Software development, testing, and validation support How Analyse-it is developed, tested, and released; your responsibilities as a user in a regulated environment; GAMP 5 classification; 21 CFR Part 11, ISO 15189, IVDR, and CLIA context.

Summary

Analyse-it is commercial off-the-shelf (OTS) statistical analysis software, developed following a documented software development lifecycle with extensive internal and benchmark testing before every release. We are responsible for producing numerically accurate, reliable software and for maintaining evidence of how we build and test it. You, the user, are responsible for validating Analyse-it for your specific intended use within your quality system — the shared-responsibility model that applies to all off-the-shelf software in regulated environments.

This page describes our development and testing process, summarises your validation responsibilities, and answers common questions from quality managers, regulatory affairs professionals, and auditors. It is written to support validation activities under the FDA Quality Management System Regulation (QMSR), 21 CFR Part 11, ISO 15189, ISO/IEC 17025, ISO 13485, EU IVDR, and CLIA ’88.

Regulatory frameworks

The principle that the end user is responsible for validating off-the-shelf software for its intended use is consistent across every major regulatory framework our customers work under. Analyse-it is used by IVD manufacturers, clinical laboratories, quality and process teams, and regulated research groups working under:

  • FDA 21 CFR Part 820 (Quality Management System Regulation, QMSR) — the QMSR took effect on 2 February 2026 and incorporates ISO 13485:2016 by reference. Assurance of off-the-shelf software used in production and quality management systems is covered by the FDA’s current Computer Software Assurance (CSA) guidance. See the Computer Software Assurance section below.
  • FDA 21 CFR Part 11 (Electronic Records and Electronic Signatures) — applies to the system that manages electronic records, not to a calculation tool in isolation. See the 21 CFR Part 11 section below.
  • EU IVDR (2017/746) and EU GMP Annex 11 — parallel expectations for computerised systems used in IVD performance evaluation and manufacture.
  • ISO 15189 (clause 7.3, medical laboratory accreditation) — the laboratory is responsible for confirming that commercial software performs as required before first use and after significant changes.
  • ISO/IEC 17025 (testing and calibration laboratory accreditation).
  • ISO 13485 (medical device quality management).
  • CLIA ’88 and CAP inspection requirements for clinical laboratories in the US.

In every one of these frameworks, the validation scope is defined by your intended use, not by the software vendor. The evidence we provide on this page — our development process, numerical accuracy testing, change control, and release history — supplies the vendor-side inputs to your validation file.

Computer Software Assurance (FDA CSA, 2026)

On 3 February 2026 the FDA issued its current Computer Software Assurance (CSA) guidance for production and quality management system software. It supersedes the version issued on 24 September 2025 and Section 6 of the 2002 General Principles of Software Validation, and is the current FDA-recommended approach for assuring software used in medical device production and quality management systems.

CSA is a risk-based, least-burdensome framework. Rather than applying uniform scripted testing to every feature of every piece of software, it directs validation effort toward the features whose failure could affect product quality, patient safety, or record integrity, and explicitly encourages manufacturers to leverage vendor documentation in place of recreating evidence from scratch.

This page, our development process documentation, the NIST StRD numerical accuracy benchmarks, and our release history are exactly the vendor evidence CSA expects you to leverage. For a GAMP Category 3 product such as Analyse-it, a typical CSA approach is:

  • Define intended use. Record the specific analyses you will rely on — for example, Passing-Bablok regression per CLSI EP09-A3, precision per EP05-A3, or control charting for in-process monitoring.
  • Assess process risk. Classify the intended use as either high process risk (failure could compromise product quality or patient safety) or not high process risk, and document the rationale.
  • Select proportionate assurance activities. Reference the vendor-supplied evidence on this page, and perform user-site checks — scripted or unscripted — proportionate to the risk you have identified.
  • Record appropriate evidence. Retain sufficient detail to demonstrate the software performs as intended for the identified risk, without creating documentation beyond what the risk justifies.

The same risk-based principle is reflected in EU GMP Annex 11, ISPE GAMP 5 (2nd Edition), and the separate February 2026 FDA guidance covering pharmaceutical quality systems.

GAMP 5 classification

Under the ISPE GAMP 5 framework, Analyse-it is Category 3 — Non-configured Product. Category 3 software is used as supplied, without user-defined configuration that affects computational behaviour. This is the lowest-risk software category and carries the lightest validation burden.

In practice, Analyse-it:

  • Is installed and used without configuration that affects results.
  • Produces identical output for identical inputs across installations.
  • Can be validated through a combination of vendor-supplied evidence (this page, our development process documentation, and our NIST StRD numerical accuracy benchmarks) and a focused set of user-site checks appropriate to your intended use.

