Legacy Modernization Index · 2026 edition
The numbers behind legacy modernization, with sources.
A short, sourced index of what is publicly known about legacy code footprint, technical debt, public sector spend and migration consequences, next to PYD ENTERPRISES' own discovery model, clearly labelled.
- lines of COBOL estimated in daily production use worldwide
- 800B+
- lines of COBOL estimated in daily production use worldwide
- Vanson Bourne survey for Micro Focus, February 2022 (estimate range 775 to 850 billion)
- accumulated software technical debt in the United States
- $1.52T
- accumulated software technical debt in the United States
- CISQ, The Cost of Poor Software Quality in the US: A 2022 Report
- of the US federal FY2015 IT budget went to operating and maintaining existing systems
- ~75%
- of the US federal FY2015 IT budget went to operating and maintaining existing systems
- US GAO, GAO-16-468, Federal Agencies Need to Address Aging Legacy Systems (May 2016)
- combined FCA and PRA fine on TSB over its 2018 core platform migration
- £48.65M
- combined FCA and PRA fine on TSB over its 2018 core platform migration
- UK FCA and PRA final notices, 20 December 2022
Explorer
Four themes. Every figure linked to its source.
Switch themes to see each figure in context. Move the estate slider to apply PYD ENTERPRISES' planning model to your own line count.
COBOL IN DAILY PRODUCTION USE · BILLIONS OF LINES
of surveyed COBOL code runs on IBM z/OS. The remaining 57% runs on other platforms, which is why discovery tooling has to handle Micro Focus dialects as well as Enterprise COBOL.
The installed base is measured in hundreds of billions of lines.
The survey estimate ranges from 775 to 850 billion lines in daily use. It is an estimate from respondents, not a census, and we show the full range rather than a single figure.
Parse time for your estate
PYD ENTERPRISES model figure~2 h
2M LOC at about 1 million LOC per hour on one 32-vCPU node (PYD ENTERPRISES measured throughput). Rule extraction and synthesis add a few hours.
PYD ENTERPRISES model figures
Our own planning numbers, kept apart from the research.
These come from PYD ENTERPRISES' engagements and benchmarks. Use them to scope a discovery phase; do not cite them as industry statistics.
- Manual discovery cost
- $0.40 to $0.75 per LOC
- PYD ENTERPRISES engagement benchmarks
- Parse throughput
- ~1M LOC per hour
- One 32-vCPU node, PYD ENTERPRISES measured
- Discovery timeline
- ~9 months to ~3 weeks
- PYD ENTERPRISES engagements
Methodology
How the index is built.
Five rules, applied to every figure. If a number does not meet them, it is not here.
License: CC BY 4.0
The index text, charts and compilation are published under the Creative Commons Attribution 4.0 International license. Reuse them with attribution to PYD ENTERPRISES and a link to this page. Third-party figures remain the work of their publishers.
- 01Only sources we can linkEvery external figure on this page comes from a published report or regulatory notice with a public URL. We cite the publisher, the date and the scope of each figure exactly as published.
- 02Ranges stay rangesWhere a source publishes an estimate range, such as 775 to 850 billion lines of COBOL, we show the range and the point estimate together, so readers can see the uncertainty.
- 03Different measures are not combinedCISQ's annual cost of poor software quality and its accumulated technical debt figure measure different things. We never add, scale or extrapolate figures across sources.
- 04PYD ENTERPRISES figures are labelledThe discovery cost per line, the parse throughput and the timeline compression come from PYD ENTERPRISES' own engagements and benchmarks. They are tagged as PYD ENTERPRISES model figures wherever they appear.
- 05What we left outModernization failure percentages circulate widely in industry decks. We could not trace them to a methodology we can cite, so the index does not include them.
Sources
| Source | Published | Figures used |
|---|---|---|
| Vanson Bourne for Micro Focus, COBOL usage survey | February 2022 | 800B+ lines in daily use (range 775 to 850B); 43% on IBM z/OS |
| CISQ, The Cost of Poor Software Quality in the US: A 2022 Report | 2022 | $2.41T cost of poor software quality; $1.52T accumulated technical debt |
| US GAO, GAO-16-468: Federal Agencies Need to Address Aging Legacy Systems | May 2016 | ~75% of FY2015 federal IT budget on operations and maintenance |
| UK FCA and PRA final notices on TSB Bank plc | 20 December 2022 | £48.65M combined fine over the 2018 migration |
Last reviewed 2 November 2026. No figures changed in this review.
FAQ
Using the index.
Can I cite the Legacy Modernization Index?
Yes. The index is published under the Creative Commons Attribution 4.0 International license (CC BY 4.0). Cite it as PYD ENTERPRISES Legacy Modernization Index (2026) with a link to this page, and cite the original publisher for any external figure you reuse.
Why is there no modernization failure rate?
Because we could not find one with a methodology we are comfortable citing. Several percentages circulate in vendor material without a traceable survey design, sample or definition of failure. We would rather publish fewer numbers than repeat ones we cannot source.
Where do the PYD ENTERPRISES model figures come from?
From PYD ENTERPRISES' own engagements: the cost of manual discovery per line of code observed on programs we have benchmarked against, our measured parser throughput on a 32-vCPU node, and the elapsed time of discovery phases with and without the platform. They are planning figures for scoping, not industry statistics.
How often is the index updated?
We review sources when a new edition of a cited report is published and at least once a year. Changes are listed with the date at the bottom of the methodology section.
Replace the estimate with your own inventory.
Run PYD ENTERPRISES on up to 100,000 lines in your private VPC and see your actual rule count, lineage and interfaces.