Home · Blog · Custom build
Guide · Custom build

When the estimating spreadsheet stops scaling

Every contractor starts with a spreadsheet, and it works for longer than anyone expects. This is about the specific point where it stops — and what an estimating and tender system has to do to be worth replacing it with.

15 SEP 2026 · 8 MIN READ · OPPERMIND PTY LTD
TL;DR. The spreadsheet does not fail because of volume. It fails on history, concurrency and revision control: nobody can say what a rate was on the day you priced, two estimators cannot work at once without merging files by hand, and an addendum arriving on the Wednesday of tender week means a new version nobody fully trusts. A system earns its place by fixing those four things — one rate library with dated history, package-level subcontractor coverage, revisions without new files, and a trail from enquiry to award. The rates stay yours.

Ask a head contractor where their pricing lives and you will usually get a slightly embarrassed answer. A master workbook, a folder of job copies, a rate sheet somebody maintains, and an estimator’s own version with the numbers they actually trust.

This is not incompetence. A spreadsheet is a genuinely excellent estimating tool — fast, flexible, and shaped exactly like the way estimators think. Most contractors should stay on one for longer than a software vendor will tell them.

But it fails in a predictable way, and the failure is not about how many jobs you price. It is about four specific things.

The four things a spreadsheet cannot do

1. Tell you what a rate was on the day you priced

A cell holds one value: today’s. When a job goes bad eight months later and somebody asks why the concrete was priced the way it was, the honest answer is usually “that is what was in the sheet at the time” — and the sheet has been updated four times since.

Without dated rate history you cannot do the single most valuable thing in estimating: compare what you priced against what the job actually cost, and feed that back into the next price. Every contractor intends to do this. Very few can, because the evidence has been overwritten.

2. Let two people work at once

The moment a second estimator joins, or an estimator and a contracts administrator need the same job open, you are merging files by hand and hoping. Most pricing errors that reach a client are not arithmetic. They are a version that was not the latest one.

3. Absorb an addendum without becoming a new file

Tenders change mid-flight. Addendum 3 lands on the Wednesday, the drawings are revised, two packages move scope. In a spreadsheet that is a new version, a manual diff, and a quiet anxiety about what was missed. In a system it should be a tracked change with a visible before and after.

4. Show coverage

Before submission the question that matters is which packages have enough subcontractor prices and which are carrying a guess. In a spreadsheet that answer lives in an estimator’s head and an email folder. It should be a screen.

What a real system does

Job Spreadsheet Estimating and tender system
RatesCurrent value onlyOne library, dated history, who changed it and when
Two estimators at onceManual mergeSame job, concurrently, with a record of who priced what
Addenda and revisionsA new fileA tracked revision with a visible difference
Subcontractor coverageAn email folderCoverage per package, gaps visible before submission
Qualifications and exclusionsRetyped each timeA reusable library, assembled into the submission
Estimated vs actualRarely possibleThe feedback loop that improves the next price
Why we priced it that wayInstitutional memoryAn auditable trail from enquiry to award

Notice what is not in that list. Nothing here prices the job. The rates, the risk allowances, the read on the client and the programme, the decision to sharpen a number to win work — all commercial judgement, all staying with your estimator. A system that offers to do that for you is solving a problem you do not have.

Where AI genuinely helps, and where it does not

The useful AI work in estimating is reading and assembly, not pricing.

Reading. A tender package is a document set: drawings, specifications, a bill of quantities, conditions, addenda. Pulling a structured quantity list out of that, and flagging exactly what changed between addendum two and three, is high-volume, low-judgement work that eats estimator hours and produces no advantage when done by hand.

Assembly. The qualifications, the exclusions, the methodology statement, the covering letter, the conformance schedule — largely the same each time, individually retyped under deadline. Drafting those from a library and assembling the submission is a real saving in the week it is scarcest.

Not pricing. Rates and risk are yours. A number your estimator did not choose is a number nobody can defend in a post-tender interview, and the first time a client asks how you arrived at it, that matters more than the hours saved.

