Software pricing feels arbitrary from the outside because the same brief genuinely can cost $4,000 or $40,000 depending on decisions nobody wrote down. Below is how the number is built, with the ranges we actually quote, which are useful whether or not you work with us.

One caveat worth stating up front: these are ranges for a competent build by a small senior team, which is the market most growing businesses buy in. Enterprise consultancies and offshore body shops both sit well outside them, in opposite directions.

Real ranges by project type

ProjectTypical rangeTime to first release
Validation MVP: one workflow, prove the idea$999–$3,0001–2 weeks
Internal tool: replaces a spreadsheet a team lives in$2,000–$6,0002–3 weeks
Single business module: inventory, purchasing, sales pipeline, payroll$5,000–$15,0002–4 weeks
Custom CRM with automation and reporting$10,000–$30,0003–6 weeks
Multi-module ERP platform$15,000–$60,000+6–16 weeks, phased
Mobile app (iOS + Android, shared codebase)$6,000–$25,0004–10 weeks
E-commerce platform with custom logic$5,000–$20,0003–8 weeks
AI agent: chat or voice$3,000–$12,0002–4 weeks
Single automation between two systems$800–$3,0003–10 days

If a number here is far below what you have been quoted elsewhere, the difference is usually scope, seniority, or what is being excluded, not magic. If it is far above, someone is either much cheaper on labour or leaving work out. Both are worth asking about directly.

The seven things that move the price

Notice that “number of screens” is not on this list. It is the thing clients estimate with and it is nearly the least predictive.

  1. Integrations. Each external system (accounting package, payment gateway, marketplace, WhatsApp, carrier) adds real work: auth, error handling, rate limits, and someone else’s undocumented behaviour. This is the single biggest multiplier.
  2. User roles and permissions. One kind of user is simple. Five roles with different visibility, approval limits and audit requirements multiplies both build and test effort.
  3. Data migration. Bringing in years of messy history (duplicates, inconsistent codes, negative stock) is frequently 10–20% of a project and is the most commonly under-quoted line.
  4. Business rule complexity. A flat price list is cheap. Tiered pricing by customer, currency, quantity break and season, with approval overrides, is not.
  5. Compliance and security requirements. Payment data, health records, audit trails, data residency: each adds design and verification work that cannot be skipped.
  6. Offline, real time, or heavy scale. Working without connectivity, live updates across users, or high transaction volume each change the architecture, not just the code.
  7. Design ambition. A clean functional interface is included. A distinctive designed product with motion and bespoke components is a separate craft with a separate cost.

Costs most quotes leave out

Ask explicitly about each of these. A vendor who has thought about them is a better bet than one with a lower number.

ItemTypical costNotes
Hosting and infrastructure$20–$300 / monthDepends on users and data; most business systems sit at the low end
Maintenance and changes15–25% of build / yearDependency updates, security patches, small changes, support
Third-party servicesVariesSMS, e-mail delivery, maps, AI usage, payment fees
App store accounts~$25 one-off + $99 / yearGoogle Play and Apple Developer
Data migration10–20% of buildOften quoted as “we will import your data” with no hours attached
Training and rolloutDays of your team’s timeReal cost even when the vendor charges nothing
Internal timeThe one nobody countsSomeone senior must answer questions and make decisions weekly

Want a real number for your project?

Two lines about the problem is enough for a scoped plan with a price range and a timeline, inside 24 hours, free, and yours to take to other vendors.

Get a free quote →

Fixed price vs time and materials

Fixed priceTime and materials
Best forA well-defined first releaseOngoing evolution and discovery work
Who carries scope riskThe vendor, which is why they pad itYou, which is why it can be cheaper
Change handlingChange requests, friction, renegotiationReprioritise next week, no drama
Failure modeVendor protects margin by cutting qualityScope drifts with no natural end

The arrangement that fails least often: fixed price for release one, so both sides commit to something concrete, then a monthly capacity arrangement for everything after. Whatever the model, insist on weekly demos. They are the only real protection either party has.

Why the same brief gets $4k and $40k quotes

  • Different scope, same words. “A CRM” can mean four screens or forty. Compare explicit in/out lists, not totals.
  • Seniority and review. A junior team with no code review is cheaper per hour and often more expensive per outcome.
  • Excluded work. Testing, deployment, migration, documentation, handover: each quietly removed line lowers a quote.
  • Process overhead. Large consultancies bill the account manager, the analyst and the process. Sometimes you need it; often you are buying it by accident.
  • Deliberate low bids. Some quotes are priced to win, with the real money planned for change requests. Ask what a typical change costs before you sign.

Five legitimate ways to spend less

  1. Cut scope, not quality. One workflow, one team, one location, in production. Then extend. Almost every project has a version that costs 40% less and delivers 80% of the value.
  2. Keep what works. Do not replace functioning accounting or payroll software as part of a bigger project. Integrate.
  3. Do your own data cleaning. Deduplicating your item and customer lists in a spreadsheet costs your time instead of engineering hours, and you will do it better.
  4. Use standard components where nobody cares. Login, notifications, file storage, reporting exports. Save the custom work for the part that is actually your business.
  5. Decide fast. The most expensive thing in any project is a question that stays unanswered for a week. Name one internal decision-maker and give them the authority.

And one non-saving disguised as a saving: skipping testing. It converts a known cost into an unknown one, usually at your busiest moment.

Deciding if it is worth it

Do the arithmetic before comparing quotes. Three questions:

  • Hours returned. How many staff hours a week does this remove? Multiply by loaded hourly cost and by 50 weeks.
  • Leakage stopped. Lost enquiries, stock write-offs, unbilled work, pricing errors. Even a rough monthly figure is enough.
  • Capacity gained. Could you handle 30% more volume without hiring? Value that at the gross margin of the extra volume.

If the annual total of those three comfortably exceeds the build cost, the decision is not really about price. It is about whether the vendor can deliver. If it does not, build something smaller until it does.

That is also how we scope our own work: pick the slice with the clearest payback, ship it in about ten days, measure it, then decide together what is next. If you want that applied to your situation, send us two lines about the problem, or read which system to build first before you spend anything.