Methodology
How Sysprobs Data collects, checks, publishes and refreshes every figure on the hub, when a figure is labelled dated, who edits it and how to report a correction.
What this section is
Sysprobs Data publishes original data on IT infrastructure: statistics pages that collect every public survey on a topic and show each figure with its sponsor, sample and question wording; calculators built from vendor list prices; and reference tables that are kept current. The audience is anyone who needs a figure with a source: writers, analysts, vendors and engineers planning a budget.
How figures are produced
- Collection. Every figure is retrieved from the organisation that produced it: a vendor pricing page or API, a statistics office, a published report, a public filing or a code repository. News articles that quote a figure are followed to the origin and are not used as the source of a figure. The one exception is set out under “Reports used where no primary document exists” below.
- Ledger. Each figure is recorded with its URL, publisher, publication date, retrieval date, the verbatim text that carries it and its context. Where the figure comes from a survey, the sponsor, the research firm, the sample size, the fieldwork dates and the question wording are recorded too. A figure that cannot be traced to a primary source is not published.
- Cleaning. Raw downloads are kept untouched. A script produces the clean dataset, and every transformation is written down in full in the page’s own methodology section, never “on request”. Derived figures state their formula where they appear, and rounding is stated once per page.
- Charts. Every chart is generated from the clean dataset. Units, periods, the vendor and the report year appear on the chart itself, and a chart drawn from vendor-sponsored surveys says so in its subtitle.
- Review. An adversarial check compares every number on the page against the dataset and the ledger, confirms that every source link resolves, and fails the page on any mismatch. A second, independent review reproduces the data and the calculators before a page goes live. The editor approves every publication; nothing is published automatically.
- Refresh. Each source has a known release cadence, shown in the meta line of every page. When a source updates, the page is rebuilt, re-checked and republished at the same URL with a changelog entry. A refresh that moves a headline figure by more than 15 percent is held for manual review before it goes live.
When a figure is labelled dated
A figure is labelled dated when the date that counts for it is more than 24 months before the page’s last data check. Each page prints its cut-off date once, in its methodology section. The CSV marks every dated row with the same cut-off, in a dated column or, in datasets without one, in the row’s note. The date that counts depends on the kind of figure:
| Kind of figure | Date that counts | Example |
|---|---|---|
| Survey result | End of fieldwork; the publication date when fieldwork is not stated | A survey fielded in July 2024 is dated from August 2026 |
| Price on a vendor’s live price list, price API or price file | The date it was read. A price the vendor still lists is current, whenever it took effect | An Azure meter effective since 2022 and still in the price API is not dated |
| Price or rule on another vendor page that shows its own last-changed date (documentation, help pages, blog posts) | That last-changed date | A traffic help page last changed in May 2024 is dated |
| Price or statement read from an archived copy | The date of the archived copy | |
| Financial filing | The end of the period it reports | Fiscal 2024 figures for a year ended 3 November 2024 are dated at the first data check after 3 November 2026 |
| Forecast or analyst statement | Its publication date | |
| Lifecycle date (end of support) or the date of an event | Never dated: it stays true until the vendor changes it, and the refresher re-checks it yearly |
Reports used where no primary document exists
A calculator sometimes needs a rule that the vendor has not published, for example an order minimum that was announced to partners and then reported withdrawn. In that case the page may use news or reseller reports, on three conditions: each is labelled “report, not a vendor document” in the text, in the source panel and in the CSV; the setting it supports is shown as an option the reader controls, not as a default figure; and it is never a headline figure. The refresher looks for a primary document at every refresh and replaces the reports when one appears.
Reading survey figures
Every survey on the hub is shown with who paid for it. A vendor that sells an alternative has an interest in one result; a vendor that resells the incumbent has an interest in the opposite. Figures from different surveys are not combined into a single number, because question wording, time bounds and sample populations differ. The source panel at the foot of each page lists, for every key, the publisher, sponsor, research firm, sample, fieldwork dates and, figure by figure, the question wording and the verbatim text.
Reuse
Data, charts and tables are published under CC BY 4.0. Use any figure with a link to the page it came from, and name the organisation that produced the figure. Each page has a citation block with ready-made text and HTML, a CSV download with the publisher and URL of every row’s source, and embed code for every chart. The calculators run entirely in the browser and send nothing to a server.
Corrections
If a figure is wrong, a source has changed or a link no longer resolves, use the contact page with “Data correction” in the subject, and include the page URL and the source. Corrections are made within five working days and listed in the page changelog.
Who runs it
Sysprobs Data is edited by Dinesh Kumar, who runs Sysprobs.com and has twenty years in Windows Server, virtualisation, Microsoft 365, enterprise networking and security operations. Collection, analysis and the first checks run through a documented, automated pipeline; every page passes an independent review and is approved by the editor before it is published. Pages in this section carry no sponsored content and no paid placements.
Changelog
- 2026-10-22: Named the editor. Added the rule for when a figure is labelled dated, the rule for reports used where no primary document exists, the independent review step, the per-figure source panel and the source columns in every CSV.
- 2026-10-08: First published.
