I haven't tried half of what I recommend

A reader wrote to me. The message was three lines long, and only the last one mattered.

You recommend a bag you’ve never carried. Why would I read you?

It’s the best question anyone has asked me on this site, and I don’t have a clever answer. I’m going to take the time to answer for real, because if the answer doesn’t hold up, nothing I write afterward holds up either.

Why I don’t run hands-on trials

Testing a backpack doesn’t mean opening the box and deciding it looks nice. It means buying five or six comparable ones, loading them identically, carrying each for three weeks under equivalent conditions, noting what chafes, what tears, what seals badly in the rain, then throwing away five bags.

I live out of forty litres and change cities every three weeks. I have neither the budget to buy six bags, nor the room to carry them, nor the stock of identical trips that would make the comparison honest. This isn’t a moral stance on consumption. It’s a physical constraint, and it isn’t going to move.

Sites that do this properly exist. They buy their own gear, publish their protocol, employ people, and live off subscriptions as much as commissions. I have no interest in pretending to be them.

What I do instead

I read the manuals. One Wednesday in January, in a café on Rua da Graça where the wifi barely held, I spent an hour and a quarter on a backpack’s manual for a single piece of information: the exact back-panel height, which appears on no sales page and decides whether the bag fits under an airplane seat. Manuals are there to download, nobody opens them, and they contain the usage limits, the operating temperatures, and the warnings the sales sheet never repeats.

Standards, when there are any, are worth more than everything around them. A water-resistance rating, a watt-hour capacity, a surge-resistance class mean something precise. The adjectives on the sales page don’t.

The average rating says nothing. I read the detailed negative feedback, over twelve or eighteen months, looking for flaws that recur across several different people in several countries. A zipper that fails for three people in Europe and two in North America over the same period is a signal. Forty glowing comments posted the same week is another signal, in the opposite direction.

And above all, I eliminate. That’s the most useful part, and the one that takes the most time. For any family of objects, nearly the entire catalogue falls away on criteria you can verify without owning the thing: dimensions that don’t fit in the cabin, no spare parts available, not sold in France, weight incompatible with the stated use. What’s left, three or four options, deserves to be talked about. The other fifty deserve an explanation of why they were cut.

The four words

For every object I write about here, there’s a word that says where I’m speaking from. It’s displayed, not buried in a methodology page nobody opens.

Owned. The object is mine, I use it, and I say for how long. The phone I’ve had in my pocket for four years falls in this category. Even this level doesn’t let me say it’s better than another one, only that it’s held up for four years under my own conditions.

Handled. I’ve held it, in a shop or at someone’s place, without extended use. I can speak to its weight, its finish, how a zipper behaves. This level doesn’t let me claim anything about durability, which happens to be the exact question everyone asks.

Analysed. I haven’t held it. I’ve read the manual, the specs, the published teardowns where they exist. This level doesn’t let me claim anything about comfort, real-world noise, or what the object becomes after six months. The text you’re reading right now falls in this category: it’s about method, it recommends nothing, and it proves nothing about any object.

Cross-checked. I haven’t held it, and I’m relying on public feedback, for which I give the count and the time period. This level doesn’t let me claim any personal experience, and it inherits every bias in that feedback, including the ones I haven’t spotted.

These four words are spelled out in what I don’t test myself and how I choose. Until I’ve bought something from a given family of objects, only the last two apply, and I stick to that.

What I won’t do

I will never write “I tested this” about an object I haven’t tested. Nor “after six months of use”, nor “I carry it everywhere”.

This rule is the only thing separating this site from what it criticises. The sentence is easy to write, it boosts conversion, and it’s false. It’s also common enough that an investigation published in February 2024 by a small testing site was able to document, with archived screenshots, how very high-traffic publications claimed dozens of tested devices on pages where, a few weeks earlier, no mention of testing existed at all (the investigation is here, published February 2024).

I’m not naming anyone, that’s not my subject and I haven’t redone their verification work. What interests me is that the fraud is undetectable just by reading. A page that claims to have tested something and a page that actually did look exactly alike. The reader has no way to tell, unless the author says outright where they stand.

And the money

There are affiliate links on this site, and I earn a commission when someone buys after clicking one. That’s stated before every link, not buried in a footer. The details are in how I make money.

What that changes about what I write: nothing about the objects I recommend, because the commission is the same regardless of which one. What it changes about which subjects I pick: everything. I choose to write about families of objects that exist in a catalogue rather than things you can’t buy. You should know that.

I didn’t answer the reader right away. It took me four days, because my first draft opened with “actually, testing doesn’t guarantee much either”, and that was a dodge. Testing guarantees something. Not everything, but something, and claiming otherwise would have been the first lie of a long series.