Your responsibilities and ours

What we provide

Analyse-it Software, Ltd is responsible for developing, testing, and releasing software that produces numerically accurate and reliable results. We maintain:

  • A documented software development lifecycle, summarised below.
  • An extensive internal test suite — more than 1,000 proprietary multi-faceted test cases covering statistical procedures, edge cases, and regression scenarios, re-run for every release.
  • Independent numerical accuracy verification against the NIST Statistical Reference Datasets (StRD), with downloadable validation workbooks.
  • A documented release history with version numbering, so you can track changes and manage upgrades within your own change control process.
  • Access to previous versions, so you can remain on a validated release while evaluating a new one.
  • This page, reviewed and maintained as our processes evolve.

What you are responsible for

As the end user in a regulated environment, you are responsible for validating Analyse-it for your specific intended use within your quality management system. Your activities typically include:

  • Defining intended use. Document what you will use Analyse-it for — for example, “method comparison analysis per CLSI EP09-A3” or “control charting and capability analysis for in-process monitoring.” Validation scope should be limited to the functions you actually use.
  • Risk assessment. Under the FDA CSA framework, classify the intended use as high process risk or not high process risk and document the rationale. Your assurance activities and documentation should be proportionate to that risk.
  • Installation qualification. Confirm that the correct version of Analyse-it is installed on a supported version of Microsoft Excel and Windows, and that the licence is activated. Record the version numbers in your validation record.
  • Operational qualification. Run a set of test cases — scripted or unscripted, proportionate to your risk assessment — and confirm that Analyse-it produces the expected results on your system. Our NIST StRD validation workbooks cover core numerical accuracy; you can supplement them with test cases specific to the procedures you rely on.
  • Change control on upgrades. When a new version is released, review the release notes, assess whether the changes affect functions you use, and re-run the relevant operational checks. For a GAMP Category 3 product, this is generally a lightweight review rather than full revalidation.
  • Documentation. Keep records of the above, and of any issues and their resolution, at a level of detail proportionate to the risk identified, as required by your quality system.

This split of responsibilities is standard across the industry. No commercial statistical software — Minitab, JMP, SAS, SPSS, or others — comes pre-validated for your intended use. The software vendor provides the evidence of development rigour and testing; the regulated user validates it for their application.

Our development process

Analyse-it is developed by a small, experienced team working to a structured release cycle. Major releases are planned on a nine- to twelve-month cadence, with maintenance releases in between as needed.

Development tools

LanguagesC#, with numerical libraries in C++ and FORTRAN
Development environmentMicrosoft Visual Studio
Source control and code reviewGit
Requirements, issue, and defect trackingAzure DevOps
Unit and regression testingNUnit, JetBrains dotCover, JetBrains dotTrace
Exception reportingIn-house exception handling and reporting, with automated diagnostic capture from the installed software
DocumentationDITA Open Toolkit
InstallerMSI-based installer with code signing

Requirements and release planning

Requirements come from customer feedback, support enquiries, partner requests, regulatory and standards updates (notably new or revised CLSI protocols), and internal development priorities. Every requirement is captured as a user story in the issue tracking system and prioritised against the development backlog.

When planning a release, the highest-priority backlog items are selected into a release backlog. A specification outlines the features to be implemented, and work is broken into short sprints with milestones that are reviewed regularly to keep code quality and test coverage high.

Coding

All code changes are recorded in source control with commit notes, providing a complete audit trail. Automated and manual testing begins as features are implemented, with test cases recorded in the test management system. Bugs are tracked and, where possible, fixed immediately to keep the codebase stable. Developers review each other’s changes to identify edge cases and confirm that design and implementation choices are sound, and performance and memory profiling tools are used to confirm the implementation is efficient.

Once the planned features for a release are implemented, the requirements are frozen. Further sprints concentrate on testing and defect resolution, with every additional change reviewed to confirm it addresses the intended issue only.

Testing

Unit, integration, and system tests are developed throughout the release. Wherever practical, tests are automated so they can be re-run to regression-test subsequent changes and releases. Our test suite now contains more than 1,000 proprietary test cases, most of them multi-faceted — exercising multiple inputs and checking multiple outputs per case.

Software testing has limitations that must be recognized and considered when planning the testing of a particular software product. Except for the simplest of programs, software cannot be exhaustively tested … Software testing that finds no errors should not be interpreted to mean that errors do not exist in the software product.

Exhaustive testing is not possible for software of any real complexity, so we take a risk-based approach: testing resources are concentrated on the areas where correctness matters most, particularly the numerical accuracy of statistical algorithms. Statistical functions are tested against other statistical packages, against hand-calculated and spreadsheet-computed reference results, and against public validation suites including the NIST Statistical Reference Datasets. Code coverage is measured at the branch/decision level so we can confirm all paths through the algorithms are exercised. Code reviews are used specifically to surface edge cases and special conditions that need to be tested.

We take this caveat seriously. Our testing strategy is designed to catch the errors that matter most, but we also rely on feedback from the diverse environments where Analyse-it is deployed to identify issues our internal testing cannot reproduce.

