How We Pick

The one-sentence version

We compile what each manufacturer publishes about its own product, print the specifications it declines to publish, do the arithmetic the listings avoid, cite every figure to its source, and put a live price next to each pick.

What we do not do

We do not test these products. There is no lab here, no sample units, no photographs of birds we took ourselves, and no “after six months in our backyard” anywhere on this site. If you find a page implying otherwise, it is a mistake and we want to hear about it.

That is worth being direct about, because the competition in this category is not. A great many pages ranking for these searches describe hands-on results across a dozen feeders. Some of those are real. Many are the same specification sheets we are working from, dressed as experience. We would rather be the site that says which one it is.

Units we claim to have tested: 0. That number is on the homepage deliberately.

What we do instead, in detail

1. Compile the published specifications

Every specification on this site was read off the manufacturer’s own material - product listing, support documentation or official site - and carries a link to where it came from. There are currently 133 cited specifications across 28 products.

Every ASIN was read off a live listing during the research pass, with the brand and model tokens checked against the title the retailer returned. None is remembered and none is guessed.

2. Print the gaps

This is the part no competitor does, and it turned out to be the most useful thing on the site. Where a maker does not publish a figure, we render the row as not published rather than leaving it out or filling it in. There are currently 64 such gaps on record.

The pattern those gaps make is itself a finding. Exactly one maker in this entire category publishes a sensor dimension - the specification that most determines whether a backlit bird renders as a photograph or a silhouette. Almost none publish a subscription price on the listing you buy from. A specification sheet with holes in it is information about the company, and hiding the holes would be hiding that.

The data model enforces this: a specification is either a value with a source, or an explicit “not published”. There is no third state, so an uncited figure is a build error rather than a matter of discipline.

3. Do the arithmetic

Three sums recur on this site, and all three show their inputs so you can check them:

• Three-year cost of ownership. Hardware plus thirty-six monthly payments, which frequently reorders the ranking entirely - the guide.
• The species-claim scale check. Set a “10,000+ species” claim against the reference checklists and it becomes legible - the audit.
• Wi-Fi band compatibility. Which cameras can join which networks, and what to do when yours cannot - the guide.

4. Verify the reference numbers against primary sources

The species-claim audit is the site’s flagship original argument, so its baselines come from the primary checklists rather than from a secondary source repeating them:

• IOC World Bird List v15.2, 13 April 2026 — 11,227 species recognized, of which 164 are extinct, leaving roughly 11,063 living species.
• ABA Checklist v8.19, January 2026 — 1,161 species for the entire ABA Area.

5. Price it live, or not at all

Every price on this site is pulled from the retailer and stamped with the date it was checked. If that data is more than 48 hours old, the number disappears and the button reads “Check price on Amazon” instead.

This is a mechanism rather than a promise. The code physically cannot render a stale price, which is why you will never find a figure typed into our body copy - a number in prose is a number nobody can refresh.

How things get ranked

In this order, which is roughly the inverse of what the listings emphasize:

1. Wi-Fi band — a camera that will not connect is not a camera.
2. True cost over three years, including any subscription.
3. Power, judged against the yard the buyer actually has.
4. Published optics — sensor and dynamic range ahead of resolution.
5. Build and capacity.
6. Resolution, and then the species count, which is close to meaningless.

Where a ranking depends on something we cannot establish, we say the comparison cannot be completed rather than breaking the tie with an invented judgment.

Every pick names its wrong buyer

A recommendation with no downside is an advertisement. Every product on this site carries a “skip this” line identifying who should not buy it, and several pages - nest box cameras most of all - open by explaining why you may not want the product at all.

What this method genuinely cannot tell you

The honest limits, stated plainly, because a method is only worth something if you know where it stops:

• Real-world image quality. We can compare published sensor specifications. We cannot tell you which produces the nicer photograph.
• Identification accuracy. Nobody publishes a figure, and we are not going to invent one.
• Battery life in your yard. Consumption tracks activity and signal strength, so a number from anywhere else does not transfer.
• Reliability and longevity. Whether a device survives three winters is unknowable from a specification sheet. Long-term owner reports are better evidence than we are - how to read them.
• App quality day to day. We have not lived with these apps for a season and will not rank them as though we had.

Who writes this

Scooter M., an enthusiast - someone genuinely interested in this category who reads the manuals, compiles the published specifications and does the math. No ornithological credential, no lab coat, and no claim to either. More on the about page.

How this is paid for

Affiliate commission, disclosed on every page carrying a buy link and in full on the disclosure page. It does not influence the rankings, and the structural reason it does not is that the ranking criteria above are fixed and public - if a recommendation contradicts them, you can see that it does.

When this is wrong

Some of it will be. Manufacturers change specifications, update listings and revise pricing, and a page written today describes what was published today. Every page carries the date it was last updated, all 53 of them.

If you find an error - a specification we have wrong, a source that no longer says what we say it says, a figure that does not add up - tell us. Corrections are handled per the editorial policy.