Every review process has gaps. The question is whether your reports show them, or bury them.
‘Not yet checked’ is not a problem. A false all-clear is.
AEC teams are under constant pressure to move fast. Packages land late. Programmes slip. The review window shrinks. And in that squeeze, status fields that should read Not Yet Checked quietly get marked Compliant instead: not through dishonesty, but through pressure and the belief that a bare guarantee looks safer than an honest blank.
It doesn’t.
Why a bare guarantee is unsafe.
A review that says everything is fine without showing the evidence is not a confident review. It is an asserted one.
An assertion can be wrong in ways nobody can trace. An evidence-backed finding can be interrogated, challenged, and updated when conditions change. When something goes wrong on site, or at Gateway 2, the question will not be “did you sign off?” It will be “what did you actually look at?”
Dame Judith Hackitt’s 2018 Independent Review of Building Regulations and Fire Safety identified a systemic pattern across UK construction: a culture of doing just enough rather than genuinely owning safety outcomes. Its core finding was not that teams were dishonest, it was that the system gave them no incentive to be transparent about what they hadn’t confirmed.
Key takeaway
Dutyholders under the Building Safety Act carry personal competence liability. A bare sign-off without a traceable record is no longer a safe position, legally or professionally.
Confidence matched to reality.
The alternative to a false all-clear is not paralysis. It is calibrated confidence: assurance that honestly reflects what the review team did and what evidence they examined.
A calibrated report holds four distinct states:
- High requirement reviewed, cross-referenced to the design, citation recorded.
- Partial reviewed but one or more conditions outstanding.
- Low reviewed against available information; a dependency or gap remains.
- Not yet checked in scope, not yet reviewed; exposure is known and visible.
None of these are failures. A report showing three items at low confidence and two not yet checked is more useful than one marking all five compliant without evidence: it tells you exactly where to focus.
In short
This is the core principle of Assurance Engineering: treat quality as a volume you raise and calibrate, not a verdict you stamp at a gate.
An honest ‘not yet checked’ vs a false all-clear.
Not Yet Checked is not a red flag. It is a piece of information, and information is actionable.
A programme manager can see it, resource it, and track it to closure. A design director can weigh it against the programme and decide whether to proceed with eyes open. A false all-clear removes that decision entirely. The problem doesn’t disappear; it becomes invisible. And invisible problems reach site.
An honest Not Yet Checked also enables continuous review. When every gap must be closed before anything can be reported, teams hold back until the last moment and submit a bulk review that creates the very bottleneck they were trying to avoid.
Evidence-backed rather than asserted.
Calibrated confidence only holds if it is grounded in evidence. A confidence level without a citation is just a different kind of assertion.
Evidence-backed assurance means:
- Citing the source the document, clause, drawing revision, or model element behind the finding.
- Recording the reasoning not just what was found, but what the reviewer concluded.
- Linking to the requirement so it is clear which item in the brief, spec, or regulation was assessed.
A review record survives staff turnover, withstands regulatory scrutiny, and passes intact to successor teams. A review opinion does none of these.
How teams miscalibrate under pressure.
Miscalibration is rarely deliberate. It happens in three recognisable patterns.
Over-confidence under time pressure. Reviewers mark items compliant based on experience with similar schemes, not on this scheme’s specific evidence. Experience is valuable. It is not a substitute for checking.
Undifferentiated scoring. A reviewer who checked ten items thoroughly but left fifteen untouched rates the whole review low confidence, obscuring the work that is in excellent shape. Treating both groups the same destroys signal.
Consensus drift. In multi-discipline reviews, confidence levels migrate toward the most optimistic position through social pressure, not evidence. The report ends up reflecting the team’s mood more than the design’s state.
All three share the same root: no mechanism forces ratings to reflect what was actually reviewed. Informal grading drifts.
Reading a graded assurance report.
A graded assurance report is a decision-support instrument, not just an output. When reading one, look for:
- Coverage first. What share of in-scope requirements have actually been reviewed?
- Confidence distribution. Is confidence concentrated on easy items, or spread across hard ones?
- Gap transparency. Are Not Yet Checked items visible and assigned, or have they disappeared?
- Evidence depth. Can you trace each rating to a specific document version and clause?
- Residual exposure. What stays unresolved under the signature, shown explicitly as a bounded position?
| Approach | Pass / fail | Graded assurance |
|---|---|---|
| Gap visibility | Hidden; every item scores 0 or 1 | Visible; the in-between is held in view |
| Not yet checked items | Often marked compliant to avoid blanks | Shown explicitly and assigned for follow-up |
| Evidence trail | Optional; assertion-based sign-off common | Required; each rating links to source and clause |
| Regulatory defensibility | Depends on who signed, not what was checked | Traceable record survives scrutiny and handover |
What this means in practice.
Tektome produces graded, evidence-backed assurance records: confidence levels tied to cited sources, honest Not Yet Checked statuses held in view, and residual exposure shown explicitly under every finding. It is not a pass/fail checker; it is an assurance instrument for teams who know that what they haven’t confirmed is exactly as important as what they have.
If your review process produces all-clears you can’t fully defend, or buries gaps that should be visible, that is a calibration problem. It is also a solvable one.
See it on your projects
Tell us what you’d like to verify and we’ll set you up with a walkthrough tuned to your data and disciplines, including what graded assurance would look like on your next Gateway 2 submission.
Book a demo