Detection is cheap relative to exposure. Exposure is cheap relative to a strike or a redesigned pile cap. The skill is knowing which side of that sentence you are on for each crossing.
Open a trial hole when a proposed excavation will occupy the same space as a service whose depth or alignment would change the design if it were half a metre out. Keep the rest of the site at QL-B. “Pothole everything” is a template, not a specification.
Non-intrusive methods still come first. Without a Type D plot and a Type B drawing you are choosing hole locations from folklore. That is how teams open the ground in the wrong place and then declare the service “not there”.
HSG47 governs the act of exposing. PAS 128 governs the record. A hole that is not surveyed into the grid is a story for the tea hut. A QL-A segment with photographs and measurements is evidence.
Read trial hole verification for how this practice scopes Type A, and quality levels for the labels you should see before and after the hole is opened.
Vacuum excavation and hand digging are methods, not brands of certainty. The quality level comes from the measurement of what you expose, provided the safe system was competent. Do not buy a machine name and forget the survey grid.
Reinstatement is part of the decision. A hole in a live carriageway that cannot be reinstated to the authority’s rules is a hole you cannot open yet, however badly the designer wants a number.
Clusters of holes along a route can still be justified — for example a new connection that crosses the same street three times. That is still a list of reasons, not a percentage of the site.
If detection and records already agree, and the design can live with the remaining uncertainty, keep the money for the two crossings that do not agree. That is the whole essay.
Write the candidate holes on the Type B drawing before you tender the excavation. A list in an email will drift. A marked-up model will not.
Attend the hole if the crossing is the one that controls a building. Photographs are good. A designer who saw the invert is better.
Do not let Type A become a way to skip Type B. Exposure without detection is a random trench. Detection without the option of exposure is how designs freeze around an interpretation. You need the sequence, not a favourite method.
QL-A is a measured fact about a small piece of ground. It is not a blessing on the rest of the site. Keep that sentence in the progress meeting when someone asks whether “we potholed it” means the whole red line is now certain.
Take this into the brief
Write the candidate holes on the Type B drawing. A list in an email will drift toward “pothole everything” the first time a director gets nervous. A marked model keeps the reasons attached to crossings.
QL-A is a measured fact about a small piece of UK ground. It does not bless the rest of the site. Keep detection for the area, exposure for the crossings that would change a design, and HSG47 for the way you open the ground. That three-part sentence is the whole comparison.