Why the Same Software Brief Gets Quotes from $2,000 to $80,000

Driver one: your brief is being interpreted, not read

A brief that says 'a booking system for our clinic' honestly supports a $900 build and a $60,000 build. One vendor reads it as a scheduling page with reminders. Another reads in patient records, audit trails, insurance workflows and a mobile app, because their last clinic client needed all of that. Both quote sincerely for the product in their head.

The spread here is your leverage: the questions each vendor asks reveal what they are actually pricing. A vendor who returns a number without asking anything is quoting their default product, not yours.

You can test this deliberately. Before requesting quotes, write down the three most important things version one must do and the three things it deliberately will not. Then watch which vendors discover both lists through their questions. The ones who do are pricing your project; the ones who do not are pricing a guess — and a guess is exactly what the published overrun statistics describe going wrong.

Driver two: production method — the 2026 divide

The largest structural driver today is how the code gets written. A traditional team hand-writes everything through a pipeline of designers, developers, testers and project managers, each billing hours. An AI-assisted studio generates boilerplate, tests and first-draft interfaces in minutes and spends senior engineering time on architecture, security and review. Same deliverable, radically different hours — and price follows hours.

The shift is measured, not claimed. Stack Overflow's 2025 Developer Survey found 84% of developers using or planning to use AI tools, with 51% of professionals using them daily; GitHub's controlled experiment on AI pair programming recorded a 55.8% completion-speed improvement on a standard task. Neither method is dishonest, in the way CNC machining and hand-carving coexist at different prices. But the gap explains most of the difference between a $2,500 quote and a $25,000 one for identical scope, and it is why quotes have spread further apart in the last few years, not closer together.

Driver three: labour rates span 5x before anything else

Underneath every quote is an hourly cost, and the published spread is enormous. Index.dev's rate research puts the global average for freelance developers at roughly $101 an hour, with North American rates commonly $80–140, UK rates around $75–95, and Southeast Asian rates $20–33 for comparable work descriptions.

Run a 200-hour project across that spread and rate alone moves the price from 200 × $30 = $6,000 to 200 × $150 = $30,000 — a 5x difference before interpretation, method or overhead enter the picture. Stack the drivers and the full $2,000-to-$80,000 spread stops being mysterious: it is rate × hours × overhead × risk, and every factor varies legitimately.

Two cautions on reading rates. A low hourly rate is not a low project price if the hours multiply — coordination overhead, rework and time zones can eat the difference. And a high rate is not padding if the hours are few: a senior engineer who finishes in 30 hours costs less than a cheaper one who takes 90. Always compare projected totals against a defined scope, never rates against rates.

Drivers four and five: overhead and risk padding

Headcount you never meet is in every quote. Account managers, office space, sales commissions and bench time between projects all land in the day rate — a general market observation, and the main reason two teams of similar skill can differ by half on price before any work is scoped. Ask a simple question of any large quote: how many of the people on this project will write or review code? The ratio of builders to non-builders in the answer explains a surprising amount of the number.

Risk padding is subtler, and the research explains why vendors do it: McKinsey's study with the University of Oxford found large IT projects running 45% over budget on average, and a vendor who has been burned prices that history into every vague brief. An unclear spec, a client who seems likely to change their mind, an unfamiliar domain — each adds contingency to an hourly estimate, because the vendor is insuring against surprises with your money. This is also why sharpening your brief is the highest-return hour you can spend: precision removes the insurance premium from every quote you receive.

Driver six: positioning and what is quietly excluded

Some firms price to signal. An $80,000 quote from a brand-name consultancy partly buys their process, their liability cover and the reassurance their logo gives your board — real things, priced whether you need them or not. At the other end, some very low quotes are anchors: a $1,500 headline that reaches $8,000 through change requests, because the initial number excluded deployment, revisions and anything not literally listed.

Exclusions are the quiet variable. Two quotes of $5,000 and $9,000 can converge once you price what the cheaper one leaves out: data migration, deployment, a warranty, testing beyond the happy path. The cheapest quote and the cheapest project are often different documents.

The tell for an anchor quote is asymmetry: a headline number stated precisely, with everything around it left vague. A genuine low price comes with a precise scope, because precision is how the vendor protects their own margin. Vagueness plus a low number means the price lives in the change requests you have not made yet.

How to make quotes actually comparable

The fix is to remove the interpretation gap before numbers arrive, then compare like with like. One page of discipline does it.

Send the same page to every vendor, unchanged, and refuse to discuss numbers until each has responded to it. The exercise costs you an evening. It repays itself the moment two quotes arrive priced against identical assumptions — something buyers who skip the page almost never get, and the reason most quote comparisons end up comparing imaginations rather than offers.

• Write a feature list with counts — user types, screens, integrations by name — not adjectives.
• State explicit exclusions: what version one deliberately does not do.
• Define done: deployed to production, data migrated, warranty stated, code in your repository.
• Ask every vendor the same three questions: what is excluded, what happens if it takes longer, who owns the code?
• Ask what they would cut to reach a smaller number — the answer shows whether they understood your priorities at all.

A worked normalisation: one brief, three honest quotes

Take a real-shaped brief — a client portal with two user types, five screens, Stripe billing and email notifications — and decompose three plausible quotes. Quote A, $3,000, from an AI-assisted studio: roughly 40 senior-review hours over generated code at an effective $75 blended rate, deployment and a 30-day warranty included.

Quote B, $18,000, from a traditional agency: about 220 hours across a designer, two developers, QA and a project manager at an $80 blended rate, with a discovery workshop and fortnightly status calls. Quote C, $60,000, from a consultancy: the same build wrapped in compliance documentation, liability cover and a named engagement manager, priced for organisations that require all three.

Decomposed, none of the three is lying — they are different products. The normalisation exercise tells you which product you are actually buying, and whether the extra $15,000 or $57,000 purchases anything your organisation needs. Most small businesses discover they were about to pay for machinery built for someone else's procurement department.

The decomposition question to ask every vendor is disarmingly simple: roughly how many hours is this, and who spends them on what? Vendors with defensible prices answer readily, because the answer is their justification. Evasion on the question is not proof of bad faith, but it does tell you the number was positioned rather than computed — and positioned numbers are the ones that move most under negotiation.

What 79 published fixed prices look like

One way to anchor the spread is a dataset with no interpretation gap at all. Across PINCLER's 79 documented projects — published project by project at /research/what-you-can-build — every quote was fixed between $500 and $2,500, the median came to $1,450, and the median delivery was 13 days, with all 79 shipping inside 30 days.

The category medians show how scope moves price inside a disciplined band: websites at $1,025 and 8 days, integrations at $1,200 and 10 days, chatbots at $1,375 and 13 days, web apps at $1,925 and 18 days. A $1,000 budget reaches 55 of the 79 projects; $2,000 reaches all of them. That is what the bottom of the market's spread looks like when it is published rather than negotiated — and it is the control group we suggest you compare any quote against.

Where we sit in the spread, and why

We publish our position rather than quoting into the fog: every project is fixed price between $500 and $2,500, scope in writing, and bigger ideas split into phases that each ship something usable. That is the AI-assisted end of the market with senior engineers on architecture and review — the production method, priced transparently, from a studio built to offer custom software development services at a published ceiling.

Whoever you shortlist, run the normalisation exercise above; it costs an hour and routinely saves thousands. If you want a data point for your own brief, the free 30-minute call ends with a written fixed quote within one working day, which makes us easy to compare against anyone.