The same principle runs through the small end of this trade too — AI for tradies follows one job from the first call to the final invoice, and lands in the same place: the document is automatable, the number is not.

Buy first. Build when the fit is genuinely wrong.

Off-the-shelf estimating packages are mature and handle a great deal. Configuration is cheaper and faster than a build, and most contractors should exhaust that route before commissioning anything.

Custom earns its cost in a narrower set of cases:

The full version of that decision, including the six shapes this work usually takes, is in AI solutions for business. The questions to ask a developer before you sign are in custom AI software: the questions to ask before you sign — question six, what it costs to run, is the one most often skipped in this sector.

The first thing to fix, whatever you do next

The rate library. One library, dated history, a named owner, and estimators working from it rather than from private copies.

It is unglamorous, it does not require new software to start, and it fixes more margin problems than anything else on this page. It is also the thing that makes any later system worth having: a new tool loaded with inconsistent rates produces inconsistent prices faster.

Where Oppermind fits

Two routes, and they are genuinely different.

If your problem is the documents around the estimate rather than the estimating itself, the Oppermind platform covers it on a subscription: a spreadsheet with around 397 functions for the rate work, a document editor with real .docx round-trip for qualifications and submissions, a full PDF editor that annotates, OCRs and marks up drawings and specs, and a project schedule with critical path for the programme. Free with no card, then Starter A$9.95, Pro A$29.99 and Pro Plus A$59.99 a month.

If the estimating system itself is the problem, that is Corporate Solutions — custom estimating and tender management systems, built to your pricing method and connected to the tools you already run, including the ERP and compliance systems the estimate has to feed.

The first conversation is a diagnosis. If the answer is configuring a product you already own, we would rather tell you that.

Whatever you’re here to make, make more of it.

Frequently asked questions

When does a business outgrow estimating in a spreadsheet?

The usual signals are structural rather than about volume: more than one estimator working at once, rates that live in several files nobody can reconcile, revisions arriving mid-tender, and nobody able to answer why a job was priced the way it was six months later. A spreadsheet is excellent at arithmetic and poor at history, concurrency and revision control.

What should an estimating system do that a spreadsheet cannot?

Hold one rate library with dated history so you can see what a rate was on the day you priced. Track subcontractor coverage per package so gaps are visible before submission. Handle addenda and revisions without a new file. Keep a clear trail from enquiry to submitted price to awarded job. And compare what you estimated against what the job actually cost.

Should we buy estimating software or build it?

Buy first. Off-the-shelf estimating packages handle a great deal, and configuration is far cheaper than a build. Custom earns its place when your pricing method genuinely differs from the industry norm, when the system must sit inside an existing chain of ERP, scheduling and compliance tools, or when the trade breakdown a product assumes does not match how you actually price work.

What is the most valuable thing to fix first?

The rate library. Rates scattered across files, each estimator keeping a private copy, is the single most common cause of inconsistent pricing and the hardest thing to unpick later. One library with dated history, owned by a named person, fixes more margin problems than any other single change.

Can AI price a tender?

No, and treat any claim otherwise carefully. What AI does well around estimating is the reading and the assembly: pulling a bill of quantities out of a document set, flagging what changed in an addendum, drafting the qualifications and exclusions, and assembling the submission. The rates, the risk allowances and the final number are commercial judgement and stay with your estimator.

Related reading

This article is general information about software and estimating practice. It is not legal, commercial, procurement or financial advice and does not take account of your circumstances; pricing, tendering and contractual decisions remain yours and should be taken with your own professional advice. Oppermind platform prices are current as at 15 September 2026, are in Australian dollars, and are subject to the plan terms at checkout. Corporate Solutions engagements are quoted individually and are not covered by platform pricing.

Describe the friction. We’ll name the solution.

Tell us where estimating is slow, inconsistent or hard to defend — we’ll identify the most practical place to start, even when the answer isn’t ours to sell.