31 March 2026 snapshot Published by NISTA · analysed 15 July 2026
Which major projects
changed direction?
ProjectLens turns disconnected public major-project disclosures into an inspectable history. See what changed, where signals conflict and which evidence deserves scrutiny.
Published portfolioOfficial releases
170 matched to last year35 across all 7 releases
Executive attention queue
Start with material change
Ranking combines published delivery confidence, year-on-year movement, end-date movement and in-year variance. It is a review queue, not a prediction.
Portfolio movement
The story is not just Red or Green
A better rating can coincide with a later published end date. ProjectLens keeps both facts visible.
Signals worth challenging
Improved confidence,
Improved confidence,
later finish.
Published narrative signals
What departments are talking about
Transparent keyword signals help organise the evidence. They do not prove root cause.
Outcome follow-up register43 records left.
43 records left.
NISTA reports 42.
Stable-ID comparison finds one more absent record than the annual report's leaver total. That may reflect an identifier or classification difference. Absence is not evidence of delivery success or failure.
Portfolio explorer 189 current projects · source-linked
Find the change.
Read the evidence.
189 projectsClick a row for up to seven years of evidence
| Priority | Project | Published DCA | Movement | End-date change | Published themes | Watch |
|---|
Comparable case finder Evidence before analogy
What happened
to similar projects?
Select a current project. ProjectLens searches projects that improved in a later annual release and explains the overlap.
Human approval required. Similarity is based on shared themes, category and department. It is a prompt to inspect evidence, not proof that the same intervention will work.
Open methodology Exact maths, bounded language
No mystery score.
No invented forecast.
ProjectLens uses deterministic comparisons for dates, ratings and variances. Text classification is deliberately simple and inspectable.
Acquire
Seven official GMPP CSV releases are retained unchanged with links to the source publication.
Match
Stable GMPP identifiers create longitudinal histories. Missing identifiers are kept but marked unmatchable.
Compare
Rating transitions, published end-date movements and annual forecast variances are calculated exactly.
Organise
A visible keyword taxonomy groups narrative signals into eight review themes.
Attention scoreA queue for scrutiny,
A queue for scrutiny,
not probability of failure.
The score deliberately prioritises obvious published signals. It does not claim to predict project outcomes.
+35Current Red published DCA
+25Rating worsened
up to +20Later end date
up to +15Material in-year variance
−8No published evidence excerpt
Model boundaries
What this cannot tell you
Source register
Inspect the original releases
Evidence expansion
Connect delivery to scrutiny, procurement and outcomes
These authoritative routes define the next evidence layer. They remain separate so ProjectLens never presents a generic source as project-specific proof.