Site selection for data centre development begins with a complete technical specification — the parameters that every candidate site must be assessed against. Power requirement, rack density, cooling type, redundancy level, resilience tier, latency constraints, and geographic boundaries are all inputs to the scoring framework. Sites that look similar on the surface — same region, similar acreage, similar planning status — can diverge significantly when scored against a specific technical specification. A site with excellent planning prospects but a five-year grid connection timeline fails a programme with a two-year delivery requirement regardless of its other merits.

Desktop scoring provides the first filter. A weighted matrix applied consistently across all candidate sites identifies which sites warrant further investigation and which can be eliminated on the basis of publicly available information. This matters because physical site visits and M&E consultant involvement are not free — the desktop scoring stage is specifically designed to protect that budget by eliminating clearly unsuitable sites before any paid physical assessment is incurred.

Site visits are non-negotiable for any programme where high-density requirements or liquid cooling creates parameters that desktop assessment cannot verify. A facility designed for 50kW per rack with direct liquid cooling has infrastructure requirements that cannot be confirmed from a desktop review — the physical configuration of an existing building, the available ceiling height for cooling plant, the floor load capacity, and the proximity of existing utilities all require physical verification.