User:Vaticidalprophet/A fool's guide to writing a DYK hook

The Did you know section of the main page is a common entryway to main page processes, a fulfilling reward for a job well done, and a great way to get some eyes on whatever topic you care far too much about. Its trivia-pub-quiz format gives it an unusual clickbait pressure the rest of the Big Four (TFA-DYK-ITN-OTD) don't quite have. DYK requires articles be new or improved, but it also requires they be interesting, or at least have an interesting fact ("hook") that can be desperately mined from them. This is hard to judge. The current wording of the criterion (previously "interesting to a broad audience", and still de facto discussed that way by many) is "likely to be perceived as unusual or intriguing by readers with no special knowledge or interest"; no one quite agrees what is unusual or intriguing, or what constitutes an impractical level of background knowledge.

I've written some significant number of DYK hooks and promoted (moved into "prep sets" that eventually appear on the main page) several hundred. Across my non-image DYK hooks (too few image hooks to measure), I've drawn an average of 382 views per hour; all non-image hooks over the same period draw an average of 217 vph. This is to say, to a greater or lesser degree, I know what I am doing. I'm also a picky promoter who mostly takes hooks I think have a fighting chance at DYKSTATS, and regularly follow up hooks I promote once they make it to the list to see where they land. It's possible for a savvy DYK regular to notice both trends and substantial randomness. Grant me the wisdom to know the difference.

Prelude

Allow us to consider two hooks:

... that XYYY syndrome, a chromosome abnormality in which a man has two extra Y chromosomes, has only been recorded twelve times?

... that with eight nominations, Peter Hylenski has been nominated for the Tony Award for Best Sound Design more often than anyone else – but only won once?

These are both real hooks by me. I wasn't sure what to expect for either of them. Neither is about the most incorrigibly mainstream of subjects; both require a little pre-knowledge. Nonetheless, both have unusual facts. I expected them to land somewhere in the middle of the DYK bell curve rather than the tails. Both ran for 12-hour slots, the former in April 2021 and the latter November that same year.

They are -- respectively -- my highest- and lowest-scoring hooks ever. The former received nearly 19k views, drawing 1,553 views per hour during its time in the sun and becoming the second-highest non-image hook of the month. (It even outscored the hook just below it in its set, ... that Hitler Nababan was beaten by an angry mob after posting a meme to a WhatsApp group?) The latter broke barely over 800, with 68 vph -- the "you have to be trying to get here" depths.

I've mulled over these a lot. I've written many hooks I think were better than the former, and many I think were worse than the latter. If you'd asked me to predict before I wrote either which would land where, I think I'd guess wrong. Over time and tracking hundreds of hooks, I've been able to determine more about how they landed where they did. DYK is more art than science, and a huge chunk of it is outright randomness. But where would we be, if we didn't try to bottle lightning?

Overview

The still-best guide to article writing, from which this guide takes its name, gives the advice that the reader is an "intelligent fourteen-year-old" -- someone bright and curious, but without extensive life experience and inclined to pick out the juicy bits. This is good advice for writing articles, and the best there is for writing hooks. People who read the main page in-depth (and have any chance of non-accidentally clicking on a Big Four) are a small subset of every hit it gets, and if you went to school in the Wikipedia age, you'll remember a huge subset of it was kids in class trying to look busy. You probably were that kid, many times over. If you keep this in mind, you will write good hooks.

There are two basic requirements for a DYK hook:

It needs to be unusual
It needs to be pithy

"Unusual" depends heavily on the subject. Some subjects are inherently unorthodox enough for them to draw clicks. This is covered more in the next section; it'll become noticeable, as you review hooks, that there are loose patterns in what breaks 1,000 vph even with equally good hooks. In most cases, though, your floor is at least "not a trivial extrapolation from the subject matter". Consider these successful examples:

... that Blockchain Chicken Farm is about chicken farms run by people who have never heard of blockchain?
— 488 vph in September 2023

... that Richard Gerald Jordan has been sentenced to death four times?
— 640 vph in December 2022

... that Jewish-Russian Zinaida Vengerova, a pioneer in Russian decadence, allowed a circle of intellectuals to drink her blood in a ritual described as anti-Semitic?
— 431 vph in February 2021

These hooks work differently to one another, but they all share a core of presenting a contrast.

Pithiness is simpler. Write hooks like every word might set your house on fire. You are writing too long. Write shorter. Compare lengths at the top and bottom of DYKSTATS. Every word you can trim from your hook is a boost in views. This is not trivial; determining what's extraneous and what's necessary context is a lesson hard learned. Many people, terrified that their hook might be insufficiently accessible to the general population, desperately layer context upon context so people get how funny the joke is. Steer yourself away from this.

