Hold Point vs Witness Point — What's the Difference?
Two terms that get used loosely, but mean genuinely different things — and mixing them up on an Inspection and Test Plan creates real risk on a live programme.
Both hold points and witness points are ways of flagging that a particular stage of work needs outside attention before — or shortly after — it happens. The difference is what's allowed to happen if that attention doesn't show up.
The core distinction
| Classification | What it means |
|---|---|
| Hold point | Work may not proceed past this point without formal, documented sign-off from the named party. Proceeding without it is a non-conformance in its own right, regardless of the outcome of the underlying work. |
| Witness point | The named party must be given notice and the opportunity to attend, but work may proceed if they don't attend after proper notice has been given and recorded. |
In other words: a hold point stops the project if the sign-off doesn't happen. A witness point invites attendance but doesn't block progress if that attendance doesn't materialise, provided notice was genuinely given.
Why the distinction matters
Getting this wrong in either direction creates a real problem. Classify too many things as hold points, and the programme grinds to a halt waiting on sign-offs that don't need to be blocking. Classify something as a witness point when it should have been a hold point, and safety-critical or contractually critical work can proceed without the oversight it actually needed.
The judgement call is usually about consequence and reversibility — first energisation of a distribution board is a hold point almost everywhere, because the consequence of getting it wrong is severe and not easily undone. A routine cable termination check is more often a witness point, or even just a review point with no attendance requirement at all.
A third category worth knowing: review points
Many commissioning frameworks also use a review point classification — documentation is checked after the fact, with no notice or attendance requirement at all. Useful for lower-risk activities where an audit trail matters but real-time oversight doesn't add much.
How to classify inspection points on an ITP
A practical approach: for each inspection point, ask what happens if nobody from the required party is available. If the honest answer is "the work simply can't proceed," it's a hold point. If the answer is "we'd rather they were there, but it's not the end of the world if they can't make it," it's a witness point. If the answer is "we just need a record of it happening," it's a review point.
Worked example: classifying common inspection points
Here's how that judgement plays out in practice, using activities that show up on almost every data centre commissioning programme:
| Activity | Typical classification | Why |
|---|---|---|
| First energisation of LV distribution board | Hold point | Consequence of an undetected fault is severe and not easily undone |
| Generator load bank test | Hold point | Contractually significant milestone, client typically wants to witness it directly and sign off before it's accepted |
| Cable termination torque check | Review point | Low individual risk, but an audit trail matters — checked after the fact, no attendance required |
| Insulation resistance test on switchgear | Witness point | Client's CxA representative should ideally attend, but work can proceed with proper notice if they can't |
| Integrated systems testing (IST) scenarios | Hold point | Represents the facility's core resilience proof — sign-off required before the facility is considered ready |
| Routine BMS point-to-point verification | Review point | High volume, low individual risk — a sampled review is usually sufficient |
Note that these are typical classifications, not universal rules — the right call always depends on the specific project's risk profile, contract requirements, and client expectations. A hold points matrix, agreed early with the client, removes the ambiguity before it becomes a dispute mid-programme.
Frequently asked questions
What is a hold point in commissioning?
A hold point is a stage in the commissioning process where work must formally stop until a named party has reviewed and signed off. Proceeding past a hold point without that sign-off is itself a non-conformance, regardless of whether the underlying work turns out to be correct.
What is a witness point in commissioning?
A witness point is a stage where a named party must be given notice and the genuine opportunity to attend, but work is allowed to proceed if they don't attend after proper notice was given and recorded. It invites oversight without making it a hard blocker.
Can a witness point become a hold point?
Yes — this happens in practice when a witness point reveals something unexpected. If a witnessed test result is inconclusive, out of tolerance, or raises a genuine safety question, it's common practice to escalate that specific instance to a hold point until it's formally resolved, even if the activity is normally classified lower.
Who decides whether something is a hold point or a witness point?
Classification is typically agreed between the client, the commissioning authority, and the principal contractor during the design and planning stage, and recorded in the hold points matrix and the ITPs. It shouldn't be decided informally on-site partway through a programme — that's exactly the kind of ambiguity that leads to disputes about whether work was allowed to proceed.
ITP guidance and a full hold points matrix
The Hold Point toolkit includes dedicated ITP guidance covering this classification in detail, along with a pre-populated hold points matrix showing typical hold points for a facility of this type — plus the CIP that ties it all together and the folder structure it all lives in.
See the toolkit