A useful review explains what the evidence can support, not just whether a product sounds appealing. Anywhere Work Lab uses the following categories for relevant reviews and comparisons. They describe the basis of an article, not a certification or a guarantee of quality.
Personally Tested
Use this only when the product or service was directly used for the task described. Identify the product, the task or use case, what was actually evaluated, and important limitations. Where known and relevant, include the plan or tier, approximate use period or scope, and whether access was self-funded, free, or provided.
A short trial does not establish long-term reliability or comprehensive knowledge of every feature. An account of an ongoing workflow is not automatically a controlled benchmark. A provider’s current documentation can supplement a first-hand account, but it does not turn a documentation check into a fresh hands-on test.
Research-Based
Use this when the product or service was not directly used for the article. Research should rely primarily on official product documentation, pricing and help pages, original research, or other authoritative primary sources. Credible independent evidence may add useful context when it supports a specific claim.
Identify what could not be verified. A provider’s claim should be attributed to that provider, particularly where performance or results have not been independently assessed. A research-only assessment must not say “in my testing” or imply direct use.
Mixed
Use this when products in one comparison have different evidence bases. Identify the basis separately for each product—for example, Apollo: Personally Tested; Outreach: Research-Based. Do not put one broad “Personally Tested” label on a comparison when only one product was used.
Essays and practical guides
Non-product articles may instead use an editorial description such as Operator’s Notes or Practical Guide. These describe the format, not proof that every suggested tool has been tested. Not every article belongs in a product-review framework.
The evidence box
Relevant reviews use a small, neutral box near the beginning. Only fields that can be truthfully completed are shown:
- Basis: The evidence category, separated by product where needed.
- Scope: The particular workflow, use period, or part of the product covered.
- What was checked: The actions or features evaluated, and the specific provider information checked.
- Limits: What the evidence does not establish, including missing detail that affects the conclusion.
- Access: How access was obtained, when known and relevant.
- Last checked: The date and subject of a factual documentation check; not an implied date of product use.
Missing facts are not filled with assumptions. A label should be withheld when its basis cannot be established. There are no certification seals, “verified” badges, star ratings, invented testing scores, or fabricated benchmarks behind these categories.
What a source establishes
Link sources next to the claims they support, or explain their purpose in a short source note. Official documentation can confirm documented capabilities and restrictions; it does not prove a personal result. AI output is not a source. Personal observations remain personal observations even when documentation is added.
Plans, prices, features, and availability change. Check the provider’s current terms before buying. Keep the original publication date and use substantive update dates only for actual editorial or factual updates. A documentation check must not be described as a new test of every product.
Relationships and corrections
Disclose relevant affiliate relationships, provided access, free products, discounts, or sponsorships without inventing relationships that do not exist. A product does not need an affiliate program to be recommended. See the Affiliate Disclosure.
Evidence boxes are being introduced to a small initial set of articles. Older unlabelled posts should not be assumed to have undergone this process. For errors or questions about evidence, contact jsooo090677@gmail.com. The Editorial Policy explains the corrections process.
Methodology published: 6 October 2026.