This page collects every standard that governs what we publish on DatingSiteSpot. It is meant to be specific enough to fail an audit against — not a slogan page.
Ranking independence
Our category rankings are not for sale. We do not adjust rank order based on commission rate, partnership status, or vendor pressure. The ranked order for every category lives in `src/content/categories/*.yaml` and is reviewed by the senior editor when the category is updated.
When two products in the same category pay materially different commissions, both rates are noted on the product review page so a reader can judge the influence themselves. See the disclosure page for the affiliate framework.
Affiliate disclosure
DatingSiteSpot earns commissions when readers click through to a partner site and complete a qualifying action — typically a signup or a paid subscription. Those commissions never determine our scores, our category rankings, or the order of products on a page.
The full disclosure is on /disclosure/.
Evidence standards
Every editorial review carries evidence. Evidence is either present on the page or absent and labelled as queued — we never imply a capture that doesn't exist.
What counts as evidence:
- A dated reviewer log on each review page (`Reviewed by` card)
- A 5-axis score with sub-scores for UX, Value, Audience, and Safety
- Pros and cons sourced from the editor's own testing
- A redacted hero screenshot from the editor's account (or an
explicit "Capture queued" placeholder)
- A pricing review tied to the reviewed_at date
- A paid-flow check when the product is subscription-only
The full evidence policy lives in `docs/evidence-review-policy.md` and is enforced by these public verifiers: `verify-screenshot-metadata`, `verify-evidence-pipeline`, `verify-evidence-freshness`, `verify-real-captures`, `verify-editorial-trust`.
Correction policy
We correct factual errors as soon as we are aware of them. Send corrections, factual disputes, or rights questions to `[email protected]`.
When we revise a review:
- The change is logged in the review's frontmatter (`reviewed_at`
bumps to the day the revision was completed).
- If the change is material — a price increase, a feature loss, a
new safety incident — the page eyebrow shows the new date and the reviewer card surfaces it.
- We do not silently rewrite history. If we change our mind about a
product's score, the new score is what ships; we do not fabricate earlier scores.
Review update process
A review is re-tested at least once every twelve months and earlier when one of these happens:
- The product ships a material change to pricing, matching, or
moderation.
- A safety incident is reported or surfaces publicly.
- The product redesigns the surface that the hero screenshot
illustrates (the hero is then re-shot or archived).
- The reviewer's own use of the product surfaces a discrepancy with
the published review.
When a re-test bumps the score, the new score ships immediately. When a re-test confirms the existing score, `reviewed_at` updates but the score is unchanged.
Screenshot policy
We do not use stock screenshots, press-kit images, AI-generated mockups, or recolored versions of any of the above. Every product screenshot we ship comes from one of our own editorial accounts on the live product.
Each screenshot carries metadata declaring:
- the capture date (`captured_at`)
- the optional re-test date (`reviewed_at`)
- the capture method (`manual-desktop` or `manual-mobile`)
- the viewport
- explicit `redacted: true` and `contains_personal_data: false` flags
A page where the screenshot isn't ready renders the editorial placeholder card with the "Live capture queued" eyebrow. We do not imply a capture that doesn't exist.
The day-to-day playbook for capture is in `docs/screenshot-capture-workflow.md`.
Evidence refresh policy
Evidence ages out. The refresh policy keeps the reader honest about when a capture was taken and when it was last re-validated:
- Every captured asset carries a `captured_at` date in YAML.
- A capture that has materially changed on the live product is
re-shot. The `reviewed_at` field bumps to the re-shoot day.
- An asset older than the freshness window (six months for priority
brands) shows a "Captured (re-test queued)" pill on the page so a reader can judge its age.
- When a product redesigns past recognition we move the old asset to
`archived` and the page falls back to the editorial placeholder.
An automated freshness check rejects future-dated captures, out-of-order review-vs-capture dates, and "captured" wording on placeholder rows before they can publish.
AI and content policy
We use writing tools to draft and edit. We do not let a model generate evidence:
- We never publish an AI-generated product screenshot.
- We never publish an AI-fabricated quote, anecdote, or testimonial.
- We never publish a price, feature list, or moderation claim that
the editor has not verified against the live product.
- We never publish an AI-generated "review-of-reviews" composite —
every score on this site is from an editor's own test session.
- AI-assisted draft text is reviewed by the editor before publish.
Editorial synthesis blocks on review pages (reality-check lines, "who should avoid this" rows, churn-risk lines) are produced by deterministic helpers from the review's own rubric data. They are not free-form AI output — every line maps to a verifiable score or site trait.
Replacement policy for closed brands
When a brand we previously reviewed closes — operator shutdown, US withdrawal, regulatory exit — the review page stays live with a deprecation banner. We do not delete the page (the URL has accrued external links and the editorial context still matters for readers arriving via search).
What changes on a closed page:
- A deprecation banner is shown at the top of the review.
- The "/go/" outbound page emits a noindex notice and routes the
reader to a sibling category rather than an affiliate destination.
- The `closed: true` flag in the site YAML excludes the brand from
ranked lists and the homepage decision matrix.
- When a `replaced_by_category` slug is set, the deprecation
notice routes readers to that category instead of the generic `/best/` index.
We do not redirect closed-brand URLs to the homepage or to other brands. That dilutes ranking signals and misleads the reader.
Contact path
If you spot a factual error, a stale price, a privacy issue with a screenshot, or a ranking call that doesn't match the evidence on the page, email `[email protected]`. We read every message and we publish dated updates when we revise a review.
For rights questions (a product operator who needs a takedown, a correction request from someone identifiable in a screenshot, a regulator with a formal request) the same address routes to the senior editor.
What this page is not
This page is not marketing. It is the standard we hold ourselves to, and it is the standard a reader can hold us to. When we drift, those automated checks fail and the change never goes live.