About Iron Pixel Works

Why we keep score

Iron Pixel Works was founded to estimate software honestly.

Our developers had all worked where estimates were optimistic by default and nobody tracked the outcome. Overruns were absorbed into scope conversations and change requests, and the gap between estimate and actual was never written down.

Without a record there is no calibration. An estimator who never sees how wrong they were last time will be wrong by the same margin next time, permanently.

Clients cannot compare firms on the thing that matters. Quotes are comparable; reliability is not, because nobody discloses it. A shorter estimate frequently signals worse estimating rather than better capability.

So we publish estimated against actual for everything we have delivered. The median overrun is eleven per cent, the worst was sixty-three, and both figures are available before you commit to anything.

How we estimate

Discovery precedes the number. A paid discovery engagement produces the scope, and the estimate follows from it rather than being guessed at from a conversation.

Unknowns are priced as ranges. Where a third-party integration or an undocumented system is involved, the estimate carries a range with the assumption stated rather than a single confident figure.

Exclusions are written down. What is not in scope appears in the document, which removes most of what becomes a change request argument later.

Comparable projects are disclosed. When we quote, we show what similar past builds were estimated at and what they actually took.