How to Estimate a Claude Code Build So You Don't Lose Money

Why Good Builders Still Underquote

Claude Code makes the first 80 percent of a build feel instant. You describe what you want, watch it come together in an afternoon, and your brain quietly files the whole project as 'a day of work.' Then you quote off that feeling and lose money for two weeks.

The demo is not the build. The demo is the part where everything goes right on your machine with clean test data. The build is everything after: the messy real inputs, the client who wants one more field, the deploy that breaks, and the hour spent explaining how to use it. Your estimate has to cover all of that, not the magic afternoon.

Estimate the Whole Delivery, Not the Code

Before you put a number on anything, break the project into the five phases that actually happen on every paid job. Most builders only price the first one.

• The core build - the working version that does the main job.
• Edge cases and real data - the ugly inputs, the exceptions, the 'what happens when it is empty' cases.
• Revisions - the client will ask for changes, guaranteed. Price for two rounds, not zero.
• Handoff - documentation, a walkthrough call, and getting it running in their world, not yours.
• Buffer - the unknowns you cannot see yet, because there are always some.

A Simple Method You Can Run in Ten Minutes

You do not need a spreadsheet with forty rows. You need a repeatable habit. Here is the one I use before quoting any build.

1. Write the deliverable as one plain sentence - what the client gets and what it does.
2. List the phases above and give each a rough time in hours, guessing high when unsure.
3. Add the phase times, then add a buffer of 30 to 50 percent for a build you have never done before.
4. Multiply the total hours by the floor rate your numbers require, then round up to a clean fixed price.
5. Sanity-check it against the value: if the build saves the client ten hours a month forever, a low quote is you leaving money on the table.

The Buffers That Keep You Profitable

The buffer is not padding you feel guilty about. It is the price of certainty for the client and the price of sanity for you. A build you have done before might need 20 percent. A build using a tool or integration you have never touched might need 50 to 100 percent, because the unknowns are real and they always cost time.

Quote a Range or a Bounded Fixed Price

When you genuinely cannot size a project - the scope is fuzzy or the client does not know what they want yet - you have two honest moves. Quote a range and explain what pushes it to the high end, or quote a small paid scoping phase first, then a fixed price once the shape is clear.

Either beats guessing a single number to sound confident. A range protects you and educates the client. A paid scoping phase gets you paid to remove the very uncertainty that would have sunk your estimate. Both keep you out of the trap of eating unpaid hours because you quoted the demo.