Choosing construction management software: a buyer’s checklist

Every construction platform demos well. The dashboards are clean, the mobile app works on the presenter’s phone, and the sample project has exactly the right amount of data in it.

These are the questions that actually distinguish them, drawn from what tends to go wrong six months after signature.

Do the registers reference each other?

This is the question that matters most and gets asked least. A platform where documents, correspondence, submittals, variations and the programme are separate modules that happen to share a login is a filing system with extra steps.

Ask them to show you a delay event, and from it navigate to the instruction that caused it, the correspondence around it, the affected programme activities and the cost impact — without leaving the platform or exporting anything. If that takes more than a minute, the links are not really there.

Who owns the data, and how do you get it out?

• Can you export the complete project record, including attachments and metadata, without paying for it?
• In what format — is it usable, or a proprietary blob?
• What happens to your access when the subscription ends? Read-only access for a defined period is reasonable; a hard cut-off is not.
• Where is the data hosted, and can you nominate the region? On Gulf projects with data residency requirements this can be decisive.

What happens on site, without signal?

Basements, plant rooms, remote infrastructure and the middle of a large slab all have poor connectivity. Ask specifically: can an inspection be recorded offline, with photographs, and sync later? Can the current drawing revision be cached on the device?

A platform that requires connectivity for field work will be used from the site office an hour later, from memory — which is how records become approximations.

Can the audit trail be edited?

Ask whether an administrator can delete or amend a log entry. If the answer is yes, the log is a convenience feature rather than evidence, and it will not help you when a position is challenged.

The follow-up question is what the log actually records. Uploads and edits are the easy part; access events — who opened a document, and when — are frequently the decisive fact in a dispute.

How does licensing scale, and what is not included?

• Per user, per project, or per volume? Per-document charges are worth avoiding — the register grows with the project, and so does the bill.
• Do consultants, subcontractors and client representatives need paid licences to review and respond?
• Is storage capped, and what happens when you exceed it?
• Which modules are excluded from the tier being quoted?

What does implementation actually involve?

Ask who migrates the existing registers, who configures the numbering convention and approval codes, and how long a single project takes to go live. Two weeks for one project is realistic; anything promising same-day is describing a signup, not an implementation.

Ask what integration is included. Most projects need the platform to talk to an ERP, a planning tool or a document system, and whether that is scoped during onboarding or sold separately makes a material difference to the real cost.

If it uses AI, who is accountable?

AI drafting is genuinely useful for high-volume correspondence. The question is whether the platform can issue anything without human approval. If it can, accountability has quietly moved to a system that cannot hold it.

Ask whether approval can be switched off, and treat “yes” as a warning rather than a feature. Ask what the log records about who reviewed each draft and what they changed.

A shortlist test

Give each vendor the same real scenario from your own project — a live variation, a submittal backlog, an EOT you are substantiating — and ask them to run it in the demo. Generic demos are equally impressive. Your scenario is where platforms separate.

If you would like to run that test with us, book a walkthrough and bring the awkward one.