A Statistical Approach to Twitter Pitching

I’ve participated in three Twitter pitch contests (#PitMad September 2014, #PitMad December 2014, and #PitchMAS December 2014) with my manuscript, Among the Red Stars, and between them, I racked up 24 favorites from agents and editors, so here I am to spill my secrets. How can you “machine” your Twitter pitch to make it more successful. What factors can you control to improve your chances, and what factors don’t matter?

I’ve erred on the side of completeness, so this will be a long post. For more helpful information, check out these posts by Lara Willard and Dan Koboldt.

The Pitches

I don't have any pictures for this post, so here's a cat in a drawer.
I don’t have any pictures for this post, so here’s a cat in a drawer.

For the September #PitMad, I carefully worked on my pitch with my wonderful mentor, Fiona. We came up with this. (The pitches for each contest are numbered for easy reference; naturally, the numbers weren’t actually tweeted.)

1. To rescue her boyfriend, a teenaged Soviet pilot must face off against both the Nazis and the perception that women can’t fight.

Since it was my first contest, I didn’t realize that you need multiple pitches. When I found out, I hastily wrote three more:

2. It’s Code Name Verity in Soviet Russia when teen pilot Valya disobeys orders to rescue her boyfriend from behind enemy lines.

3. 18yo pilot Valya must rescue her boyfriend from behind enemy lines–even if it means being branded a traitor by the Soviets.

4. Armed only with obsolete biplanes, a Soviet air regiment of teen girls faces the Nazis–and the idea that women can’t fight.

I labeled these three as “comp-focused,” “plot-focused,” and “concept-focused,” respectively. You’ll notice that, because I was in a hurry, these are all fairly similar.

In December, working off the information from September, I dropped the least successful pitch (plot-focused) and wrote eight more, some reworkings of the ones I already had, some new. These I broke into four categories (comp titles were not counted as their own category, but analyzed separately). Plot-focused pitches lay out the main conflict of the story:

1. To rescue her boyfriend, a teenaged Russian pilot must confront both the Nazis and the perception that women can’t fight.

2. It’s Code Name Verity in Soviet Russia when teen pilot Valya disobeys orders to rescue her boyfriend from behind enemy lines.

7. 18yo Valya disobeys orders to rescue the boy she loves in this adventure of the Russian pilots the Nazis called Night Witches.

11. When her boyfriend is trapped behind enemy lines, 18yo pilot Valya must risk both the Nazis and the gulag to rescue him.

Concept-focused pitches dispense with the details of the story and instead describe the setting and what makes it unique:

3. Armed only with obsolete biplanes, a Soviet bomber regiment of teen girls defies the Nazis–and the idea that women can’t fight.

4. One Soviet bomber regiment so terrifies the Nazis that they earn the name “Night Witches”–and the pilots are teenaged girls.

5. Teenaged girls in wooden biplanes combat both the Nazis and the idea that women can’t fight. Code Name Verity in Russia.

Character-focused pitches provide a little insight into the protagonist:

6. Soviet teen pilot Valya longs to fight the Nazis. When her boyfriend is trapped behind enemy lines, she gets her chance.

9. The Nazis call these Russian pilots “Night Witches,” but in her fragile biplane, 18yo Valya knows she’s all too human.

And finally, voice-focused pitches may cover any subject, but use a more casual, conversational tone:

8. Who says girls can’t fight? Not the Soviets. A teen aviatrix rescues a soldier from behind enemy lines in this WWII adventure.

10. Nazi fighter aces, blinding searchlights, flammable planes: Nothing teen girls can’t handle. Code Name Verity in the USSR.

For #PitchMAS, I took a similar approach; I dropped the least successful pitches (#1, #5, and #9) and created some new tweets, including several that focus on the romance aspect and the epistolary aspect. Since #PitchMAS wasn’t very successful, I won’t list the pitches here.

Preliminary Results

In September, I tweeted 14 times and received five favorites, distributed as follows. (For statistical purposes, mistaken favorites from people who are not agents or editors were not counted. They were instead classified as retweets.)

September

In December, I tweeted 25 times and received 17 favorites. The best performers were #7 and #10, each with four favorites. Here are the results divided up by pitch focus. The measurement, favorites per tweet, will be used throughout.

December

In #PitchMAS, I tweeted 25 times and received just two favorites, one from the popular #10 and one from a new tweet combining #5 and #7. #PitchMAS took place just eight days after December #PitMad and the day after the PitchMAS blog contest, in which I was also featured, which explains my lack of success; very little information can be derived from it.

These initial results are already surprising. The carefully-crafted pitch I created for September was outperformed by two of the others in September and received no favorites in December. The pitch I created by combining the two most successful pitches from September (#5) also received no favorites.

Voice-focused pitches were the only category that performed well across the board. However, voice isn’t an essential ingredient: The popular pitch #7 is one of the most concrete and straightforward.

