Every bird feeder design, scored on the same attributes

The short answer

• Only two designs score full marks for squirrel resistance without correct pole geometry: wire-cage feeders and weight-activated feeders.
• Platform feeders reach the widest range of species and score worst on squirrel resistance and hygiene, which is the central trade in feeder choice.
• Capacity is not a virtue. A large hopper in a quiet yard holds seed long enough for it to spoil.
• Every design here has one job it does better than the others, which is why most yards end up with two or three rather than one.
• Fill a feeder to what your birds clear in seven days, whatever its capacity says. A feeder should be empty when it is due a wash.

Every feeder design is a set of compromises between four things: how many species can use it, how well it keeps food dry, how easily it can be raided, and how much work it is to keep clean. No design wins all four, and the ones that market themselves as winning all four are usually trading away the first.

The trade that runs through the whole table#

Read the platform row and the caged row against each other and you have the entire argument.

A platform feeder takes every food in the dataset and is used by every guild of feeder bird, from doves on the tray to woodpeckers on the edge. It also has no defense at all against squirrels, no protection from rain unless covered, and it accumulates droppings on the surface birds are standing on, which makes it the highest-maintenance design in normal use.

A caged feeder stops squirrels, grackles and starlings without any mechanism to fail. It does that by excluding anything larger than about a house finch, which means you have also excluded cardinals, jays and grosbeaks.

There is no design that keeps out the birds you do not want while admitting all of the ones you do, because "unwanted" is defined by preference rather than by size, and feeders can only filter on size, weight and food type. Understanding that saves a lot of money.

What the scores mean#

Squirrel resistance (0 to 5). How the design performs against a determined gray squirrel with no help from pole geometry. Zero means the design offers no obstacle at all. Five means a squirrel cannot reach the food even standing on the feeder. Correct baffle placement can lift any of these to effectively five, which is the point made in the baffle calculator.

Weather protection (0 to 5). How well food stays dry. Roof depth and overhang do most of the work here, along with whether the food sits in an open surface or inside a column.

Typical capacity. A midpoint for the design, in fluid ounces of food. Read this as a warning rather than a feature: a feeder should hold roughly what your birds will eat in a week, and no more. Anything larger is storing seed outdoors.

Capacity as a liability#

The capacity column is the one people shop on and the one that should make them cautious. A feeder does not preserve food. It holds it outdoors, in a container birds stand on, so every extra ounce is another day the food waits.

The fill-rate rule: put in what your birds clear in seven days, whatever the feeder holds. Seven days because that is the shortest routine cleaning interval for a seed feeder in warm or damp weather, and a feeder should be empty when you take it down to wash. Seed still sitting there at the interval means you are either throwing food away or topping up onto old seed, and topping up is how feeders go bad.

Worked at our conversion of about 1 lb of black oil sunflower to 25 fl oz of volume:

A 100 fl oz hopper in a top-row yard holds eight weeks of food. Birds take from the top, the bottom two inches never move, and the first humid week sets them into the plug that seed feeders die of. Measure your own rate once: mark the level, read the drop seven days later, multiply the fraction gone by the capacity, then repeat in June because the two numbers will not be close.

Capacity earns its price in three situations only: a property you visit weekly, a cold snap that empties a hopper in two days, and a roofed design that protects food while it waits. Otherwise a big feeder is a small feeder you clean more often, and a source of moldy seed.

Matching designs to guilds#

Most yards need two or three rows of that table covered, not all six. Which rows depends on your region and habitat, which is what the species index and the feeder selector are for.

Pairs of designs that duplicate each other#

Most overspending in bird feeding is a second design that does what the first already did. Three pairs are worth checking against the table before you add hardware.

Tube and caged. A caged feeder handles everything a tube handles (black oil, hearts, safflower, shelled peanuts) plus suet, at the same 40 fl oz capacity, and scores 5 for squirrel resistance against a tube's 2. The only access it removes is for birds larger than about a house finch, which a bare tube with short perches was already excluding. If squirrels are the problem, a cage replaces a tube rather than joining it. The same logic pairs a hopper with a weight-activated feeder: both are large roofed seed feeders scoring 4 for weather, and the mechanism only adds exclusion.

Platform and ground tray. Every food the ground row carries is also on the platform row, and the species using one mostly list the other: juncos, sparrows, towhees, doves. A platform set low is a ground feeder. Run both for separation, not for coverage.

Nyjer feeder and a tube of hearts. Across all 34 species in the matrix, not one rates nyjer above sunflower hearts. Goldfinches and siskins tie them at the top of the scale, and house and purple finches rate hearts higher. So a nyjer feeder reaches no finch that hearts cannot. What it buys is exclusion, because house sparrows, starlings, grackles and blackbirds rate nyjer at 1 or 0. Buy it for a mobbed yard, or not at all.

One pair looks redundant and is not. Oriole and hummingbird feeders both hold nectar, but the two hummingbirds in the dataset use one design only, while orioles use an oriole feeder, a platform or a mealworm dish. Neither stands in for the other.

