How we evaluate optics

What this site is, and what it is not

Most of our guides are research-based roundups. We read the retailer's product page, the manufacturer's own specification sheet and warranty documents, independent tests from the shooting and outdoor press, and the full owner-review record, then we synthesise. Where we have handled a product ourselves we say so and describe what we did. We never claim a hands-on test we did not perform, and we do not publish photographs of gear we do not have.

That is a real limitation and it belongs here rather than buried at the bottom. A research-based guide cannot tell you how a scope feels on your rifle at last light. What it can do, and what most listicles do not, is make sure every number you are given is real, current and traceable.

Where every number comes from

A figure is only allowed on a page if it came from a source we opened ourselves. In practice that means a fixed order of checks, each of which exists because skipping it burned us once:

• Price and stock come from the product page, never from a category grid or a filter. A filter can show a product as in stock while the exact model you would receive is backordered. We found precisely that on an Aimpoint listing whose real ship estimate was the following year.
• Every variant is read, not just the first one. Retailer structured data often lists a factory demo or open-box unit beside the new one. Reading only the first entry once gave us the price of a factory demo unit, some fifty dollars below the new one, so now every variant is listed and demo units are excluded by name.
• Being listed is not the same as being buyable. A Bushnell Banner scope appeared in the retailer's live category with an ordinary product tile; its own product page said Product No Longer Available. A product you cannot buy does not get ranked.
• Awards are verified at the awarding body, not at whoever repeated the claim. If we cannot find the award itself, it does not get printed. A retailer's own "Best Rated" badge is a shop label, not an award, and we say so when it comes up.
• Owner ratings stay attributed. When we write 4.8 out of 5 from eleven reviews, that is the retailer's number and the page says whose it is. It is never presented as ours.

We do not take the manufacturer's word either

This is the part that surprises people. A maker's own product page is a primary source, but it is not automatically correct. While writing a single group of pages we caught one brand contradicting itself three separate times:

• A product page called its optics fully multi-coated in the marketing copy and listed Multi-Coated in the specification table on the same screen. Those are different specifications, and they are not a rounding difference. We take the table, and we tell the reader why.
• Another advertised 30 MRAD of elevation travel in its description while its own table gave 29 MRAD as the dial travel limit. One mil is 36 inches at 1,000 yards, so on a scope bought for dialling that gap is the whole point of the purchase.
• A third published an identical specification block on two different models, the 8x42 and the 10x42 of the same binocular. Arithmetic settled it: exit pupil is the objective divided by the magnification, 42 divided by 10 is 4.2 mm, and the 5.3 mm printed on both pages is the 8x42's figure. Anyone researching the 10x42 on the manufacturer's own site was reading the wrong numbers.

Retailers make the mirror-image mistake. One listed Battery Life: 6 hours for a riflescope the manufacturer rates at 5,000 hours, having printed the illumination auto-shutoff timer in the battery-life row. Taken at face value, a feature looked like a disqualifying flaw. Where sources disagree and arithmetic or documentation can settle it, we settle it and show the working. Where nothing can settle it, we print both figures and say which we used.

Deep research is a lead, not a source

We use AI-assisted deep research to find claims worth checking: awards, competitor comparisons, independent tests, recurring owner complaints. Nothing reaches a page in the form the report delivered it. Every load-bearing claim is re-verified by opening the cited source itself, and reports routinely fail that step. In one recent round four separate claims were dropped: a warranty term that does not appear in the manufacturer's current warranty document, an award that on inspection had gone to a different product, a country-of-manufacture assertion the maker publishes nowhere, and a stock claim we could not confirm. A summary that is ninety percent right is still a summary that will publish something false.

What we refuse to publish

• Our own star ratings as review schema. Our editorial score is our opinion, and dressing it up as aggregate rating markup would tell search engines that customers rated the product when no customer did. SEO tools recommend this because competitors do it. We decline.
• Reviews with no downsides. Every product involves trade-offs; a page without them is advertising. We also separate a genuine trade-off (weight, glass at that price tier, limited adjustment travel) from one faulty unit, which tells you nothing about the product itself.
• Hidden availability. If something is on backorder we say so with the estimated date, even when that costs us the sale. Ranking a product you cannot receive, without saying so, wastes your time.
• Invented dates and invented precision. If a figure cannot be sourced it does not appear, and where a page has no verified content date we publish none rather than a plausible one.

How we handle prices

Prices move constantly and we are not a price tracker. Rather than scatter separate timestamps across the site, which is how a site ends up quietly stale in a hundred places at once, every price is checked in one sweep and the site renders that single verified date wherever a price appears. The date you see is a real date on which someone opened that product page, not the day the site happened to be rebuilt. In running text we prefer an honest band to a figure with cents on it, because the exact number belongs beside the buy button where it is actually checked. Confirm the current price on the retailer's page before buying.

What our build refuses to publish

Most of the rules above would decay into good intentions if they lived only in a style guide, so the ones a machine can check are enforced when the site is built. The build fails, and the page does not ship, if any of the following is true:

• A retailer link appears that is not in our verified URL register, so no plausible-looking address can be typed from memory.
• The same product carries two different prices anywhere on the site.
• A page hand-writes a price-verification date instead of rendering the swept one.
• A page ranks several products but publishes no list markup, or a list entry carries no link.
• An internal link points at a page that does not exist, or a video ships without complete schema.
• Our cookie and privacy pages describe behaviour that does not match what the site actually renders.
• A published file still contains a note-to-self such as TODO or "to be updated".

Every one of those exists because we got it wrong once. And every one was deliberately broken after it was written, to watch it fail, because a check nobody has seen fire is not a check.

When we get it wrong

We will. When a reader or a later review finds an error we correct the page, and where the fault was in our process rather than in one figure we change the process so it cannot recur. Corrections are made in place with the page's modification date updated, not silently. If you find something wrong, tell us and it will be checked against the source.

Independence

We earn affiliate commissions on some links (full disclosure), and that is how the site is funded. Commissions do not decide rankings: the order of picks is settled before payout is considered, products we would not buy are included where they help you compare, and a product that fails on the facts is not rescued by being lucrative. More about who publishes this site.