How our software cost calculator works

What every estimate assumes

• A senior US team. The base ranges assume a boutique team of senior people based in the US. Other team models are applied as a multiplier.
• A first sellable version. The baseline scope is an MVP: the first release you could put in front of paying customers or your whole team. Larger scopes are multipliers.
• Build cost is one-time. The headline number is the cost of getting that version built. Ongoing costs are shown separately as maintenance and infrastructure.
• US dollars, rounded. Results are rounded to the nearest $1k, and amounts of $1 million or more are shown in millions to two decimal places.

Not modelled, and so not included in any figure: software licenses and subscriptions the product depends on, third-party audit or certification fees, marketing and launch costs, data migration and content entry, your own staff's time, inflation and financing.

Base ranges by product type

The starting point is the product you are building. Each type has a build range for an MVP with a senior US team, and a typical monthly infrastructure and operations range used for running costs.

Scope and team multipliers

Scope multiplies the base range. A v1 product goes beyond the MVP; a full platform has several modules.

The team factor is applied last, to the whole build including add-ons. It does not change infrastructure costs. When offshore is selected the calculator shows a warning: the cheapest quotes usually exclude scope, and rework risk is real. These factors are our planning assumptions, not a survey of market rates.

Complexity add-ons

Five items move budgets more than anything else, so each adds a fixed range. The low figure is added to the low end of the estimate and the high figure to the high end.

Maintenance and running costs

Maintenance and evolution is set at 18% of the build cost per year, applied separately to the low and high ends. It covers updates, fixes and the small improvements that keep software current. The 18% is our planning assumption, not a measured statistic; how much you actually spend depends on how actively you keep developing the product.

Infrastructure and operations per year is the midpoint of the product type's monthly range, times 12. If the product is an AI system, or AI features are selected, that figure is multiplied by 1.6 to allow for model inference costs. It is a single figure rather than a range.

The formula, step by step

1. 01Build low = base low × scope multiplier. Build high = base high × scope multiplier.
2. 02Add integrations × $8k to the low end and × $15k to the high end, plus each selected add-on's low and high figure.
3. 03Multiply both ends by the team factor.
4. 04Maintenance per year = 18% of build low and 18% of build high.
5. 05Infrastructure and ops per year = midpoint of the monthly range × 12, × 1.6 for AI.
6. 06Total for N years = build + N × (maintenance + infrastructure and ops), shown for years 1, 3 and 5.

Worked example

These are the calculator's default settings: a B2B SaaS platform, MVP scope, senior US team, two integrations, no other add-ons.

The width of that range is deliberate. Before scoping, nobody knows whether a SaaS MVP needs two screens or twenty, so the honest answer is a range that a conversation narrows.

Where the service-page ranges come from

Service pages lead with Eltherion's starting price, as low as $1,000. The comparison range shown beside it is what a typical US agency charges: the calculator's base range for the matching product type, at MVP scope with a senior US team and no add-ons. We set the two to match so the page and the tool never disagree.

Legacy Modernization, Software Architecture & Strategy, Platform Engineering and Technical Due Diligence show no range. That work depends too heavily on the existing system to price from a product type, so it is scoped individually. Where the site says a focused internal tool starts around $60k, that is the internal tool base range.

The build vs. buy calculator

Our build vs. buy calculator compares the cumulative cost of three paths over five years. You can change every input.

• Buy: one-time integration cost, plus the vendor's annual cost growing by your chosen percentage each year.
• Build in-house: the build estimate plus 25% for hiring and coordination overhead, then your chosen number of full-time engineers at a loaded annual cost to own it. The 25% is our assumption.
• Build with a partner: the build estimate, then a yearly maintenance cost as a percentage of the build.

Defaults are a $60k per year vendor growing 15% a year, $25k of integration, a $250k build, 18% maintenance, half an engineer in-house and a $200k loaded cost per engineer. At those defaults, buying is cheapest over five years (about $430k, against $475k with a partner and about $813k in-house), and the calculator says so.

For context on the $200k default: the U.S. Bureau of Labor Statistics reports a median annual wage of $135,980 for software developers in May 2025, and benefits made up 30.0% of private-industry employer compensation costs in June 2026. If that average applied, a median-paid developer would cost about $194,000 a year in wages and benefits, before recruiting, equipment and management time.

Limitations

We would rather you know where this model is weak than trust it too much.

• It reflects typical agency pricing as we see it, not Eltherion's own prices, which start as low as $1,000. Firms price very differently.
• Multipliers are linear. Real projects do not scale in straight lines, and add-ons interact with each other.
• Team factors capture price, not outcome. Rework, communication overhead and quality differences can erase the savings of a cheaper team, and the model does not try to measure that.
• Running costs are typical figures for each product type. Heavy usage, especially for AI products where inference cost grows with volume, can push them much higher.
• Maintenance is a flat 18%. A product under active development will cost more; a stable internal tool, less.
• It does not adjust for region, inflation, deadlines or the state of any existing system.
• It is not a quote. A real price needs scoping: what you are building, what exists today and what done means.

When the calculator or the service-page ranges change, we update this page at the same time and change the date at the top.