Most lightning protection certificates that cross our desk are too short. A single page saying the system was inspected and found satisfactory tells the building owner nothing, tells the insurer less, and gives next year's engineer no baseline at all. A certificate is a record of measurement, not a receipt.
Start with the individual readings. Every earth electrode should be listed by reference, with its measured resistance, taken with a calibrated instrument and the method noted. The three pole fall of potential method is the reference method where the ground allows the spikes to be placed; clamp measurement is a reasonable alternative on constrained sites, but the certificate should say which was used, because the numbers are not directly comparable.
Then the combined value. Individual electrode readings mean little on their own; what matters is the resistance of the earth termination system as a whole, calculated or measured for each protected structure. A building with several electrodes in parallel can show a healthy combined value while one electrode is quietly failing. Both figures need to be on the page, which is why the individual readings matter even when the overall result is a comfortable pass.
Continuity comes next: the bonds at each down conductor position, the connections to roof plant, ductwork, handrails and cable containment, the joints and the test clamps. Then the visual inspection, recording corrosion, mechanical damage, missing or loose fixings, tape lifted by roofing works, new plant installed since the last visit and never bonded, and any alteration to the building that changes the protected volume. New rooftop equipment appearing without anyone telling the lightning protection contractor is by some distance the most common defect we find.
A good certificate also records what is not compliant and does not pretend otherwise. Legacy early streamer emission systems, for example, do not conform to BS EN 62305 and must be reported as such rather than tested as if they did. Where a reading is drifting year on year but still passing, it should be flagged as attention with a recommendation and, ideally, a price, so it becomes a planned remedial in next year's budget rather than a surprise failure.
Finally, the administration: the instrument serial number and calibration date, the engineer, the date of test, the structures covered, and the next due date. Annual testing is the norm for most systems, with the interval shortened for critical structures. One certificate per site, covering every structure, is far easier to file than seven certificates covering one building each.
The commercial case for keeping a proper test book is not about the standard. It is that a documented, measured history of a system costs a fraction of one strike related outage, satisfies insurers without argument, and means that when a building is sold or refinanced the compliance file is already complete. The certificate should never be the thing you cannot find.