...but don't divest yourself of context completely. Every once in a while, you see a bland hook at DYKN with a comment going "this is actually really interesting, because X". Put X in the damn hook. This usually happens in topic areas where many people are vaguely familiar with them, but most lack in-depth knowledge -- see #Sports and pop music for more on this failure mode.

Genres

DYK hooks fall into rough clusters. Some of these do better than others. There is disagreement over whether "every article has a hook" -- I personally think most "some things aren't hookable" is a failure of imagination, but it's true that some articles have a lower ceiling than others. Not everything will have a gold-standard hook, a sharp, succinct, unusual idea, but most subjects anyone cares to write about in the first place should have something usable.

For all of these I'm giving examples of successful hooks, because you can't very well pick on hooks you didn't write yourself. This should not be taken as meaning all these strategies are universally successful -- in fact, several have prominent and prevalent failure modes. The view counts themselves for the general-purpose hooks should also not be taken at face value, rather than the general theme that some ideas are successful. A strategy with lower-viewed example hooks is not a "worse" strategy than another; in many cases the opposite is true, if a given strategy is a high-variance one. Study DYKSTATS before embarking on a specific route, and see what articles similar to yours have landed where. The failed hooks will often be more educational than the successful ones.

To contextualize these examples: as of 2023, you can assume a median of roughly 250 views per hour for a non-image hook. These numbers vary drastically over time; if you are chronologically displaced, check your local standards. A non-image hook with at least 400 vph is batting well above average. One above 800 is outstanding, and one above 1,000 is spectacular. Once you exclude multi-hooks, which break the stats by consolidating all their (usually individually poor) views, there are very few non-images breaking four figures each month.

The obvious

Some DYK strategies are obvious. Some subjects are simply more "clickable" than others, and these articles have a significant lead. If you've ever written for attention (e.g. running a Youtube channel or working in some forms of journalism) somewhere else, you'll have a good sense of what these are. These three strategies overlap heavily, but wordplay in particular is applicable to a decent variety of articles.

Inherently interesting

... that some men purposely have sex in order to become infected with HIV?
— 1,315 vph in June 2021

... that an air traffic controller's confusion of two planes' locations caused sixteen skydivers to drown after they unknowingly jumped over Lake Erie?
— 1,552 vph in July 2023

... that joking about the Chinese military can cost you US$2 million?
— 1,256 vph in July 2023

For some subjects, their existence itself is remarkable. All you need to do is find the shortest, simplest way to express them. While this is the easiest starting point, don't get complacent! If you overcomplicate your explanation, you can fumble even this. There's an all-too-common mistake where a nominator assumes, or a reviewer insists, that some additional fact needs to be layered on the base hook. The resulting "did you know that X did Y and Z" Frankenstein's-monsters fall precipitiously from their height. There are contexts where you want to combine a strong core with another fact, but this is an expert strategy, not a beginner one.

Note that "interesting" here is a bit...tabloid-y. If you work the main page, you notice quickly that most biases people want to ascribe to editors are more fulminant in readers. If you believe a bias towards true crime or pop culture or anything else is reflective of some flaw in the editor base, filling preps should quickly disabuse you of this notion (also noticeable at OTD, where serial killers sometimes outscore the TFA). The comparison has been made between what is inherently popular for DYK purposes and what could appear in a Ripley's Believe It or Not! reader; this is, so far as I'm concerned, exactly what you should expect when titling your section "Did you know" and presenting it as a collection of fun facts.

Wordplay

... that Dr. Young's Ideal Rectal Dilators were forcibly withdrawn after officials clamped down on them?
— predates DYKSTATS, but come on

... that it's pronounced "gif", not "gif"?
— 1,453 vph in December 2021

... that Nancy Reagan was hexakosioihexekontahexaphobic?
— 2,392 vph in August 2023

Hooks that play with language itself tend to be popular, as long as they don't stretch it so far as to be incomprehensible (a failure mode I've seen, though sometimes one I've seen in the top half of DYKSTATS). This is a skill, and not everyone has it. It's of course also necessary that the article itself permit such wordplay. The first two examples here both overlap highly with "Inherently interesting" -- I don't know if it's even possible to write a sub-400 vph hook about rectal dilators, though I'm sure someone could with sufficient dedication -- and the last mentions American politics. Still, you'd be surprised sometimes what you can spot in an article when you really look through it.

