who asks where the 20 percent came from. It is useless as a reference after the go-live if you want to know if the system actually worked. And it doesn't answer the really relevant question: not "What could be gained?" but "What does the current operational infrastructure actually cost, and what minimum improvement is needed to justify the investment?" This guide shows how to calculate the ROI of custom software and digital transformation with numbers that are anchored in your own operations before a partner writes a line of code.
Why most ROI calculations for individual software are misleading
The standard ROI formula is simple: total benefits minus total costs, divided by total costs, as a percentage. The formula is not the problem. The inputs are.
Most ROI models are built from the benefit side. A partner estimates what the system will improve, assigns percentages to those improvements and multiplies them by sales or headcount to calculate a return on investment. The problem is that percentages applied to an unmeasured baseline produce numbers that seem precise but are not. A 20 percent increase in productivity means nothing if you don't know how many hours per week your team is currently spending on the process you want to improve. Likewise, the error and rework costs of the current situation are almost always missing, as well as the overall cost of doing nothing: the coordination hours, the SaaS subscriptions, the workarounds, the key person dependencies and the orders that are missed because processes are too slow.
What faulty ROI models use | What credible ROI models use |
"Estimated increase in productivity: 20%" | Hours measured per job, per person, before build |
"Assumed error reduction: 15%" | Number of errors and costs from the last 12 months |
"Forecasted increase in sales: 10%" | Specifically named orders, lost due to process speed |
Maintenance costs: not included | Maintenance: 15 to 20% of construction costs per year, explicitly budgeted |
Costs of the actual state: not modeled | Cost of doing nothing: measured and documented before the offer |
Start with the cost of doing nothing
Before a single benefit can be calculated, the costs of the current situation must be documented. This is the figure that makes a reliable investment case possible in the first place. It is also the figure that surprises most companies.
The cost of doing nothing is not the same as the cost of your current tools. It's the full operational cost of operating on the current infrastructure: the labor absorbed by coordination, the lost contribution margin due to pricing inconsistencies, the revenue that doesn't materialize due to slow processes, and the risk borne by dependence on a few key people.
Cost category | How to measure | Where the data is located |
Manual coordination effort | Hours per order x costs/hour x annual volume | Time records, team interviews, payroll accounting |
Error and reworking costs | Number of faults x average rectification costs x frequency | Complaints, credit notes, operating log |
Delayed or lost orders | Offers not sent x average order value x completion rate | CRM, sales log, pipeline data |
Key person dependency | Days lost per personnel change x daily turnover | Personnel files, company investigation report |
Current tool and workaround costs | SaaS subscriptions + time spent on workarounds | Finances, IT subscriptions, team time log |
A regional freight forwarding company conducted this analysis for the first time and added up three items. Manual coordination in the quotation and scheduling process took up 22 staff hours per week; at a full-cost rate of 50 EUR and 48 working weeks, this amounted to approximately 53,000 EUR per year. Errors and rework on order documents cost an additional 34,000 EUR per year in corrections and customer credits. And three to four tenders per year were lost to faster competitors solely due to slow response times, amounting to approximately 40,000 EUR given an average contribution margin of 10,000 EUR per order. The costs of the current situation totaled 127,000 EUR per year. The commissioned system cost 88,000 EUR to set up and 16,000 EUR annually for maintenance. The business case was thus clear, without a single estimated percentage.
Counterexample: When the numbers don't support the build
Not every analysis results in a positive investment case, and a credible model must reflect that as well. A B2B service provider with €12 million in revenue ran the same calculation. Manual coordination took up 6 staff hours per week, or about €14,000 per year. Errors and rework cost about €7,000, and lost orders were difficult to quantify. The costs of the current state amounted to around €24,000 per year.
A custom-built core operational system would have cost at least €75,000 to set up and €14,000 annually in maintenance. Even if the system eliminates 70 percent of the actual costs, the realized benefit amounts to approximately 16,800 EUR per year. Over three years, costs of 121,000 EUR are offset by benefits of around 50,000 EUR. The investment case is negative, and the payback period exceeds any reasonable timeframe.
The right recommendation here isn’t a custom-built solution, but rather a better-configured SaaS solution or the targeted partial automation of the most expensive individual process. That’s exactly what reliable data tells you: it also clearly indicates when you shouldn’t invest.
The real costs of the investment: what needs to be included
Credible ROI models for custom software must include four cost categories, not just one.
1. build costs
The investment in development is the starting point. For a B2B service provider with revenue in the range of 10 to 50 million EUR, a well-defined core operational system typically costs between 75,000 and 150,000 EUR, depending on the depth of integration, data structure, and AI requirements.
2. maintenance costs
A production system requires ongoing maintenance: bug fixes, dependency updates, interface maintenance, and a small backlog of improvements. Budget 15 to 20 percent of the original development costs annually. For a €100,000 system, that amounts to €15,000 to €20,000 per year. Over a three-year evaluation period, this is a significant cost item that significantly alters the ROI result if it is omitted.
3. Internal time and transition buffer
Your team will invest time during the build: in workshops, testing, process customization. For a 90- to 120-day implementation, 80 to 160 hours of internal operations team time is realistic, i.e. 4,000 to 8,000 EUR. You should also plan for a short drop in productivity in the first two to four weeks after the go-live, experience has shown that throughput is reduced by 10 to 15 percent during this phase. Both items are one-off, easy to estimate and significantly lower than expected if the system was introduced carefully and the team was involved in good time.
