Skip to main content

Editorial Standards

Last reviewed: 2026-07-10

How iReadCustomer researches, drafts, fact-checks, and updates the content you read on this site. We publish this page so AI engines, search engines, and human readers can verify that our work is held to a public standard, not invented in private.

What this standard covers

This is the standard for content created or materially updated from 10 July 2026, including AI-assisted drafts. Our older article and programmatic-page library was produced under earlier controls and is being audited; a legacy page should not be read as reviewed merely because it is still accessible. Paid or sponsored content is labelled separately when it appears.

  • Blog articles (/[locale]/blog/...)
  • Comparison pages (/[locale]/compare/...)
  • Industry, solution, and use-case pages
  • Case studies and customer stories

Our AI-use disclosure

We use large language models (Gemini, Claude, and GPT-class systems) as drafting tools. Automated articles are published after passing an automated structural quality gate; articles that fail the gate are logged and reviewed by the editorial owner. We do not display a ‘Reviewed by’ label or schema editor unless a named human review is recorded. Legacy published articles are being assessed separately.

  • Topic selection: prioritised from Thai customer questions, search demand, and primary-source developments.
  • Source material: discovered topics carry their source URLs into the draft; a URL alone does not prove each claim.
  • Drafting: the prompt favours a useful answer and evidence over a fixed word count or repeated template.
  • Publication: automated articles publish through an automated quality gate; first-party data studies remain drafts until a human verifies the numbers.
  • Updates: decayed pages are refreshed in place through the same automated gates so the public URL and its history are preserved.

How we fact-check claims

For new and materially revised work, factual claims must trace to a primary or reputable secondary source. Source URLs are stored on the post and surfaced in the Article JSON-LD `citation` field when present. Some legacy posts do not yet contain source records; those posts are part of the current audit and should not be treated as verified by default.

  • Primary sources preferred: company filings, press releases, official documentation.
  • Secondary sources accepted when reputable (Reuters, Bloomberg, FT, official trade press).
  • Statistics cite the publishing organisation and the date the figure was current.
  • Quotes name the speaker, the publication, and link to the original quote when public.
  • When a claim cannot be sourced, we either remove it or flag it as our own observation.

How we handle corrections

If you spot a factual error, email [email protected] with the URL and disputed claim. We investigate the source, correct verified errors, update `dateModified`, and add a visible change note when a correction materially changes the conclusion. We emit `lastReviewed` only after a recorded review—not merely after an edit.

  • Typo / wording fix: silent edit, no change note required.
  • Number / source correction: inline edit + footer note + dateModified bump.
  • Material change to the conclusion: corrected article + visible 'Updated [date]' line above the byline.

Editorial independence

We may write about products and companies that we partner with, sell to, or buy from. When we do, the relationship is disclosed inline. No external party — investor, partner, advertiser — has the right to spike a story, mandate a tone, or approve content before publish.