These hooks can be contentious. Many of them get pushback from nominators, queuers, or ERRORS aficiandos. Many more get the request that they appear specifically on April Fools' Day, rather than year-round DYK. The latter is a thorny issue. DYK traditionally runs themed humour sets for April Fools', which are great fun, but they mean the hook appears alongside strong competition rather than having the chance to shine for itself. Consider whether you'd rather have a big fish in a small pond, and don't accept a binding/certain suggestion of a date request if you aren't sure.

Debauchery

... that known books bound in human skin include a highwayman's memoirs bound in his own skin, a novel about a man being left by his lesbian wife, and a BDSM erotic poem?
— 1,100 vph in September 2023

... that the Bill & Melinda Gates Foundation supports the creation of shit flow diagrams?
— 930 vph in August 2021

... that a dead woman's knickers helped to solve the Charing Cross Trunk Murder?
— 747 vph in August 2022

Anything that's sexual, true-crime-based, or vulgar will probably do numbers. This is not an all-purpose strategy. There is a limit to this. It's also inappropriate for some topic areas (e.g. almost all BLPs). Nonetheless, if you have a subject that it works for, it might help. It's an obvious one and can be very fun, but do it with caution.

You might notice all three of these strategies overlap significantly. This is not coincidence. There's a kind of article that does spectacularly well at DYK for its own sake. Nonetheless, you can succeed at DYK even if your subject is much less obvious.

General purpose

What if your article is less obvious? There are a number of strategies that work generally well at DYK. The first of these in particular isn't necessarily applicable to all articles, but most subjects can have something pulled out of them.

Visionary

... that the play-by-email game Blood Pit was so complex that even its programmer had trouble winning?
— 879 vph in September 2023

... that tens of thousands of New Yorkers traveled to see Dreamland burn?
— 715 vph in August 2023

... that the newly discovered red tigrina may already be extinct?
— 783 vph in July 2023

This is almost one of the first categories. These hooks have a strong, singular idea, expressed succinctly. This is the gold standard, and the thing to use if you have an article that offers it. Do whatever you can to get this one to run. Do not offer alternatives that might tempt a reviewer or prepbuilder over them, unless you're really, truly unsure that it's approvable in the first place. Not every article has one of these, and that's okay. Failed visionary attempts, such as where the singular idea is bland, clog up the bottom of DYKSTATS. This is the paradox -- if you have a true visionary you need it to run, but if you have a false one you'd be better going for a usually weaker genre, and there is just enough randomness built into views that you can't always tell which you have.

One real spanner in the works is that visionaries are exactly the type of hooks that showcase the issues with strict readings of "no special knowledge or interest". As discussed later, many successful visionaries assume knowledge of some non-universal concept. Is your non-universal concept commonly known enough for your visionary based on it to be successful? Is your reviewer wrong to reject it? Great questions! These get even more fun once you add prepbuilder discretion in which of several hooks to pick (not always flawless, especially with non-universal visionaries), admin queries at WT:DYK if the queuer is one of the ones who challenges hook interestingness, and everything that has ever happened at ERRORS in history. You'll get this wrong sometimes, and that's okay. Keep trying. It's the best there is, when you get it right.

Lists

... that Ben Phillips replaced his friend's hair gel with superglue, put Viagra in his sports drink, and placed him on a lake while he slept on an inflatable mattress?
— 866 vph in July 2023

... that the construction of 666 Fifth Avenue involved cartoon decorations, a papier-mâché Santa Claus, and a 78-by-155-foot (24 by 47 m) American flag that ripped as it was being unfurled?
— 518 vph in August 2021

... that for three years, an illegal gold-mining settlement on the Amur river went on to host high-class hotels, have public healthcare, and even have a casino?
— 565 vph in April 2021

An easy way to write a hook is to list several interesting things from the article and hope they combine greater than the sum of their parts. This is more or less the only time you should be writing a hook above 150 characters. This strategy combines well with others (note its use for List of books bound in human skin), and is one of the few good ways to hook, well...lists. Ideally, subjects in the list should be incongruous with each other, the subject, or both -- the second example here showcases these well -- but strong enough subjects or items can permit more descriptive lists.

