How a TrueBallpark estimate is built

TrueBallpark is an independent software-cost benchmarking system. It is not an agency, and this page is not selling you a build. We do not take referral fees and we do not pass your idea to a development shop — there is nobody downstream to route you to.

What we produce is a range: a ticket-by-ticket breakdown of the work your idea implies, sized against published market rates and a small number of real, delivered builds. It is a benchmark, not a bid. Nobody here is offering to do the work for the number shown, and the number is not a quote.

You describe your idea in your own words. We classify it and ask up to five clarifying questions — only the ones that could actually change the price, ranked by how much the hour total would swing if the answer went the other way and how uncertain we currently are about it. Simple ideas get fewer questions; complex ones get more, up to a hard ceiling of seven.

Once you confirm the scope we think you mean, it is broken into modules and then into individual tickets — the same unit of work a developer would put on a board, with a stated reason for its size. Each ticket is sized Small, Medium or Large, and each size carries a fixed hour range. We price two versions of every idea from the same ticket list: your complete idea as described, and the leaner scope we think you could actually launch first, with the difference explained in plain language.

The site describes tickets as matched against roughly 100 standard components. Here is what that is, plainly: there is no static catalogue of 100 named line items. What exists is a pattern library — the expected modules, workflows and integrations for about ten product categories (marketplace, SaaS, ecommerce, social, booking, fintech-adjacent and a few others) — combined with three fixed hour bands, Small, Medium and Large, that every ticket is placed into. "100 components" is shorthand for that combination, not a literal count.

The bands themselves were calibrated against published freelance and agency figures across six app archetypes, then checked against two real, delivered builds with known final costs — a recruitment platform and a deals aggregator, both landing inside or within 7% of the priced range. Two builds is a small truth set, and we say so rather than round it up. Growing it — by asking paying customers what they were actually quoted — is ongoing, not finished.

The US benchmark rate is $85/hour, with a $75–95/hour spread underneath it as the secondary "depending on who you hire" figure. Both are sourced from Gun.io and Toptal's published freelance rate bands, not from any of our own transactions.

UK (£65/hour) and EU (€70/hour) rates are provisional. Our most recent market research pass covered only US-published figures, so the UK and EU bands are derived mechanically from the US numbers by the same ratio — not independently sourced. Every UK or EU estimate is flagged provisional until a dedicated research pass replaces them.

Agency comparisons apply a 1.8–2.2× multiplier to the benchmark range, clamped to a published agency band — $25,000–$200,000 in the US, scaled proportionally for UK and EU.

The range prices the tickets in your confirmed scope, at a hire-someone-yourself rate. It does not include ongoing costs after launch — hosting, content operations, customer support, marketing or acquisition spend. It does not include the account-management and delivery-risk premium an agency charges; that premium is what the separate agency-comparison figure represents, and it is deliberately kept out of the benchmark headline.

It does not include anything outside your confirmed scope — if a requirement never came up in the questions you answered, it is not in the tickets. And it does not include region-specific legal, compliance or industry-certification work; that can surface as a clarifying question, but it is never automatically priced.

The hour bands assume a competent professional using modern tooling — which already means some amount of AI-assisted coding, not development the way it looked five years ago. Both of the real, known-cost builds behind the current calibration were delivered with heavy code generation, and both landed in the lower half of their priced range, roughly where an AI-accelerated build would be expected to sit.

Studios built specifically around AI-assisted delivery can come in lower still. Treat that as a real, directional effect rather than a fixed discount: one figure surfaced in our market research, a $12,000–$45,000 "AI-assisted" MVP claim, read like a sales-funnel hook for one shop's service rather than a representative price, and it was excluded from calibration for that reason.

Archetypes barely separate at the low end today — a CRUD marketplace and a real-time, GPS-heavy app currently price close together, because the Small/Medium/Large system doesn't yet carry a separate "integration-heavy" flag. That's a known gap, not an oversight we're unaware of.

The truth set behind the bands is two delivered builds, both accelerated by heavy code generation — a normally-staffed, non-accelerated build could reasonably land higher in the range, not lower. Highly specific scale requirements, like a fixed number of data feeds or a fixed number of generated pages, aren't priced from any published source we found; they only appear as tickets if a clarifying question surfaces them. And run-to-run variance on an identical, unchanged scope is 3–10% of the range — one reason we show a range and not a single number.

We ask everyone who buys a spec, and everyone who goes on to hire someone, what they were actually quoted or actually paid. Right now that sample is two known-cost builds, both inside or within 7% of the full-scope benchmark range — informative, but not yet a track record.

We will publish the estimate-vs-reality gap once the sample is large enough to say something meaningful, and not before. Until then, every number on this site is a benchmark built from the best public data and real-build evidence we could find — not a promise about what your specific build will cost.