Titles in Fantasy: Weighing What Works and What Doesn’t

scales-book It really is the seed of your entire marketing plan, arguably, even more important than the cover art, but many of the authors I talk to cower in fear before the comparatively small task of writing those 3-10 crucial words. The title. So, as I mentioned in a previous posting, I decided to conduct a little primary research on the matter. Thus was born my Fantasy Title Fights site, where real people weigh in on their preferences, picking the better (or least bad) title from two randomly chosen titles. Well, two weeks later and almost 4,000 head-to-head match-ups later, I’m now ready to share the results, so pull up a chair and let’s dig into this, shall we?

First up, let me start with some background statistics. As of right this moment, there have been 3,843 fights, and Google Analytics tells me those fights were conducted by a total of 634 visitors. So on average, each user conducted 6 fights and then moved on. There were a total of 27 different titles in the pool, each of which appeared in a matchup between 250 and 300 times. The titles were all created at random, using a couple of web-based fantasy title generators (this one and this one). Bear in mind that this has imparted a rather formulaic feel to the candidates, and several visitors have commented on that fact, but I don’t see this as a problem. If anything, the shared procedural origin of these titles should result in a collection of weak and unappealing titles. And that strikes me as a good place to begin our examination, because if I can show that people have preferences even among weaker titles, that should underscore the value of coming up with actually good ones.

For the record, here’s the list of titles that were judged, shown in alphabetical order.

  • A Conquered Goddess
  • Beyond Fury
  • Black Stone
  • Blood Song
  • Born in a Grave
  • Claw of the Sea
  • Dangerous Journey
  • Edge of the Solstice
  • Eye of the Demon
  • Fairy Tails
  • Hell Bent
  • Hills of Challenge
  • Honor Song
  • Lord of Discovery
  • Mage Fire
  • Oracle of Ice
  • Stone Heart
  • The Accidental Prince
  • The Blade’s Tale
  • The Creator’s Apprentice
  • The Death Forge
  • The Game Without End
  • The Legacy of Kings
  • The Nightingale Chronicles
  • The Pleasure Thief
  • The Street Wizard
  • Valor of the Crown

Were All Titles Created Equal?

The sad truth is that, in a world that can revere both Led Zeppelin and Air Supply, um, Marilyn Manson and Justin Bieber, if we poll enough people, the sheer variety of their aesthetic tastes might smear the entire list into an average melange. And if this experiment had produced that kind of result, it would suggest that the title itself doesn’t much matter – that we could pick any title, and some group of people would like it, while others wouldn’t. I like to believe, however, that titles do matter, and hopefully you’ll agree that the data seems to back me up on that.

When I first launched the site, and the first few users started recording their preferences, there was a significant difference between the most successful and least successful titles. This made sense, since the system had captured the judgements of about 50 people. In those early hours, when there had only been a few hundred fights recorded, the weakest title had been victorious in a paltry 14% of its fights, while the champ had gone all Hulk-SMASH, taking 87% of the bouts it appeared in. But as more users checked in to record their preferences, that window began to narrow. In fact, for a while, I worried that we really would see a flattening of the graph, with some people loving one title, and others hating it, in favor of some other.

But fortunately, the trend did not continue.

Overall, the gap did compress a bit, but the compression started leveling off once we hit 1500 – 2000 fights, and there is still a very strong preference-gap, which seems to have stabilized now. Over the course of the most recent 800 fights, the range between worst and best candidate has remained stable (24% – 66%). And regardless of the distance between them, the relative positions of the titles in the results list have not changed much since the first 500 fights, either.

All of this goes to suggest that, at the very least, there really are such things as “good titles” and “bad titles.” Moreover, there appears to be some degree of consensus among fantasy fans as to which titles are which. Now all we have to do is figure out what qualities influence that judgement.

The Dregs

BottomOften, we can learn more by looking at the failures than we can from the successes. So let’s start by looking at the bottom of the success chart. On the right, you’ll see the five weakest performing titles of the set. I have some personal theories about why they didn’t do so well, but I’ll save that discussion for the attributes section below. For now, I invite you to read them over and see if you note any commonalities among them.

Handy Tip: I have run these stats each time another 500 fights are fought, and in every case, Hills of Challenge has been in dead last position. If you’ve been thinking about using a title along those lines, you might want to give some serious thought to re-assembling your brainstorm team.

The Cream

TopAt the other end of the spectrum, there are the “good” titles. Unlike with the last place titles above, as the number of fights has increased, there has been a bit of jockeying among the top titles. Not a lot, mind you, but The Accidental Prince was the runaway champ for the first 1,500 fights, eventually falling to these other contenders in later bouts.

The case of Blood Song  warrants particular mention. At the time I drafted the list, I had read Anthony Ryan’s book, but I misremembered it’s title. Thinking that it had been called Song of Blood, I decided to keep Blood Song in the mix. For the first 500 bouts or so, this title placed in the middle of the pack. But as soon as I started getting visitors from the open invitations that I posted on Reddit and Google+, Blood Song started climbing the charts. I don’t know that the fact the book actually exists had any influence on the votes at all, but I do know that my first dozen visitors were drawn from friends and family – none of whom are much aware of the indie book market – and the later visitors were coming from online communities in which indie books (including Blood Song) are discussed with some regularity. Consequently, I tend to ignore the results for this one title, because I have to believe that a non-trivial number of respondents would have been influenced by more than just the title by itself.

Again, I invite you to muse over these titles and see if you can spot any reasons or qualities that might explain why these titles are more appealing to potential readers.

What Might Make Them Tick?