Connections

... that oversimplifying a fictional character over the course of a show's run is called Flanderization, after Ned Flanders of The Simpsons?
— 1,526 vph in March 2021

... that Beyoncé paid tribute to O'Shae Sibley, a gay man who was killed after vogueing to her music, on her official website?
— 592 vph in August 2023

... that Catechumen, a Christian first-person shooter, was funded only in the aftermath of the Columbine High School massacre?
— 1,019 vph in September 2023

Definitionally, subjects that run DYK tend to be obscure -- if they aren't GAs, they had to either be redlinks or underdeveloped. Drawing a connection to a better-known topic is a useful way to get attention. Having said that, newcomers often overestimate how broadly this works. Some topic areas (most markedly, songs by popular musicians) perform poorly on average despite obvious connections to a well-known subject; there are confounders here, such as these subjects being hard to hook and tending to run in too close proximity, but this is not a guaranteed views-printer. Still, if you have it, it's a safe bet.

Like visionaries, there's complex overlap between these hooks and hooks which technically require "special knowledge or interest". At a certain point they cross over with "inherently interesting" (Flanderization is arguably a case of this), and at another certain point you get too niche, but human interests are fractal. Hooks connecting to celebrities don't seem to do as well as those connecting to non-celebrity pop culture, like fiction ("Dark Souls of X"), politics (we have so many Trump hooks), or true crime/shocking subjects/etc (the Columbine example here). Consider this when weighing up your options.

A risk for these hooks is that they can direct traffic to the better-known connection, rather than the actual subject. In particular, do not suggest an image hook where the image is of the connection! If you get your monkey's-paw wish, the views will go there, not to the article you intended to highlight. This preference for the subject of image hooks is exploitable at OTD and probably (insufficient firsthand experience) ITN, but it's an entirely unintended consequence at DYK.

Quotes

... that the theme song for the tabletop role-playing game F.A.T.A.L. was described by a reviewer as "sound[ing] like the Cookie Monster chasing a drum kit being pushed down a flight of stairs"?
— 454 vph in February 2021

... that one journalist said that the proportions of the limbs of the statues at the Monumento a los Indios Verdes are a "sin against anatomical laws"?
— 561 vph in December 2021

... that Lord Adolphus FitzClarence was said to be too slow due to his "most necessary appendage"?
— 928 vph in June 2023

One reasonably safe bet is to quote someone who already unknowingly wrote the hook for you. This is one of the few non-agonizing ways to hook fiction (discussed more later), so it's seen extensively for such subjects. The big thing to remember here is that negative reviews are better than positive. Negative or at least comedic reviews are a lot more fun to read and a lot less likely to see promotionalism accusations. If you're hooking from a work itself, draw the weirdest quote you can (... that Zealot: A Book About Cults suggests someone might join a cult when their lifestyle "doesn't let them have enough sex with aliens"?).

As can be seen from the final example, this format also works for quotes that aren't reviews. While you'll often have a better example in such cases, they're a good way to introduce some mystery or misdirection.

Like connections, this is a strategy with a high failure rate, and it doesn't tend to go stratospheric even when it works. If you have a good ear for it, though, you can turn out consistent solid performers.

Juxtaposition

... that Bird Thomas Baldwin, the director of the first child research center of its kind, died due to a shaving accident?
— 697 vph in August 2023

... that Ivan Beshoff, the last survivor of the mutiny on the Potemkin, emigrated to Ireland where he established a fish and chip shop that is still run by his descendants?
— 608 vph in September 2023

... that the New Amsterdam Theatre, once described as "a vision of gorgeousness", later had dead cats in the basement and mushrooms growing through the floor?
— 465 vph in October 2021

This is a whole phylum of hooks. The core of this superset is that they intentionally invoke two ideas that are incongruous with each other or make for a more fun fact than they would independently. Many other hook genres are essentially subsets of these -- lists, for instance, often amount to compound juxtapositions.

A lot of hooks are somewhere under the juxtaposition umbrella. It's the best fallback when you aren't sure what else to go for, so they can swing wildly in quality. Common juxtaposition genres, with some fake illustrative examples, include:

X, a member of Y profession, was a former member of/trained in Z different profession ("supermodel majored in chemical engineering")
X, who did Y, supported Z ("army officer attended peace demonstrations")
Rags-to-riches, riches-to-rags, or similar narratives ("millionaire died homeless")

Over time, DYK hooks trend roughly in this direction. If you know how to write a good juxtaposition, you can survive DYK. However, this doesn't mean it's the be-all and end-all of hooks. In particular, juxtapositions are worse than visionaries or inherently-interestings. Many nominators or reviewers, worrying that a visionary isn't strong enough, try to layer fact upon fact and end up with a muddled mess. Much of this is a failure of the Iron Law of DYK: be concise. The fundamental problem with juxtapositions is that they are, by their nature, not concise; if your choice is between a potentially muddled juxtaposition and a potentially bland concise hook, be careful.