Two worked stations#

Both come from crossing the feeder rows against the species records: a species counts as reached when it lists that design and rates at least one food the design handles at 3 or better. These are counts from the dataset rather than from observation, and the caged and weight-activated rows score low on any count made this way, because species records name the underlying design a bird uses rather than the squirrel-proofing built around it.

Three feeders. The widest combination in the dataset is a platform or tray, a suet cage and a nectar feeder, which between them reach all 34 species we score: the cage takes the clingers that ignore a flat surface, the nectar feeder takes the two hummingbirds nothing else reaches, and the platform takes the rest. The price is labor, since the platform scores 0 for squirrel resistance, 2 for weather and worst in the table for mess. The quieter version is a tube, a suet cage and a low ground tray: 29 of the 34, giving up the hummingbirds, the American Robin, the Baltimore Oriole and the Evening Grosbeak, and removing the one surface every bird stands on. Add nectar in season and it is 31.

One feeder. A covered platform: 28 of 34 species, against 17 for a hopper and 12 for a tube. That is the honest answer to which single feeder attracts the most birds, and also the one that fails most often, because it puts every bird in the yard onto a surface they stand on. It needs a roof and a real weekly clean with the bleach routine. If either is doubtful, take the hopper and its 17 species: you lose the ground feeders and the fruit eaters, and you gain a roof and food that never sits where birds perch.

What this table cannot tell you#

It says nothing about build quality, and build quality is what determines whether a design that should work actually does. Two things are worth checking in person on any feeder: whether it comes apart completely for cleaning, and whether the ports are metal or plastic where a squirrel can reach them. Those two questions predict more about a feeder's useful life than the price does. See feeder materials.

The three build details that decide service life#

Within a category, three attributes separate a feeder that lasts a decade from one that is rubbish in two winters. None is scored above, and all three can be checked in thirty seconds with the feeder in your hands.

Full disassembly. Every internal surface must be reachable with a brush and every part must come off without tools. The test is practical: could you take this apart at a sink in a minute and rebuild it with cold wet hands? Glued bases, fixed inner tubes and bodies needing a screwdriver all end the same way, with the feeder rinsed instead of washed, which is how it becomes a disease risk.

Metal where teeth and bills reach. Ports, perch sockets, lids and hangers are the failure points. A squirrel can open a soft plastic port into a hole in an afternoon, and after that the feeder leaks every time it is filled. Woodpeckers widen plastic suet openings more slowly and just as surely. Metal costs more once and removes the failure mode for the life of the feeder.

Drainage. Water gets in; the question is whether it gets out. Look for holes at the lowest point of the base and a raised or mesh floor holding food above any water that pools. An open tray with a solid bottom is a dish of soaked seed after one storm; the same tray with a mesh floor drains by afternoon. On tubes, check the base cap, because seed left standing in a sealed base is the commonest way a tube feeder ends. Our working judgement, from replacing our own feeders rather than from any study: those three predict service life better than price does.

Frequently asked questions#

Which bird feeder attracts the most species?

A covered platform or tray feeder, because every bird can land on it and it accepts every food from sunflower to mealworms to fruit. It is also the design that needs the most cleaning, offers the least protection from squirrels, and wastes the most seed in rain. That trade is the core of feeder choice: reach and convenience pull in opposite directions.

Are squirrel-proof feeders worth buying?

They are worth buying when you cannot solve the problem geometrically, which mainly means hanging sites and small yards where a pole cannot be placed 8 ft from everything. If you can mount a pole in open ground, a baffle at the correct height is cheaper and more reliable than any mechanism, and it protects every feeder on the pole rather than one.

What does weather protection mean in this table?

How well the design keeps rain and snow off the food. A hopper with a deep overhanging roof scores high, an open tray scores low, and a tube sits in between because rain runs down the outside and wets the seed at the ports. It matters more than people expect: wet seed is the beginning of nearly every feeder hygiene problem.

How big should a bird feeder be?

Big enough to hold a week of your yard's consumption, and no bigger. Measure it once: mark the seed level, wait seven days, read the drop, and multiply by the capacity. A quiet yard eating half a pound a week needs about 13 fl oz in the feeder at any time, which is a third of a typical tube and an eighth of a large hopper. Anything beyond that is seed stored outdoors, and seed stored outdoors is what turns into the damp plug at the bottom of the column.

Do I need a nyjer feeder as well as a sunflower feeder?

Only for exclusion. Across the 34 species in our matrix, not one rates nyjer above sunflower hearts: goldfinches and siskins rate both at the top of the scale, and house and purple finches rate hearts higher. So a nyjer feeder does not reach finches that hearts cannot. What it does is exclude almost everything else, because house sparrows, starlings, grackles and blackbirds rate nyjer at 1 or 0. If your seed feeder is mobbed, that is worth buying. If it is not, it is a duplicate.

Sources and further reading

1. Types of Bird Feeders Cornell Lab of Ornithology, Project FeederWatch Feeder designs and their typical species
2. Methodology Bird Feeder Web How the five-point scores were assigned