Editorial
How we evaluate
The value of a review site rests entirely on what it refuses to do. This page sets out our method so you can check whether the rest of the site follows it — and hold us to it when it does not.
Last reviewed August 2026. We update a page when the underlying evidence changes, not on a schedule.
Our process, in order
- 1
Start from systematic reviews, not from products
Where a systematic review exists, it takes precedence over any individual trial, and both take precedence over a manufacturer’s white paper. For this category the anchor is the 2023 Cochrane review of blue-light filtering lenses, because its method was published before its results were known.
- 2
Read individual trials for their design, not their headline
Sample size, blinding, whether there was a placebo lens, who funded it, and how long follow-up ran. A twenty-person unblinded trial and a pooled analysis of seventeen trials do not carry the same weight, and we do not present them as if they do.
- 3
Label every claim with its confidence
Strong evidence, professional consensus, or unresolved. If we cannot place a claim in one of those categories, it does not go on the site. Readers are entitled to know how much to lean on any given sentence.
- 4
Mark manufacturer claims as claims
When a company states a filtering percentage or a material specification, we attribute it to them. We do not have the instruments to verify optical claims, and we say so on the page rather than in a footnote.
- 5
Publish no number we cannot stand behind
No star ratings, no review counts, no lens weights, no filtering percentages presented as our findings, and no hard prices. Prices in this category move weekly; a figure printed on a review page is a small lie waiting to happen, so we give bands and link out.
- 6
Make a recommendation, and name who should skip it
Every pick says why it won, who should buy it, who should not, and what the alternatives are. Where the honest answer is “see an optometrist instead of buying anything”, we give that one too — it appears repeatedly on this site, and it costs us money every time.
How we evaluate products without a lab
We have not put these lenses in front of a spectrophotometer, worn them for a month, or run controlled comparisons — and we will not use the word "tested" for anything we have not done. What we do instead is synthesise four kinds of information, and the wording on every review tells you which one a given claim came from.
- Published research. Systematic reviews and randomised trials, stated directly and cited. This is the only tier we assert as fact.
- Manufacturer specifications and retailer terms. Filtering figures, materials, warranty and return policies. Always attributed — "GUNNAR publishes…", "the listed warranty is…" — and never restated as something we measured.
- Aggregate buyer reports. Recurring patterns across public retailer reviews, phrased as patterns: "buyers commonly report…". We do not publish counts we did not count, and we do not present a single anecdote as a trend.
- Our own reasoning. Inference from the three tiers above, flagged as inference — "which suggests…", "in practice this likely means…" — so you can disagree with it separately from the facts it rests on.
Anything that fits none of those tiers does not go on the page. Every review also carries a list of what we could not verify, because the limits of a review are part of the review.
Our confidence labels
- Strong evidence
- Supported by a systematic review of randomised trials, or by the settled position of a major professional body. You can act on this.
- Professional consensus
- What clinicians and their professional bodies advise, where the underlying trial evidence is limited but the reasoning is sound and widely accepted.
- Unresolved
- A plausible mechanism with trial evidence that is small, mixed or poorly controlled. Worth knowing about; not worth spending much on.
How we handle the conflict of interest
Affiliate links create a real incentive to recommend purchases. We manage it in three ways, all of which you can verify by reading the site.
- The inconvenient evidence stays on the selling pages. Our homepage and our buying guide both carry the finding that these lenses probably do not reduce eye strain, alongside the affiliate links. It is the first thing a commission-driven site would cut.
- Disclosure sits next to the links. Not only in the footer. Any block of retailer links carries a notice immediately above it, and every commercial link is marked
sponsored. - We point readers at the cheap half of the category. Our standing advice is to spend on the frame rather than the filter, and to skip the purchase entirely where an eye exam is the better use of the money. Commission is a percentage of the sale, so that advice works directly against our own revenue.
How we handle updates
Pages carry a single "last reviewed" date, and it reflects when someone actually reread the page against its sources — not an automated timestamp bumped to look fresh. We do not add a review date to a page we have not reviewed, and we do not claim a publishing cadence we cannot keep. If the underlying evidence changes, the page changes; if it does not, the page stays as it is and the date stays honest.
What would change our conclusions
Stating this in advance is a discipline worth having. We would revise the eye-strain conclusion given an adequately powered, properly blinded randomised trial — or an updated systematic review — demonstrating a clinically meaningful benefit. We would firm up the sleep section given larger placebo-controlled trials using well-characterised lenses and objective sleep measures. We would revise our position on any specific product if the manufacturer published a full transmission curve from an independent laboratory.
If you know of evidence that meets one of those descriptions, send it to corrections@blueblockreview.com.