Juxtapositions are stronger the more recognizable the ideas are. Many juxtapositions amount to trivia and fall to the bottom of DYKSTATS. This is to say that they're markedly more vulnerable, as a genre, to the "no special interest" clause than some other hook genres. See the Hylenski hook in the prelude for a good example of a failed juxtaposition -- these are generally common in topic areas where the general reader is unlikely to recognize what makes something interesting, or where the "unexpected" element is a bit of a wet noodle.

Just get weird with it

... that you can't be .sexy in private?
— 712 vph in June 2021

... that vampire amoebae are naked?
— 426 vph in May 2023

... that the dwarf merry widow is not very brave?
— 606 vph in September 2023

Hell, it works. This hook genre overlaps with wordplay, but it's a bit more of a blunt instrument. Their relationship to the article's content tends to be somewhat strained. From a purely informative perspective, they're not spectacular. But we always need quirkies, and it's fun to try contort something into this.

Note that none of these hooks are above fifty characters long. Write accordingly.

Other strategies

There are many ways to write a hook. The above strategies are generally safe, but a number of others can be used to produce well-performing hooks. Again, only successful examples are given for ethical reasons. This is even more unrepresentative than for the previous categories; definitionally, they have a higher failure rate than any of the general-purpose strategies except juxtapositions. Hone your judgement.

Just the facts

... that Dark Archives reveals that most books bound in human skin were made by respected doctors?
— 443 vph in July 2023

... that pentasomy X, in which a girl or woman has five X chromosomes, is sometimes mistaken for Down syndrome?
— 967 vph in May 2021

... that medieval monks could be punished by being walled up alive in a special ecclesiastical prison called the Vade in Pace?
— 592 vph in September 2023

Purely factual hooks, without a particular twist on them, can work for some subjects. This is sort of a milder version of the inherently-interesting strategy. The real tricky thing about it is that you don't necessarily know if your subject is good enough for this until you've done it -- I'm still trying to piece together how the chromosome disorder hooks did so well (the lowest performer was 727 vph). These are nonobvious, which is exactly the problem. If you can write a literal description of your subject and see where the "unusual or intriguing" element lies, you might be better off trying for this than overcomplicating it by layering twist upon twist.

The first example here is particularly illustrative in two ways. One is that it's a hook about a work of media that solely references that media. You can do this for nonfiction, but not for fiction, because the main page might burst into flames or something. This is why nonfiction-book hooks do numbers while fiction-book hooks all sound something like "did you know the author of X, something you have never heard of and have no reason to care about, wrote it because they were bullied in high school?". The other is that it has a link to another, more surface interesting-looking article in it. After all was said and done, that article got about as many views as it did -- which sucks, because that article was terrible! Do not include tempting non-bold links in your hooks.

Firsts

... that the 1979 video game Superman was one of the first console games with a pause feature?
— 604 vph in July 2023

... that after becoming one of the Mongolian Armed Forces' first female recruits, Bolor Ganbold is now its first female brigadier general?
— 565 vph in May 2023

... that "She's a Woman" was the first song by the Beatles to include a reference to drugs?
— 705 vph in July 2021

These are contentious! I personally am a fan of first-to-do-X hooks -- as you can see from these examples, they do very well when the first-thing is recognizable to our audience. (A note: it was surprisingly hard to find good examples, not because there aren't many highly successful first hooks, but because too many highly successful first hooks were in the image slot.) Nonetheless, there's a real strain of thought that they're bland or unapproachable, or that specific variants of them are problematic (much ink has been shed on "first woman"). More genuinely problematic is that they're not always trivial to verify, which is where you can see the "one of the first" hedging in the Superman example.

Firsts can be vulnerable to mutilation if DYK's hivemind-pendulum is currently on the "first hooks are bad" end. The argument goes that, obviously, something being the first-X isn't interesting enough for its own sake, so we need to add an additional clause ("did you know X, the first to do Y, said Z?"). Remember the Iron Law of DYK. Do not turn a good first into a lurching synthesis.

