View 1 · Portfolio
Corpus overview
What the seven-year archive actually contains — and where it is thin. Every figure is measured from the extracted records, not asserted. This is a bid archive: nearly all effort was worked back from price ceilings, so read it as bid unless marked actuals.
Opportunities
147
9 non-opportunity records held aside
Bid years
2021–2026
15 undated
Functional domains
19
keyword taxonomy
With role-level staffing
101
97 carry a person-month total
Records flagged actuals
2
the only calibration-grade data in the archive
Total archive value
$1.50B
Assumption competing variants and template carryover inflate this
Opportunities per bid year
15 undated, excludedArchive value by domain
top 8 by summed TCV Derived MetricContract value vs total person-months
the single most informative view in the corpusBid effort (worked back from price)Mixed (bid + internal work plan)Delivered actuals
95 of 147 opportunities carry both figures. 50 have no person-month total at all — effort reads “Not available in source data” rather than zero. 6 points share byte-identical totals carried over from an uncustomised template and do not describe their own deal — see Data gaps.
Person-months vs DDI duration
fitted · slope 62.54 · intercept -509 · R² 0.512 (n=67)⚠ This line describes pricing, not delivery
Slope reads as average concurrent FTE. Because every point except one is BID effort, this line describes how Cardinality prices, not how work behaves. Do not present it as a predictive parametric model.Derived Metric Slope reads as average concurrent FTE. Axes follow the source fit: x = duration_months_ddi, y = total_person_months.
Field coverage
60 fields audited · worst 26 shownWorst-covered first, so nobody mistakes this for a complete dataset. The zero-coverage rows are not missing by accident — interface, integration, report, form, data-source and site counts do not exist as structured fields anywhere in the source workbooks. See Data gaps & decisions.