Most smart building projects that disappoint were not badly installed. They were badly started. The technology arrived before anyone had written down, in plain language, what the building was supposed to do differently once it was finished, and every later argument traced back to that missing page.
The order that works
- Write the owner’s requirements first, in outcomes rather than product names.
- Scope against a recognised checklist so the brief is complete rather than fashionable.
- Prove one measurable use case on one zone before committing a portfolio budget.
- Put data ownership, configuration handover and remote access in the contract, not in an email.
Start with a document, not a device
The discipline already exists and is borrowed from commissioning practice. ASHRAE Guideline 0, The Commissioning Process, is built around the Owner’s Project Requirements, a document stating what the owner expects the facility and its systems to do, against which the design, the installation and the training are all later verified. The design team responds with a Basis of Design, and a Commissioning Authority verifies that what was built matches what was asked for.
Written properly, that document contains sentences a facilities manager can test. Ventilation matches actual occupancy in the meeting rooms. Out-of-hours plant operation is visible within a day rather than at the quarterly review. A tenant can be given and revoked access to a floor without a site visit. None of those name a supplier, and all of them can be demonstrated or failed on site.
Written badly, it contains ambitions instead: a connected building, a data-driven estate, a platform for future services. Those cannot be tested, which means they cannot be refused at handover either.
Scope against a checklist rather than a wish list
For anyone working on European non-residential stock, there is a ready-made scoping instrument. The recast Energy Performance of Buildings Directive, Directive (EU) 2024/1275, already requires building automation and control systems in non-residential buildings above defined output thresholds, with the obligation broadening later this decade. CEN published CEN/TR 18276:2026, a checklist of compliance of building automation with EPBD requirements, aimed at policy makers, planners, owners and inspectors.
Used early, a checklist of that kind changes the character of the brief. Instead of asking a supplier what their platform can do, the owner arrives with a list of functions the system has to demonstrate, which is a far stronger negotiating position and a far better basis for comparing two bids that look nothing alike on paper.
Prove one use case on one zone
The most reliable way to lose credibility internally is to instrument an entire estate before anything has been shown to work. A pilot confined to one floor, one plant room or one building answers the questions a business case cannot: whether the existing cabling and network can carry it, whether the integrator understands the site, whether the facilities team will actually use the interface, and what the measured effect is against a baseline rather than a projection.
A pilot also produces the numbers for the second phase. Costs per point, hours of integration labour, the real duration of commissioning: these are the figures that make a portfolio budget defensible, and they are almost never transferable from a vendor case study on a different building.
Budget the lines that usually get missed
Hardware is the visible cost and rarely the one that derails a project. The following lines appear on every project we look at and on very few initial budgets.
| Budget line | Why it appears | When it hits |
|---|---|---|
| Network and cabling | Existing infrastructure rarely reaches the plant | During installation |
| Integration labour | Legacy plant speaks its own dialect | Throughout, and over estimate |
| Licences and subscriptions | Platforms are rented, not bought | Annually, from year one |
| Cybersecurity and remote access | The system joins the attack surface | At design, then continuously |
| Commissioning and training | Sequences need tuning and people need to read them | At handover and beyond |
Four clauses worth arguing about before signature
- Data ownership and export. The trend data belongs to the owner, in a documented format, retrievable without the supplier’s cooperation.
- Configuration and points list at handover. The as-built points list, the control sequences and the configuration files are deliverables, not the integrator’s private working papers.
- Open protocols named explicitly. Specify the communication standards, BACnet being standardised as ISO 16484-5 and KNX as ISO/IEC 14543-3, rather than accepting a brand as a substitute for interoperability.
- Remote access under the owner’s control. Vendor access is time-limited, logged, and revocable by the owner rather than permanently open for convenience.
That last clause is not theoretical. The KNXlock incident of October 2021, documented by Limes Security in a German office building, saw an attacker reach the building network from the internet and lock a large share of several hundred devices, leaving lighting, presence detection and blinds inoperable and the building run from its circuit breakers for weeks. The recurring vector in cases of this kind is remote access left open for an integrator’s convenience.
Who owns it on the Monday after handover
A smart building is an operated asset, not a delivered one. Somebody inside the organisation has to hold the alarm list, decide which alerts require a response, review trend data against the baseline, and authorise changes to sequences. Where that role is unassigned, overrides accumulate, alarms get muted, and within two or three years the system reports green on a building nobody is actually managing.
This is also the argument for keeping the first phase small. A facilities team that has run one floor successfully knows what it is taking on when the estate follows. One that inherits an entire instrumented portfolio at handover generally does not.
Questions the steering group will ask
Does this only make sense for a new build? No. Retrofit is where most of the stock is, and the constraint is usually the state of the existing plant, cabling and network rather than the age of the building itself. It does change the sequence, since a survey of what is already installed becomes the first cost rather than an afterthought.
How long does a realistic first phase take? Long enough to see the building through more than one season. A pilot judged on six weeks of autumn data will mislead in both directions, which is why the baseline conversation belongs at the start rather than at the review.
Should the platform decision come first? We would put it last among the major decisions. Once the requirements, the scope and the acceptance tests exist, the choice of platform becomes a comparison against a specification instead of a reaction to a demonstration.
Two directions from here
The wider context of what a smart building is now expected to deliver, and the specific discipline of getting the energy layer right.
The energy strand of the project has its own sequence, covered in our guide to implementing a building energy management system.
Published in 2025. Updated on 13 August 2026. Sources: ASHRAE Guideline 0, The Commissioning Process; Directive (EU) 2024/1275; CEN/TR 18276:2026; ISO 16484-5; ISO/IEC 14543-3; Limes Security reporting on the KNXlock campaign, 2021. National transposition and local regulation govern any specific project.