This is the single subject area for which "study your DYKSTATS equivalents" is the most important advice -- perhaps even moreso than juxtapositions. The nature of the structure means I can only give examples of successful firsts. But firsts routinely fail, because this is another topic area where general-audience recognizability really, truly matters. To be clear, for both firsts and juxtapositions, "this hook will probably land low on the scoreboard because it's not recognizable to a general audience" is not necessarily a bad thing -- a hook can be well-crafted and basically good, and still only recognizable to a narrow audience, and accordingly wither and die. I still think the Hylenski hook was good -- I'm not that much of a theatre guy, and it's still an interesting juxtaposition to me that someone can be the most prolific nominee for something but win it few times out of that. But readers aren't interested in sound designers, and they aren't interested in who was the first narrow-demographic to do some incredibly-narrow-thing, and sometimes you gotta roll with the punches. Find the best hook you can and use it.

Misdirection

... that Darth Vader's anal shield has a "pronounced bell shape"?
— 821 vph in April 2022

... that Mr. Bean accurately predicted the results of many American elections?
— 898 vph in December 2021

... that there are dragons in Buckingham Palace?
— 1,338 vph in April 2021

These are fun -- basically the logical conclusion of quirkies. The problem is they can be too fun. As you can extrapolate from the months, many of these are forced into April Fools' sets. The AF set is fun, worth wracking your brain for a good one, but there's a fairly broad range of hooks that are just real enough that you can run them any month while just trollish enough that they get suggested for AF. The main consequence of this is to take viable quirkies out of the year-round pool while clogging AF up with candidates well before April rolls around, which is a bad outcome for everyone -- there are tons of viable AF potentials every year, and the phenomenon where people start holding things in June means it's harder to start plotting out a good AF set. (My ideal is something like "any AF hold before March is challengeable".)

These hooks are also vulnerable to ERRORS criticism. ERRORS is...it's vitally important. If you ever get involved in DYK's backstage, or deep in any other mainpage process, you need to watchlist it. It's one of DYK's major checks-and-balances. But a large chunk of it is inane. (Did you know it's possible to get pblocked from the ref desks for disruption? I do, because I've seen underlined names pop up on my watchlist from ERRORS.) Misdirection hooks are exactly the thing to be careful with, because a lot of the time criticism of them will be wrong, but sometimes it will be right, and you can't always tell from the inside when you've crossed that line. Be careful, when writing misdirection, that you don't left-turn into inaccuracy. Double-check the hook with someone stricter than you are.

Like debauchery hooks, this is not a license to do whatever you want.

Image hooks

... that the Bank of England has a device to prevent unwanted deposits (pictured)?
— 2,750 vph in July 2021

... that Kid Canfield (pictured) is the first known person to die live on radio?
— 2,452 vph in May 2021

... that Ava Cherry (pictured), David Bowie's partner and muse, spent a year searching for him in Europe after he cancelled a tour of Japan on which she was to be a backup singer?
— 1,358 vph in March 2021

Someday, after ages of searching, you will finally have a nomination with a usable image. Jumping for joy, you submit it for DYK, awaiting the opportunity to take the best spot of all -- the Image Hook! Already anticipating how many views you're going to get, you watch as the reviewer happily rubber-stamps the image and keep your fingers crossed as it waits at DYK. Then you watch someone promote it to slot five. Image hooks are the one true "can't please everyone" at DYK. There is one image slot in every set, and too many nominations to use them all. The first rule of image hooks is accepting that you will never get an image hook, not once; then you will be enlightened.

Image hooks perform differently to non-image hooks. While non-image hook performances generally have some vague relation to one another month-by-month, any attempt to give a true "average" view count for an image hook is folly. At the time of writing, other people (not you) can vaguely assume 550-600 vph as the "average" for an image hook, but the small sample size means this depends much more on individual hook performances than it does for non-image hooks. Still, other people (not you) treating 600ish as the "number to beat" in the same sense as 250ish for non-images is...reasonable enough. 1,000 vph is a more common achievement/realistic goal for images than non-images, and 2,000 or more is within the realm of possibility, though happens at a rate well under once a month.

Something unintuitive about image hooks is that the image, itself, doesn't seem to matter very much. The second example here is a good demonstration -- the image is not high-quality, but it was still the best of its month by a substantial margin. Don't worry too much about image quality -- you won't get the slot either way, and when other people (not you) do, it won't impact performance that much. One disappointing corollary of this is that great images don't benefit much either. Beautiful landscapes or detailed artworks or similar seem a bit lost at thumbnail. It's still nice to put something pretty up on the main page when you can, but don't expect it to be the key to high views.

From a prepbuilder point of view, I like using the image slot to prop up hooks that are good but might get buried a little in the middle of the list (... that Glyn Johns (pictured) was the recording engineer for Led Zeppelin's debut album, the Beatles' Get Back Sessions, and the Rolling Stones album Let It Bleed?) -- this tends to get them respectably high views (1,100 vph in July 2023, in this case). In general, image hooks seem even more vulnerable to "much of DYK is random" than non-images.

