The quoting spreadsheet, and the week it stops working

Every vendor in this market opens by insulting the quoting spreadsheet, and every job shop that has one knows the insult is unearned. It has priced work for 9 years. It has never been down. Nobody has ever needed training on it.

It is also the single largest piece of undocumented business logic in most manufacturing companies, and it usually has one author. This post is about both of those facts, because the case for replacing it is real and the case for respecting it first is stronger than software people admit. What replacing it involves is covered in the buyer’s guide.

What the quoting spreadsheet gets right

It encodes judgement that exists nowhere else in the building. The margin that goes on this customer because they pay late. The extra hour on anything with a cosmetic finish because the last 3 came back. The material multiplier that looks arbitrary and is not.

None of that is written down anywhere else. It accumulated one correction at a time over years, each one added the week after a job went wrong, which is the most reliable way anything ever gets learned on a floor.

It is a record of every lesson.

It also costs nothing, opens instantly, and every estimator can read it without a login. There is no vendor, no renewal, no roadmap and no support queue. When a customer asks for a breakdown in an unfamiliar format, somebody builds it in 20 minutes.

Any replacement has to beat all of that, and most replacements are only compared against the failure modes below rather than against the whole thing. That comparison is how floors end up paying for a system their estimators route around.

The formula that breaks quietly

The first failure mode is the one nobody sees, because a spreadsheet does not report errors, it reports numbers. Somebody inserts a row above the rate table. A lookup range that used to end at row 40 now ends one row short. Every quote using the last work center in that table starts pulling the wrong rate.

Nothing turns red. The quote looks normal, goes out, and is possibly won. The error surfaces months later when a job runs and the margin is not there, and by then 40 quotes have used the same broken lookup.

The specific mechanics vary and the shape does not. A dragged formula that lost an absolute reference. A hardcoded number pasted over a calculation during a rush. A column inserted for one customer that shifted everything to its right.

The reason this failure is specific to spreadsheets is that there is no separation between the data and the logic. In any other system the burdened rate table is a table and the calculation is code, and changing one cannot silently corrupt the other. In a workbook they occupy the same grid and a row insert touches both.

The usual defence is locking the calculation cells, which helps until the first time somebody opens them up for a genuine change and then leaves the sheet that way. Protection switched off under time pressure is protection that is off during exactly the weeks it was needed.

There is a cheaper habit that catches most of it. Keep 3 parts you have quoted before as test cases, priced by hand once and written on a card, and run them through the file after any structural edit. If the answers moved, something else moved with them.

The two versions that drift apart

The second failure arrives the day a second person needs to quote at the same time. Somebody saves a copy to work on a big package. The original gets a rate update on Tuesday. The copy gets a new customer’s form built into it on Wednesday.

Neither person did anything wrong and there is now no correct file. Both contain work the other lacks, and merging them means a human comparing formulas cell by cell, which nobody has time for and which therefore does not happen.

Shared drives and cloud versions help with the crudest form of this and do not solve it. The problem is not simultaneous editing. It is that a spreadsheet has no concept of which change was a policy decision and which was a one-off for a single quote.

The tell is when an estimator says they use the one on their desktop because the shared one has something wrong with it. At that point there are 2 pricing policies in the company and the difference between them has never been discussed.

The key person who resigns

The third failure is the one that ends careers and it has a name in every building. One person built it. One person understands why the finish factor is 1.18. One person knows which tabs are live and which are abandoned experiments.

When that person retires, the spreadsheet keeps working and stops being changeable. New customer, new form, no one to build it. Rates move, nobody trusts themselves to touch the rate table. A formula breaks and the fix is a guess.

The logic left with the person.

This is the risk that justifies the whole software category, and it is not really a software problem. It is that the argument behind each number was never written down, only its result, and a result without an argument cannot be maintained by anybody else.

The mitigation costs nothing and almost nobody does it. Sit with whoever owns it for 2 afternoons and document why each factor is what it is. That document is worth more than the file, and it is the thing any replacement project will need on day 1 anyway.

The week volume outruns it

The fourth failure is arithmetic rather than accident. A spreadsheet takes a fixed amount of human attention per quote, and that attention does not compound. At 20 requests a month it is comfortable. At 120 it is 2 people and a queue.

The queue is where the money goes, because requests that sit for 8 days lose to the ones that arrived on day 2. What a quote never sent actually costs works through the arithmetic of that, and it is larger than any licence in this market.

The signal is not that quoting feels slow. It is that requests are being triaged out. When somebody decides on a Friday which 30 of the 90 to answer, the spreadsheet has already stopped working and the company has adapted around it without deciding to.

Volume also breaks the file physically. Nine years of quotes in one workbook means a file that takes 40 seconds to open and recalculates on every keystroke, and the standard fix, starting a fresh file each year, is how the history stops being searchable.

The queue is the real cost.

That last consequence is the expensive one and it is rarely counted. The value of 9 years of quoting is that you can ask what work like this cost last time. Split across 9 files with 3 different layouts, nobody asks, because answering takes an afternoon and the quote is due Thursday.

At that point the spreadsheet has quietly stopped being a record and become a calculator. It still produces numbers. It no longer produces the one thing it was uniquely good at, which was remembering.

What to audit before you replace it

Do this before any demo, because it decides which of the alternatives you actually need. Pull the file apart and answer 5 questions in writing.

Which rates are current, and when was each last reviewed against what a burdened rate should include? Most files carry at least one rate that predates a machine purchase.

Which factors exist and why? Every multiplier gets a sentence explaining what it is compensating for. The ones nobody can explain are the interesting ones.

Where does the file get material cost, and is that a live source or a number typed in 2 years ago? Where does it get outside processing, and is that a real quote or a remembered one?

What does it not do at all, that somebody currently does by hand afterwards? That list is usually longer than expected and it is the part a should-cost tool will not cover either.

Which of the 4 failures above has already happened here? A floor that has lived through the second and third needs something different from a floor whose only problem is volume.

Run that audit and you may conclude the spreadsheet plus a rate review buys another 2 years, which is a perfectly good outcome and cheaper than anything a vendor will propose.