Excel or a custom application — where the break-even point lies
Łukasz Mikucki · Published Jul 20, 2026 · 9 min read
The question "Excel or a system" is badly posed, because it suggests this is about a tool. It is about something else: how much work it takes for the number one person reports to be the same number another person sees on their own screen. A spreadsheet has no problem with that — as long as there is one person and one copy.
What spreadsheets do not get wrong
Let's start with the honest side of the argument, because trying to convince anyone that Excel is bad ends in lost credibility. A spreadsheet is an excellent tool when:
- the data model is not yet known and changes every week,
- a single role uses it, entering and reading the data themselves,
- the work is a one-off analysis rather than a repeatable process,
- what matters is time to first result, not durability of the solution.
These are not edge cases. A large share of the preparatory work on a contract — analyses, variants, quick calculations — should stay in a spreadsheet, and there is no reason to change that.
Where the hidden cost begins
The cost of a spreadsheet rarely shows up as a budget line. It shows up as time nobody measures:
Two people have two files and different numbers. Before a decision can be made, someone has to establish which file is current. That is not work — it is a tax on not having a register.
The question "why was it 320 in April and 290 now" means opening old copies. A spreadsheet overwrites state; it does not store events.
Progress from one file, materials from another, acceptances from a third. Every board report costs a few hours of copying — repeated every month.
The formulas and logic of the spreadsheet exist only in the author's memory. The author's absence halts the process, and their departure erases part of the organization's knowledge.
That last point is in practice the most expensive and the least often priced in.
Six decision criteria
Instead of asking "are we big enough yet", walk through concrete questions. The more answers land in the right-hand column, the stronger the case for a register.
| Criterion | Spreadsheet is enough | Time for an application |
|---|---|---|
| Number of roles using the same data | One, maybe two | Three or more, in different locations |
| Nature of the data | One-off analysis | A register of events kept continuously |
| Need for history | Current state is enough | You need to know who changed what, and when |
| Consequences of an error | Corrected on the spot | Payment, deadline, liability |
| Data source | One person enters | Site enters, office and management read |
| Repeatability | Occasional | The same flow every week, on every contract |
Note that none of these criteria mentions headcount or revenue. A small company running three parallel work fronts hits this problem earlier than a large one running a single front.
The settlement model moves the break-even point more than company size does
There is one more factor that rarely comes up in this discussion, yet it decides more than headcount: how the employer requires you to settle. This is not uniform practice — three contracts can carry three different regimes.
| Model | What it demands of the register | Where it appears |
|---|---|---|
| Re-measurement contract | A continuous record of quantities in the units of the bill of quantities, entries by the Site Manager confirmed by the Engineer, permanent works measured net | GDDKiA technical specification — „Rejestr obmiarów” (the measurement register) |
| Lump sum settled against the RCO | Lump-sum items settled against the RCO (rozbicie ceny ofertowej — the itemized breakdown of the contract price); the notion of a bill of quantities does not appear in the PFU at all | PKP PLK PFU |
| Lump sum with an indicative BoQ | The bill of quantities explicitly excluded from the basis of pricing and from the scope definition — provided for orientation only | CPK OPW |
The difference is fundamental. On a re-measurement contract, the quantity register is a settlement document — it must be kept as the works progress and carry the other party's confirmation trail. Under a lump sum, the same register is an internal tool: useful, but nobody will enforce it on you.
The practical takeaway: if you run re-measurement contracts, your break-even point comes much earlier than company size would suggest. A spreadsheet nobody countersigns and which keeps no change history is a weak basis for settlement in that model. And conversely — under a clean lump sum you can stay with the spreadsheet longer without real risk.
Three signals that the point is already behind you
In practice, the break-even point is usually crossed before anyone notices. You recognize it by symptoms, not by a metric:
- A spreadsheet about spreadsheets appears. Someone maintains a list of which file is current and who last edited it.
- The board report takes a week to prepare. Not because the data is complex, but because it first has to be reconciled.
- The answer to a status question depends on whom you ask. The site manager, the commercial department, and the contract manager give three different numbers — and each is right within their own file.
The third signal is decisive. It means the organization has no single state of truth, only several parallel ones — and every decision is made on one of them, at random.
What a custom application will not fix
Honesty requires naming the limits. A system will not solve problems that are not IT problems:
- An unagreed process. If there is no agreement on who approves a material request, an application will not settle it — it will only fossilize the dispute in the form of an approval workflow that everyone bypasses.
- Data nobody wants to enter. A field that serves no purpose for the person entering it will be filled with anything. A register survives when it gives something to the site, not only to head office.
- No data owner. Every field needs a role accountable for its correctness. Without that, the system produces complete tables with incorrect content.
That is why the first stage of a sensible rollout is a process audit, not a technology choice.
How to run the numbers for yourself — without an ROI spreadsheet
Do not build an ROI model on parameters you do not measure. A simpler approach gets you a result faster:
For example: "how much did we complete this month on front X, and how much of it has been accepted".
Do not estimate — measure it once, honestly, including the time of the people who had to send something back.
This question comes up every week or every month, on every contract. The product is usually surprising.
Not the whole system — one register that answers that question. The scope of the first stage should be exactly that narrow.
That last remark matters. The most common mistake when moving off a spreadsheet is trying to replicate everything at once. A rollout that starts with a single register with a real owner has a far better chance of surviving contact with the construction site.
The takeaway
Excel does not stop being enough because the company grew. It stops being enough at the moment the same number has to be simultaneously true for several roles — and from that moment on, every week of delay is not a saving but a deferred cost of reconciliation.
Sources
The substantive basis of this text. Items marked as own practice do not come from a document — they are delivery experience that cannot be verified with a publisher.
- DocumentGeneralny Dyrektor Dróg Krajowych i Autostrad / GDDKiA Oddział w KatowicachSTWiORB DM.00.00.00 (GDDKiA technical specification, in Polish), item 7 — re-measurement contract, measurement units and „Rejestr obmiarów” (the measurement register) (pp. 31–32)
- DocumentPKP Polskie Linie Kolejowe S.A.PFU (TOM III SWZ) (PKP PLK employer's requirements, in Polish) — settlement of lump-sum items against the RCO price breakdown (p. 26)
- Own practiceObservations from running settlements and data flows on the contractor side of infrastructure projects
Terms used in this article
Related articles
Where information gets lost in construction document flow
Seven points where information on an infrastructure contract disappears or loses credibility — and what can be done about them without deploying a system.
Construction dashboard: 12 data points that need an owner
A ready list of fields without which a construction management dashboard shows opinions instead of state. For each field: responsible role, source, and update frequency.