Supply-demand for non-bio images tends to be a little less horrible than for bio images. Other people (not you) have somewhat better odds of landing in the slot if they have a non-bio image, but it's still poor odds on average. It's possible to carve out a niche of getting many more image hooks than average by working in an area where non-bio images are common, like architecture, though this can backfire.

Assorted other points:

You won't get the image slot on April Fools'. I mean, you won't get it either way, but other people (not you) won't either. If other people (not you) want an image slot, don't ask for AF. Especially do not ask for AF nine months in advance.
I can't guarantee this generalizes across prepbuilders, but I personally prioritize article quality higher for image slots -- if there's a GA with a decent hook and a relevant image, I'm going to put it there every time. There aren't enough GAs or sleeper-GANs to put in the image slot every time, but that's how I do it when there is.
Please stop nominating images that aren't of the subject of the article. See #Connections for the practical consequence of doing this.
Per "image quality matters less than you think", if you're reviewing off a checklist version of the DYK criteria and it includes a "clear at thumbnail size" criterion, think very carefully before disqualifying images on its basis.

"No special knowledge or interest"

Special cases

Some article types perform strangely, or are hard to hook, or are just worth explicitly calling out for one reason or another.

Biographies

Biography articles (both BLPs and BDPs; if an alternative category emerges I assume it'll act the same) have strange view patterns. A small minority are extremely popular, while most barely beat Special:Random background noise. Not many fall between these extremes. This phenomenon is noticeable at DYK, where biographies can and do perform extremely well, but many more of them falter even for experienced hook-writers. Every hook I've nominated that fell below 200 vph was a bio. Follow this advice and do the best you can, but keep in mind any DYK-eligible biography just isn't that likely to be of true broad interest.

BLPs have the additional consideration that "hooks focusing on negative elements" are barred. Interpretation of this varies, but it rules out a lot of categories. BLPs who are notable for negative reasons may not be hookable at all. (He's no longer a BLP, but I remember the joke suggestion of ... that the American criminal justice system is guaranteeing a fair trial to Nathan Larson for his alleged child sex offenses? It's not technically a negative aspect, you see -- it's just talking about the system's commitment to fairness! To be clear, don't do this.) DYK's power means it has a history of ideological or commercial misuse; many people have tried their best to push through negative hooks on those they disagree with. Do not corrupt yourself.

Also, don't nominate BLPREQUESTDELETEs. I can't believe I have to say this.

Stand-alone lists

Lists are hard. Many of them fall low on DYKSTATS. Commonly observed list strategies include:

Listing interesting entries (the List of books bound in human skin example), as has been mentioned as a workable strategy for many articles
Referencing the list itself, common for articles split off subjects ("List of performances by X", "List of awards and nominations received by X", etc being structured as "did you know X has appeared in Y films")
Pulling a fact from an individual list entry

All of these can work, though the first and third work a lot more often than the second. Play around with the format, see where you land.

One common problem with the second category in particular is that it can fall into the connections paradox, where a hook drives far more attention towards something other than its subject. Do what you can to avoid this. One strategy I've contemplated is changing the scope of the bolding. Many hooks for, say, a list of performances are formatted something like ...that [actress] appeared in X films?; a more profitable method might be ... that [actress] appeared in X films?. Neither of those hooks are good -- unless the number happens to be remarkable ("has appeared in over a thousand films", "won a major award for every one of her roles", etc) -- but you can hopefully see the shape here. In practice much of the splitting seems to be from a desire to construct an image hook; as repeatedly mentioned, this is actively undesirable.

Accolade-lists ("List of awards and nominations received by X" and "List of accolades received by X") are a special case. There's an argument to make that these sorts of articles shouldn't exist. They sort of snuck into being a thing because FLC didn't set a precedent of explicitly rejecting them, so now there are hundreds. While I'm loosely skeptical about the idea there are large swathes of article conceptspace with no possible hooks, inasmuch as there's anything it's true for, it might be these. Tread carefully.