Measuring benefits without inventing numbers
Every benefit category in a credible ROI model needs a source: a number from your business, not an industry benchmark.
Time savings: Measure actual time first
Identify the process to be improved. Measure how long it currently takes. Multiply by volume and cost. If your dispatching process today takes 45 minutes per order, involves 30 orders per week and is handled by one employee at a full cost rate of EUR 45 per hour, that's EUR 1,012 per week. If the new system is reduced to 10 minutes, the weekly saving is EUR 787, or EUR 40,900 per year. Record this measurement before the build and repeat it six months after the go-live. The difference is your actual realized benefit, not a projection.
Error costs: First count, then extrapolate
Pull the last 12 months of customer credits, rework orders and corrections from your accounting or operations log. Assign a cost center to each: the direct remediation cost plus the time spent. Add it all up. If your business generates EUR 28,000 per year in measurable error costs and the new system eliminates 70 percent of that through structured processes and automated checks, the benefit is EUR 19,600 per year.
Sales impact: name orders, do not project a percentage
Sales increases are the most tempting input to puff up and the most difficult to defend. The credible approach names specific cases instead of projecting a percentage: How many bids per month are late because the current process can't keep up? What is the average order value of an offer that is sent late or not sent at all? Have you lost any tenders in the last twelve months where response time was explicitly cited as the reason? Three contracts at an average value of EUR 14,000 is a defensible figure. A 10 percent increase in turnover to EUR 20 million annual turnover is not.
Case study: Logistician, quotation and scheduling system
Company: Regional logistics provider, €24 million in revenue, 40 employees.
Position | Amount (EUR) |
COSTS | |
Build costs (offer and disposition layer) | 95.000 |
Annual maintenance (18% of construction costs) | 17,100 per year |
Internal time during implementation | 10,000 (approx. 200 hours × 50 EUR/hour) |
Total costs over 3 years | 154.300 |
MEASURED BENEFITS (stationary, annual) | |
Coordination time eliminated (8 hours/week × 48 weeks × 55 EUR/hour) | 21.100 |
Error costs eliminated (12 errors/year x EUR 1,800) | 21.600 |
Recovered contribution margin (3 orders/year × €9,000 contribution margin) | 27.000 |
Solved SaaS tools | 9.600 |
Annual total benefit | 95.200 |
Annual fixed income | 79.300 |
Benefits in the first year (70% launch ramp) | 55.500 |
Total benefit over 3 years | 214.100 |
ROI over 3 years: 37% ((214,100 – 156,300) / 156,300)
Payback period: about 25 months
With a payback period of about 25 months and a 3-year ROI of 37 percent, this isn’t a spectacular investment, but it’s a sound one that a CFO can consider. Two factors make it solid rather than inflated. The recovered revenue is counted as contribution margin, not gross revenue, because a recovered order contributes its margin, not its invoice amount. And the benefit is only realized at about 70 percent in the first year because a system takes time to be fully adopted. Over the typical five-year useful life, the ROI rises to about 96 percent because the implementation costs are one-time, while the benefits are annual.
Why the amortization period counts more than the ROI percentage
A 3-year ROI of 85 percent and a 3-year ROI of 200 percent look very different on paper. But if the first is amortized after 19 months and the second after 36, the first is probably the right choice for a company that needs operational improvement now.
The payback period also has a practical use that the ROI percentage does not: it tells you how long the system has to run before it has paid for itself. A system with a payback period of 19 months that runs for five years has delivered a significant return on investment, even if it requires a partial rebuild in the fourth year. A system with a payback period of 36 months that is not adopted in the first year has simply lost money. Calculate both figures. Use the payback period to make the decision. Use the ROI percentage for internal communication to the CFO and management.
What this means for the partner conversation
A credible ROI model for custom software cannot be provided by the partner. It must come from you, based on your own operational data, before a proposal is written. If a partner presents an ROI model in their proposal, the first question is: Where do these benefit figures come from? If the answer is industry benchmarks or standard assumptions, the model is not a company-specific document, but a sales document. Instead, ask for a process that generates your own figures: an analysis phase, a process audit, or a scoping session in which the current state is assessed before any benefits are projected.
The Strategic Audit that appleute conducts prior to each engagement provides the cost of the current state as the first deliverable. Before we design a system or write a line of code, we measure what the current infrastructure costs the company in time, margin and operational risk. This number determines whether the investment makes sense, what scope it justifies and what success looks like after go-live. If the numbers don't support the investment, we say so. The audit is a fixed-price, stand-alone engagement for exactly this reason.
What to do now
When building an investment case for a core operational system, first work through the five cost categories in the table above. This takes two to three hours with your operations leadership and provides the only input a credible ROI model actually needs: the real cost of the as-built.
Once you have this figure, the investment decision becomes clear. Either the cost of the current state is high enough to justify the investment and the payback period is short enough to make the decision viable, or it is not. You know this before the first offer arrives.
If you would like external support with this analysis: The appleute Strategic Audit captures your operational process, identifies where costs are concentrated and provides a cost-as-is figure alongside a build breakdown and a payback estimate. A fixed-price engagement that provides a document you can present to your CFO or senior management with conviction. If the numbers don't support the investment, we'll say so too.
If you want this clarity: We are ready.
Arrange a free, no-obligation consultation with our team.
EN 
DE 
FR 
ES