Review Methodology

Last updated: August 17, 2026

This page explains exactly how Camp Gear Data produces its comparisons: what data we collect, how we turn inconsistent manufacturer figures into comparable numbers, what we refuse to publish, and — most importantly — the specific ways manufacturer specification data is known to be wrong.

1. What We Do, and What We Do Not Do

We do not test camping gear. No one at Camp Gear Data has slept in these bags at altitude, run these stoves in wind, or left these lanterns burning to see what happens at hour twelve. Every number on this site is a published manufacturer specification or a figure computed from published specifications.

What we do instead is read an entire category at once. A reviewer who has field-tested nine sleeping bags can tell you how those nine felt on cold nights — genuinely valuable knowledge we cannot offer. What that reviewer usually cannot tell you is where a tenth bag sits against every comparable bag on sale. That distributional view is the only thing this site is built to provide.

Both approaches have failure modes. A field review generalises from a small sample. A specification comparison inherits every inaccuracy in the manufacturer’s own marketing. We think the honest response is to be explicit about which one you are reading and where ours breaks down, which is what the rest of this page does.

2. How Specifications Are Collected

For each category we gather the products actually selling in that category — not a hand-picked shortlist — and collect the published specification data for each. That data arrives in two very different forms:

• Structured specification fields. The labeled attribute table on a product listing: “Temperature Rating: 20 Degrees Fahrenheit”, “Maximum Weight Recommendation: 300 Pounds”. These are the fields a manufacturer has filled in deliberately.
• Marketing copy. Figures that appear only in the product title, feature bullets, or description: “1000 Lumens”, “20,000 BTU”, “65L Internal Frame”.

We record which source every single value came from, and we say so on the pages. These are not equally trustworthy. A structured field is a deliberate declaration; a number in a feature bullet is a sales claim written by a copywriter. Blending them into one table without distinction would quietly destroy the reliability of the whole comparison.

3. How Figures Are Made Comparable

Manufacturers publish the same quantity under different labels and in different units. Weight appears as “Item Weight”, “Weight”, or “Product Weight”, stated in pounds, ounces, kilograms or grams. Cooler capacity appears in quarts, liters, fluid ounces and cubic inches — sometimes several within one category.

We map each label variant onto a single canonical field and convert every value to one unit. Where a label is ambiguous, it is scoped to the categories where it is unambiguous. “Capacity” is the clearest example: it means a person count on a tent, a volume on a cooler, and a weight limit on a sleeping pad. Treated as one field, it produces nonsense — an early version of our pipeline computed a sleeping-pad category median of 300 persons before that scoping existed.

4. The Coverage Rule

A specification is only published when enough products in its category carry it to make a comparison meaningful. Our threshold is that the field must appear on at least 40% of the category, with a minimum sample size below which no baseline is computed at all.

This rule is why some categories on this site have no ranking. Sleeping pads are the clearest case. R-value is the number that defines a pad’s usefulness, and it appears on roughly a quarter of listings — and that is the ceiling, not a limitation of our collection, because only about a quarter of pad listings mention R-value anywhere at all. We could rank pads on weight and thickness instead. We do not, because that would be ranking them on the wrong thing while implying we had ranked them on the right one.

5. Baselines Use the Median

Every category baseline is built on the median and the quartiles, never the mean. This is not a stylistic preference. A single mis-converted unit in a category of sixty products moves the mean by several times and leaves the median untouched — which is exactly what happened when cooler capacity stated in cubic inches was briefly read as liters, pushing the category mean to five times its true value while the median stayed correct.

When you see a “category median” on this site, it is the midpoint of the products we could measure, and the quartile spread tells you how tightly the category clusters around it.

6. Known Failure Modes in Manufacturer Data

These are documented problems we have measured, not hypotheticals. Each one affects how much confidence a figure deserves.

Multi-size listings publish one variant’s specification

This is the most serious problem we have found. Retailers list several sizes of the same product under a single entry — a tent titled “10 Person / 11 Person / 12 Person” — while the structured specification carries only one variant’s figure. The result is a value that is real, structured, and describes a different product than the title.

In our tent data, more than half of listings state a size range, and nearly half of published capacity figures disagree with their own product title. Coolers show the same pattern less severely. Sleeping bags and backpacks are barely affected.

No coverage check can catch this, because every individual value looks perfectly plausible. Where a category is materially affected, we say so on the category page, and we exclude multi-size listings from direct product comparisons.

Specification completeness falls away from the bestsellers

Established brands fill in their specification fields; smaller sellers frequently do not. As we look further down a category past the best-known products, the proportion of listings carrying a usable figure drops — in some categories by twenty percentage points. A comparison built only from the most popular products will therefore look better-documented than the category really is. We compute baselines from a deliberately wider sample so this decay is visible rather than hidden.

Units collide with model names

At least one major lantern manufacturer abbreviates lumens as “L” in its product names, which collides directly with liters. Figures like these have to be read in the context of the category, not pattern-matched blindly, and we sanity-check every category’s numbers against known real-world ranges before publishing.

7. What We Will Not Publish

• Star ratings and review counts. Aggregate sentiment is not a specification, and republishing it would add nothing you cannot see on the retailer’s own page.
• Exact prices. They change constantly and any figure we printed would be wrong within days. Where cost affects a decision we describe it in tiers, and cost-based ratios are always relative.
• Fabricated test results. We have not tested anything, so there are no results to report.
• Specifications we could not verify were comparable. If a figure is measured on inconsistent bases across brands, we either flag it clearly or leave it out.

8. Who Applies These Rules

The decisions described above — which ranking rule a category gets, which specifications are complete enough to publish, whether a computed figure is plausible — are made by the site’s editor, writing under the byline Ellis Hartley. That is a pen name, disclosed in full on the about page.

It carries no claim of field expertise, and deliberately so. Nothing on this site rests on the editor having used the equipment, because the editor has not. What the role involves is defining the rules on this page, checking the output against known real-world ranges, and documenting the failures — every problem in section 6 was found that way rather than reported to us.

9. Corrections

If a figure on this site is wrong, we want to know. Specification data changes when manufacturers update listings, and our parsing has been wrong before — every failure mode in section 6 was found by checking our own output against reality, not by being told. Report an error through the contact page and include the page and the product.