What a price list does not tell you
The licence is the visible part. Migration, training, per-user pricing and the cost of leaving are larger, arrive later, and are rarely quoted at all.
The licence fee is the part of the cost that arrives first, is easiest to compare and is smallest.
This post has no figures in it, because the figures are yours and they vary enormously. What does not vary is the shape of the number. Once you know the shape, you can ask for the parts nobody volunteers.
The costs that appear on somebody's invoice#
Implementation and configuration. Setting up your companies, sites, cost heads, item masters, tax settings, approval limits and user roles. Sometimes included, often quoted separately, occasionally quoted as an estimate that grows. Ask whether it is a fixed price or a day rate, and what happens if the work takes longer than expected.
Data migration. This is nearly always underestimated, and the reason is not technical. It is that the old data is dirty. The same supplier exists three times with three spellings. Two item codes mean the same steel. Old balances do not tie. Somebody has to decide what dirty means and what to do with each case, and that somebody must be from your side, because only your side knows which of the three spellings is the real vendor. Budget for your own people's time, not just the vendor's.
Training, and retraining. The first round is planned. The second is not. Site staff turn over, particularly after a season, and a system nobody was taught is a system that gets bypassed. Ask what a training session costs after the implementation period ends, because you will buy several.
The per-user question. Ask early what a user costs, and then ask a different question: what does it cost to put the storekeeper, the gate guard, the site engineer and the supervisor on it. A per-user price is a tax on getting the record from the place it actually happens. It pushes you towards shared logins, which destroy the audit trail, or towards paper at the gate typed up later in the office, which is the failure the software was bought to fix. Ask whether there is a lighter, cheaper class of user for people who only record events.
The people who now do data entry that did not exist before. Every system creates keystrokes. If the answer to the per-user question is bad, some of those keystrokes are done by a new person in the office whose entire job is retyping what the site wrote on paper. That is a permanent salary caused by a pricing model.
Integration. Accounts, banking, attendance, whatever else you already run. Ask what is included, what is a one-time build, and who maintains it when either side changes.
The annual increase. Almost every agreement has one, and it compounds quietly for the entire life of the system.
The cost of leaving. Extraction, cleaning the extract, loading it elsewhere, and running two systems for a period. This is invisible at purchase and it is the reason people stay with tools they dislike, so it deserves a question at the start rather than at the end — the tests are in getting your data back out.
The costs nobody invoices#
These are larger than the ones above and they never appear in a comparison.
The parallel register nobody stopped keeping. If the notebook at the gate is still being filled, you are paying for the record twice: once in licence fees and once in the storekeeper's evening. Worse, you now have two histories of the same events, and when they disagree the argument costs a day. This is the single clearest signal of a failed adoption, and why it belongs in how to choose construction software.
The process bent to fit the software. A system that cannot express a disputed quantity teaches the office to record undisputed ones. A system that cannot hold a short delivery teaches the storekeeper to accept the full figure and telephone about the shortage. Nothing is invoiced. You simply have a slightly less honest archive from that month onwards, and you find out during a dispute two years later.
A slow screen multiplied. Take the commonest transaction in your office and count how many times a day it happens across all sites. Now add ten seconds to it. That number is a salary, and it recurs every year, and it never appears in any comparison because nobody measured the screen.
Support you cannot reach at the hour you need it. A lorry at the gate at nine in the evening is not a business-hours problem.
How to ask about all this#
Three requests, all reasonable, all in writing.
- A three-year total, itemised. Licence, implementation, migration, training, integration, support, and every one-time item named separately. Not a monthly figure. Vendors quote per month because it is small; you are buying three years at least.
- What the renewal increase has been for existing customers. Not the cap in the contract — the actual history. A vendor who will not answer this has answered it.
- What it costs to add a person for one month at a peak. Construction is seasonal and lumpy. If adding fifteen temporary users for a busy quarter means an annual commitment for all fifteen, you will not add them, and the record will be made on paper instead.
One more worth asking, though it rarely has a clean answer: what is the price if we double our sites, and is that in writing anywhere.
What actually decides the return#
Not the feature set. Adoption.
A tool that costs more and is used by everyone at the gate produces a complete record. A cheaper tool used by three people in the head office produces a partial one, and a partial record is not proportionally useful — it is a record you cannot trust anywhere, because you never know which events are missing.
So the money question and the legibility question are the same question. If the storekeeper cannot read the screen, no licence price is cheap, which is the argument in software your staff can read. And the only honest way to find out whether people will use it is a trial designed so that they might not — a pilot that tells you something.
Say it plainly to whoever is comparing quotations: the cheapest tool that is used beats the best tool that is not.
The short version#
The licence is the visible part of a number whose larger parts are migration, training, per-user pricing and leaving. Ask for a three-year itemised total, the real renewal history, and the cost of one extra person for one month.
Then price the things nobody invoices — the parallel register, the bent process, the slow screen — and remember that adoption, not features, decides whether any of it was worth spending.