Contrary to popular wisdom, character-focused pitches performed very poorly, netting only one request total. I suspect that one tweet simply isn’t enough space to establish anything interesting about a character. Conversely, concept-based pitches performed fairly well. Again, this goes against the common wisdom that the setting is not the interesting part and that pitches should focus on character and plot instead. (Voice is probably the factor that explains why concept-focused pitches performed better in September than December. The concept-focused pitches were rather plainly written.) And common wisdom states that stakes are an essential element, but the only tweet that mentioned the stakes outright (#11) did not receive any favorites.

Our first conclusion is clear: Common wisdom should go out the window.

The Time Factor

The most important outside factor is, thankfully, also the easiest to control: Time. In September, I tweeted manually. In December, I used TweetDeck to schedule my tweets. This was a major benefit. In September, there were large gaps during which I didn’t tweet at all and any agents who came online would have missed me. In December, I’m confident that the vast majority of agents who looked at the hashtags saw my tweets.

Here’s how the distribution of favorites and retweets broke down over time. My time zone, Pacific time, is used throughout.

September times

December times

PitchMas times

In both #PitMads, activity is highest in the morning, peaking around 7 AM. There’s a midday lull. Then, around 2 PM, there’s an uptick in favorites, but not in retweets. This correlates with New York agents checking the feed before they go home. I’ve never received a favorite after 3 PM, despite a sharp day-end uptick in retweets in September. Perhaps because of the confusion about start and end times, #PitchMAS is more of an amorphous blob with one favorite, as expected, in the early morning and another, unusually, in the middle of the day.

Retweets

Retweets are a complicating factor. As shown above, there’s a loose correlation between number of retweets and number of favorites. But is it because frequently-retweeted pitches get seen by more agents and thus favorited more, or simply because good tweets get more attention from both agents and friends? There isn’t enough data to say.

In December #PitMad, the most retweeted tweets also had the most favorites, but in both September #PitMad and #PitchMAS, some of the most-retweeted tweets received no favorites. So I wouldn’t bother tailoring a pitch to what garners the most retweets. Focus on the favorites.

Keyword Analysis

In addition to using different pitches, I varied the wording of individual pitches, and in December #PitMad, I received enough favorites to examine the success of various keywords. Naturally, these are specific to my manuscript, but they’ll give you an idea how word choice can affect a pitch’s success.

KeywordsIn gray, we see miscellaneous keywords. The most successful pitches of all, averaging well above one favorite per tweet, featured the word “adventure.” “Night Witches” also performed well. In keeping with the general success of plot- and concept-focused pitches, this shows that agents responded best to pitches that clearly indicated the sort of story they were in for.

Conversely, setting-specific keywords did not affect a pitch’s success. Agents were not just looking for books with Nazis in them. There’s hope for humanity after all. (However, all my pitches included enough information to determine that the setting was the Eastern Front. I wouldn’t attempt a pitch that left the setting unclear.)

I intentionally didn’t focus on my story’s romantic aspect, but some pitches mentioned a love interest and others did not; these also had no effect. It appears that agents did not find romance, or the lack thereof, to be an essential element.

In red, we see that “Soviet” significantly outperforms “Russian.” Soviet Russia is such an untapped resource in fiction that it was effective at catching agents’ attention.

Finally, in cyan, we see the strange effect of genre hashtags. The neglected historical fiction has no agreed-upon hashtag; it’s split between #hf and #hist, both of which it must share with historical fantasy. I used both hashtags, plus some tweets with neither. Bizarrely, tweets without hashtags vastly outperformed tweets with hashtags. However, looking at the individual tweets reveals two mitigating factors. First, the tweets without genre hashtags used the longest pitches, which were also some of the most popular. And second, crucially, tweets without hashtags were skewed towards the beginning of the day, when the most agents were active.

Still, it’s safe to conclude that using a genre hashtag didn’t improve a tweet’s chances. Again, this goes against the common wisdom that genre hashtags are the only way agents can filter the chaotic #PitMad feed. But note that all my pitches included the #YA hashtag.

The final factor I examined was the use of comp titles.

CompsHere, for once, the data is clear: Pitches featuring comp titles vastly outperformed those that didn’t in every contest.

Conclusions

Comparing my data to Lara’s data from #SFFPit, we see a striking amount of agreement. Her analysis of timing shows an initial peak within the first few hours and another peak in the afternoon, though her afternoon peak skews a little earlier than mine.

She also found that concept- and voice-focused pitches had the highest return rates, and she also had success with comp titles. Thus, we can conclude that these types of pitches are more successful in general, and that it’s not simply that I had a particularly good concept or a particularly sought-after comp title. She also found that character-based pitches weren’t very successful.

However, other factors are more difficult to generalize to other manuscripts. For example, “adventure” was probably successful because it’s such an unusual genre these days. I doubt you’d have similar success including words like “dystopian” or “epic fantasy.” Conversely, genre hashtags might work better for those categories, since agents are more likely to search for them.

In the end, the only way to know how you’ll do is to get in the trenches and try Twitter pitching yourself.

Leave a Reply