Fiction

In real life, no matter how unlikely anything is—the deaths of William Shakespeare and Miguel de Cervantes on the same date in 1616, or one man being struck by lightning five times—if it really happens, we do not question that it would happen. Our credulity is not stretched to the breaking point, causing us to stop participating in the world and go looking for another one that is more convincing. Thus, God can work with the most mind-bending coincidences, far-fetched plot devices, and perverse dramatic ironies, never giving a moment's thought to whether or not his audience will buy it. You do not have that luxury.
— "Why Your Job is Harder than God's", How Not to Write a Novel

One of DYK's rules is DYKFICTION, which holds that hooks for works of fiction can't be solely about the work itself -- they need to have a "real-world connection". This was unilaterally added years ago and inherited, and no one agrees how it should be interpreted. When you consider the constraints of good hooks (that they need to be concise and have the minimum possible number of clauses), it quickly becomes apparent that "complying with DYKFICTION" and "giving any context at all for what a work is" are frequently at odds. Some juxtapositions can pull off both, but that requires a work's plot itself be incongruous with some real-world element of it ("did you know that X picture book was written by a splatterpunk horror novelist?"), which is not a given.

No one agrees what DYKFICTION is meant to prevent. I think it should be interpreted as narrowly as possible as preventing in-universe hooks ("did you know that Ash Ketchum wants to be a Pokémon Master?"), which is a common enough and perfectly reasonable ruling. Others think it's meant to prevent any hook about something that happens in a work, using the argument that "because anything interesting could happen in fiction, nothing in a work can be interesting". This is an admission to writing bad hooks -- the example hooks given in these circumstances are always something like "did you know in X video game you kill three-eyed aliens?", which should be rejected because it's a bad hook (a bare statement of fact where the fact itself is too routine to pass). The fact "anything can theoretically happen" in a work of fiction does not mean anything unusual in a work of fiction is purely equivalent to anything else unusual. It is common for three-eyed aliens to appear in sci-fi action games; it is not common for them to appear in Regency romance novels.

The basis of high-reading DYKFICTION is "if it didn't exist, we'd be flooded with these terrible hooks and unable to do anything about them". This is only true if you do not consider reviewers empowered to query hook interestingness. There are points in the project's history where that was true, but it's not true now -- if anything we have the opposite problem (many reviewers haven't yet internalized the Iron Law of DYK and ask for too many additions). Given this, there is nothing gained by them but stupid arguments. There is a limitations-breed-creativity quality to some element of it, but once you've already gotten the idea of "don't use the work itself literally every time", there is no additional benefit -- it only rules out excellent hooks and sparks constant tension.

Because of these prohibitions, hooking fiction is a huge pain. Generally one tends to fall somewhere around "quotes from reviews", hoping something is funny enough to make it past the waterline. This has the interesting consequence that it incentivizes the nominator to represent a work negatively, because negative reviews are funnier (and hookier) than positive. There are areas of fiction (literature originally published in a language other than English, most severely) where the project has serious coverage gaps; the functional inability to write coherent DYKs on these clearly doesn't help the cause of writing them.

This specifically applies to fiction, and not nonfiction media, because of the project's complex history around and relationship with its fictional coverage (which has never had unintended consequences, not once). Nonfiction media routinely run hooks about the work without issue, which in turn routinely pull solid numbers. There is no moral to this story.

Sports and pop music

Subjects that are accessible and approachable to the general Anglosphere population tend to do better than those they're unfamiliar with. Normies are way into sports and pop music, so those hooks must do well, ĉu ne? Well, it would make a lot of sense if that were true, wouldn't it.

Paradoxically, these hooks tend to do poorly. There are a few reasons for this, but I think a big, slept-on one is that while these subjects are popular, knowing about them in-depth is less common. There are sizable populations of people into the nitty-gritty of sports or pop music (as we know from our excellent coverage of those topic areas), but the average casual fan doesn't have that connection. Meanwhile, because the gist of the subject is so widely understood, the author (who is someone with that nitty-gritty interest) assumes the casual fan must know this too, and writes a very inside-baseball (literally, as the case may be) hook that the casual fan doesn't grok. This is, to be clear, a just-so story. But I think there's something to it, having seen and promoted a lot of these hooks.

Alternatively, maybe readers just don't understand our genius. We've had some fantastic hookwriters in this area who still performed poorly on average. Who knows?

Pop music requires a special note. Song hooks tend to be clustered close together -- when an album drops, the album and its notable songs are all written in quick succession, thus DYK-eligible around the same time. Pop music is also an area particularly vulnerable to topic saturation. While DYK's bread and butter has a lot of subject matter repetition, "works by an individual artist" is a lot more visible than "radio stations", "Manhattan buildings", or "Indonesian politics". If you're working on an album, or any kind of ultra-narrow limited topic area (e.g. a single video game series), it may be good to request such hooks be intentionally spaced out.