User site testing

Towards the end of development, interim releases are shared with selected domain experts for feedback on installation, usability, feature implementation, and numerical correctness in real-world use. Feedback and bug reports are captured in the issue tracking system, prioritised, and either addressed in the current release or deferred to a future one. Automated exception reporting inside the product captures diagnostic information from runtime errors and relays it directly to the development team.

Release

When development and testing are complete, a production release candidate is distributed to selected users. The highest-risk regression tests are re-run on the candidate across supported versions of Excel and Windows. Once the release candidate is stable, the software is released to the public.

Maintenance releases

After a major release, issues reported through customer support and the in-product bug reporting feature are captured in the issue tracking system, prioritised, and scheduled for maintenance releases as needed. Maintenance changes are recorded in source control, regression tests are updated and re-run, and documentation is updated in step with the code.

Numerical accuracy

Statistical software is only useful if its results are correct. We verify the numerical accuracy of Analyse-it against the NIST Statistical Reference Datasets (StRD), the benchmark developed by NIST’s Statistical Engineering Division specifically for evaluating statistical software. NIST StRD datasets carry certified values accurate to 15 significant digits, and each tested result is scored by log relative error (LRE) — the number of significant digits of agreement with the certified value.

Tested against the NIST StRD, Analyse-it performs consistently among the best and outperforms several popular statistical packages. Full tables of results and comparisons with published benchmarks for other software are on our numerical accuracy page. You can download the NIST StRD validation workbooks and re-run the analyses on your own installation as part of your operational qualification.

Separately from our own testing, Analyse-it appears in the published methods of peer-reviewed papers we have verified one by one, each listed with the sentence naming it quoted from the paper itself. That speaks to scrutiny and long use in regulated laboratory work; it is the NIST results above that address numerical accuracy.

Version and change management

Every release of Analyse-it has a unique version number, and all user-visible changes are recorded in our release history. You can use the release history to:

  • See exactly what changed between the version you have validated and any later release.
  • Assess whether the changes affect the functions within your validation scope.
  • Determine what, if any, revalidation is required under your change control procedures.

If you need to remain on a previously validated version while evaluating a newer release, previous versions are always available. This supports the common pattern of running a validated production environment alongside a test environment where upgrades are assessed before being adopted.

21 CFR Part 11 — electronic records and signatures

21 CFR Part 11 places requirements on systems that manage electronic records and electronic signatures. It does not impose requirements on individual calculation tools used within such a system.

Analyse-it is a statistical analysis tool. It computes results and writes them as content into standard Excel workbooks on your computer. It does not provide electronic records management, audit trails, or electronic signatures of its own, nor does it need to — those functions belong to the quality management system that governs how the resulting Excel workbooks are stored, controlled, reviewed, and signed.

For Part 11 purposes, the controls that matter — access control, change control, audit trail, retention, signature binding — live in your document management system, your SOPs, and your IT infrastructure. Analyse-it’s outputs are standard .xlsx files that integrate into whatever document control system you already operate.

Frequently asked questions

Is Analyse-it FDA-validated?

No off-the-shelf statistical software is “FDA-validated” as a product. The FDA requires the regulated user to validate software for its intended use. We provide the evidence of our development rigour and testing — this page, the NIST StRD benchmarks, and the release history — to support your validation activity.

What GAMP category is Analyse-it?

Category 3 — non-configured, commercial off-the-shelf product. This is the lowest-risk category and requires the least validation effort.

How does the FDA’s Computer Software Assurance (CSA) guidance affect how I validate Analyse-it?

CSA, finalised in September 2025, explicitly encourages manufacturers to leverage vendor documentation and to apply risk-based, proportionate assurance activities rather than exhaustive scripted testing. This page, our NIST StRD benchmarks, and the release history are exactly the vendor evidence CSA expects you to leverage. See the Computer Software Assurance section above.

Do you provide a validation kit?

We provide the resources most quality managers use to build their own validation file: this page, the NIST StRD validation workbooks, and the release history. Because validation scope is defined by your specific intended use, your quality team typically produces the IQ/OQ protocol itself, referencing our documentation as the vendor-supplied inputs. If there is something specific you need from us to support your validation, please get in touch.

Does Analyse-it comply with 21 CFR Part 11?

Analyse-it is a calculation tool, not a records management system. Part 11 applies to the system that manages your electronic records and signatures — see the 21 CFR Part 11 section above.

Does my data leave my computer?

No. All statistical computation happens locally inside Microsoft Excel on your machine. Your data is not transmitted, uploaded, or stored by Analyse-it. The only network communication the software performs is licence activation, update checks, and, if enabled, automated exception reporting — none of which involve your analysis data.

What should I do when I upgrade to a new version?

Review the release notes for the new version, assess whether any changes affect functions within your validation scope, re-run the operational checks that cover those functions, and document the outcome under your change control procedure.

Can I stay on a previously validated version?

Yes. Previous versions are always available. This is a common pattern: a validated version runs in production while a newer release is assessed in a test environment before being promoted.