> ## Content Index
> Fetch the complete content index at: https://connect.ghost.preview.themeanax.com/llms.txt
> Use this file to discover other available public pages before exploring further.

# The estimate and the invoice never match
- URL: https://connect.ghost.preview.themeanax.com/estimate-and-the-invoice/
- Published: 2026-10-28T09:00:00.000Z
- Updated: 2026-10-28T09:00:00.000Z
- Description: Six years of comparing what I said a project would take against what it took. The gap is consistent, which means it is not an accident.
- Author: Priya Raman
- Tags: Systems, #Import 2026-11-13 08:50

I have kept every estimate I have given since 2019 next to what the work actually took. Not to feel bad about it — to find out whether the error has a shape. It does, and the shape is boringly consistent, which is the part worth writing down.

## The gap is not where I expected

I assumed I was underestimating the hard parts. I was not. The difficult, technically uncertain work lands close to the estimate, because it is the part I think carefully about before quoting.

What runs over is everything surrounding it: the review cycle, the third set of feedback, the migration nobody listed, the week where two people are away. None of it is hard. All of it is real, and none of it appears in a plan that was written by thinking about the interesting problem.

> I estimate the work. The invoice bills the project. Those have never been the same thing and I keep quoting as though they are.

## What I actually changed

Not the multiplier. Multiplying a bad estimate gives you a bad estimate with padding, and the padding gets negotiated away first because nobody can say what it is for.

- I estimate the surround separately and name it: review, revision, handover. It is a line item, so it is a conversation rather than a surprise.
- Anything I have not done before gets a range, not a number. A range is honest and, unexpectedly, easier to defend than a confident figure.
- I record the actual on the day the project closes. A week later I am already telling myself a story about why it ran long.

## What has not improved

My estimates are not much more accurate than they were. The difference is that the overrun is now something both sides saw coming, which turns out to matter more to a client relationship than precision ever did.

I no longer think accurate estimation is the achievable goal. Predictable disagreement about scope, early enough to act on, is — and that is a different skill entirely.