Early sites get attention, senior oversight, and careful documentation. The failure mode is drift — small inconsistencies that compound quietly until nobody can prove what was installed where.
Programs rarely fail loudly. They drift, quietly, until nobody can prove what was installed where.
The first four sites of a deployment program are the best work the program will ever produce. A senior project manager is watching. The install standard is fresh in everyone's mind. Photographs get taken because someone asked for them that morning.
By site forty, none of that is true. The senior attention has moved to the next program. The standard has been relayed through three layers of subcontractor. And the photographs — if they exist — are in an inbox somewhere, unnamed, unlinked to a location.
The technician at site forty is not less competent than the one at site four. They are working with less context. Nobody told them that the mounting height was revised after site eleven, or that this region's locations have a different cabling path, or that the customer now wants a serial number capture on every unit.
What separates programs that hold their standard is not heroics at the far end. It is instrumentation put in place before the first technician is dispatched — a defined evidence requirement, a single place it lands, and a review cadence that catches deviation while it is still cheap to correct.
On multi-site programs we set the evidence requirement during survey, not during install. By the time work is scheduled, every technician receives the same capture list and the same submission path — and a missing item is an open exception, not a discovery at acceptance.
Name every photograph, measurement, serial capture, and signature required at a location. Ambiguity here becomes inconsistency at scale.
One destination, one naming convention tied to the site identifier. Email attachments are how documentation goes missing.
Decide in advance what counts as an exception and who adjudicates it. Without a threshold, everything is escalated or nothing is.
Weekly sampling against the evidence set. The purpose is to catch drift at site twelve, not to audit at site sixty.
Failure at scale is drift, not incompetence — the far end of a program is working with less context, not less skill.
Evidence requirements belong in the survey phase, where they are cheap to define and impossible to forget.
A weekly sample beats a closeout audit by fifty sites and several months of rework.