For The Cream, I’ve listed the top-producing 25% of the candidates, each of which won at least 60% of their fights. There’s nothing magic about either number, but I think it give us a good basis for culling, and focusing our explorations about what makes a good title. In particular, it’s that 60% metric I’ll be coming back to. If we could identify factors that produce wins in 60% or more of their bouts, I think we’d be safe to judge that, at the very least, the title is not hurting us in the market. But before we get into those results, let’s talk about what sorts of qualities of a title might be measured.

If you think about successful titles in literature, overall, there are a number of qualities that seem to recur. For example, many well-liked books have had titles based on Shakespearean quotes. Some much-read authors have a consistent naming convention, such as Robert Ludlum’s The Adjective Noun titling pattern. For a while, I considered trying to measure which titles were actually some sly reference to some other title or famous quote, but I quickly realized that this highly subjective. How many readers would know that Serenity Valley Blues was a popular culture reference? Would younger readers know the Shakespearean origin of a book called Serpent’s Tooth? So I abandoned the notion of trying to measure allusions like this, although I did hang onto the notion of titling patterns, as you’ll see.

But is that all there is to it?

When I began this project, I had two suspicions. I suspected that titles that called to mind an image of an identifiable, specific character, would do better than titles that did not. And I also wondered if titles that conveyed some specific action, conflict, or scenario would out-perform others. To my mind, the most engaging titles are the ones that serve as more than just labels – the ones that convey some micro-hint of story, character or situation. But it turns out that not everyone agrees.

I went through all 27 titles, and coded each of them, against 4 different criteria. Those were:

  1. Person – does the title seem to refer to a specific character
  2. Scenario – does the title invoke some specific action, conflict, or situation
  3. OfThe – does the title take the form Blank of the Blank
  4. Specificity – does the title convey either a person or a situation (combining #1 and #2)

After discussing this list with fellow author, Nathan Lowell, he suggested a few more conditions, and so we went through the list and added:

  1. LeadingThe – does the title start with a definite article
  2. Syllables – how many syllables does the title have
  3. Cadence – what is the pattern of hard and soft stresses on the syllables

The Binary Attributes

binaryOf that list of qualities, the first 5 are all binary attributes. A title either has that quality, or it does not.

From among them, only one performs as well as the (arbitrarily chosen) 60% wins that were reported by the most popular titles: Specificity. Of all the titles that referenced a person or situation, when competing against a title that connoted neither, they won their fight exactly 60% of the time. (For the purposes of these investigations, I ignored cases in which two titles with the same attribute faced each other, since the attribute in question was unlikely to have played a role in the reader’s decision.)

The Class Attributes

classThe remaining two qualities: syllables and cadence, are of a classifying nature. I don’t believe that 3 syllables are better than 2, and that 4 syllables beats 3, but it is possible that the English-language ear has a taste for 5-syllable phrases, and maybe 12-syllable phrases, but hates titles with exactly 6. Might a short, staccato title be better than a long, meandering one? Is there such a thing as too long?

It turns out – in this experiment, at least – that there is no consensus on a particular number of syllables that is best. 7-syllable titles (The Nightingale Chronicles, The Creator’s Apprentice) scored the highest number of wins, over other syllable lengths, but at 57%, they did not do so often enough to break through the 60% threshold we’ve set. And they were only marginally preferred over the second-best length of 2, shown in candidates like Blood Song, Black Stone, Mage Fire, and a few others. These two-fers were only half a percentage point behind the length-7 candidates, score wins in 56.5% of their fights.

At first glance, the other class attribute I explored did much, much better. The cadence measure, which was a metric initially suggested by Nathan, performed surprisingly well. Each title was coded for its sequence of S(oft) and H(ard) syllable stresses, and to my utter shock, 3 different stress patterns dominated the metrics list, outperforming my own weak-assed favorite, Specificity.

In third-best place, the stress pattern SHSHSH (The Accidental Prince) won 62.6% of its fights. But in fairness, since there was only one candidate with this pattern, it’s hard to tell whether it was the cadence or some other factor that gave it the appeal. In fact, this caveat holds true for the other high-scoring classes as well. The pattern SHSSHSS (The Nightingale Chronicles), scored 64.6% wins. And in first place, winning 66.3% of its fights, the pattern SHSSH, which applies only to The Game Without End.

And that’s the problem. Each of the three strong cadence classes had but a single title in them, and those titles performed well, so the cadence class scored well, too. Would those cadences score so well if they were each represented by twenty titles? One hundred? I don’t know, but I hope to explore that more fully in a new version of the site that I’ll be launching soon.

Take Home Message

To be honest, most of these metrics are sensitive to the fact that our population of titles was rather small. The most important finding, I think, is simply the fact that readers do see a difference in quality of titles. Your title does matter. So take some time, test them with friends. Don’t shy away from the job of writing a title just because it intimidates you. In the near future, I hope to announce a new version of Title Fights, in which authors can submit their own titles and let the public provide feedback, so stay tuned. But that system isn’t ready yet, so until then, you’re on your own.

As for more specific recommendations, it’s too early to tell. We need to run the data on a much larger pool of candidates before any useful generalizations can be drawn. But if you need to build a title today, and you’re looking for a little guidance, I find myself falling back on that 60% workhorse, Specificity. Craft a title that invokes a little conflict, or have it make the reader think about a specific character. After all, you’re trying to get them to read a story about specific people getting into specific conflicts, right? Maybe having the title start them down that road is a good way to go.

And if you think any of your writer friends might find this article interesting, please click on one of the share buttons below.

Character Cards in Scrivener
The Arcane Art of Notes To Self

About the author

Jefferson Smith is a Canadian fantasy author, as well as the founder, chief editor and resident proctologist of ImmerseOrDie. With a PhD in Computer Science and Creativity Systems compounded by a life spent exploring most art forms for fun and profit, he is underqualified in just about everything. That's why he writes.