Key Takeaways
Text patterns identified about 80% of sampled New York Times bestsellers.
Success leaves traces. Archer and Jockers argue that blockbuster fiction is not simply an inexplicable stroke of luck. Their computer model examined nearly 5,000 novels, including just over 500 New York Times bestsellers, and compared thousands of textual features. Across three classification methods, it correctly separated bestsellers from other books about 80% of the time. One method reached 90% in their tests. The model did not know an author's name or reputation, so a manuscript's language and structure mattered to its predictions.
A bounded prediction. This result applies to the books and bestseller definition in their research, not every market or future hit. A bestselling book was one that appeared on the New York Times list, not necessarily the book with the most sales. The model also made mistakes, including missing the success of The Help.
The compelling shift is from asking whether a hit can be foretold with certainty to asking whether repeated reader preferences leave measurable traces. That reframes publishing from pure intuition toward testable judgment. Yet an 80% result is meaningful only relative to the chosen sample, labels, and time period. The model identifies associations that helped classify past books; it does not prove those features caused sales. This distinction matters for writers: the findings can sharpen revision and editorial discussion, but they cannot turn creative risk into a guaranteed investment.
Give one or two themes a third of the novel's oxygen.
Focus creates coherence. Across the novels studied, bestsellers tended to concentrate roughly 30% of their pages on one or two topics. Less successful books needed at least three topics to occupy a similar share. To reach 40%, a bestseller used about four topics on average, while a non-bestseller used six. The authors see this as evidence that a focused story gives readers a strong through-line instead of asking them to track an encyclopedia of subplots.
Pair focus with friction. John Grisham devotes about a third of his work to law, while Danielle Steel gives a similar share to domestic life. Their other recurring topics create pressure around those familiar centers: legal threats endanger ordinary lives, and medical emergencies disturb domestic security. The advice is not to copy their subjects, but to make a central concern collide with a meaningful threat.
This finding offers a useful design principle: narrative focus is not the same as simplicity. A novel can contain many details while asking readers to care most about a small number of organizing concerns. The strongest examples use secondary topics to intensify the primary one, so variety feels like escalation rather than distraction. Still, topic counts are an imperfect proxy for artistic coherence. A deliberately sprawling novel may make multiplicity its subject. The practical question is whether each strand deepens the reader's central experience, not whether the manuscript obeys a numerical quota.
Make human closeness the engine, not a decorative subplot.
Connection travels across genres. The book's most useful predictive topic is what the authors call human closeness: scenes of shared intimacy, chemistry, and emotional connection between people. It is not limited to romance. Even Grisham's lawyers need ordinary moments of companionship amid legal danger. The theme appears in very different kinds of fiction because relationships give readers a personal stake in events.
Fifty Shades in context. The model gave Fifty Shades of Grey a high score despite its explicit sex, which was generally less common in bestsellers. Its largest topic was human closeness, at about 21% of the novel; intimate conversation accounted for another 13%. Sexual topics appeared too, but were not the largest elements. The authors argue that emotional negotiation between the characters, not simply erotic content, helps explain its wider appeal.
Human closeness is a persuasive bridge between plot mechanics and reader involvement. A threat becomes more consequential when it endangers a bond, and a quiet exchange can give action a human scale. This also helps explain why intimacy can matter in thrillers, literary fiction, and romance without making them the same genre. The concept should not be reduced to a requirement for warm or healthy relationships. The dysfunctional marriage in Gone Girl shows that alienation itself can make closeness central, because readers feel the damage to a bond that should provide safety.
Build a regular emotional pulse, not merely a dramatic plot.
Readers feel the pacing. The authors used sentiment analysis, a method that tracks positive and negative emotional language, to chart how a novel's emotional terrain changes from scene to scene. They found no single overall plot shape that guaranteed bestseller status. What distinguished some major hits was a steady rhythm of emotional turns: conflict, relief, renewed conflict, and another release.
Two unlikely cousins. Fifty Shades of Grey and The Da Vinci Code differ in genre and subject, but both show a notably regular sequence of emotional peaks and valleys. In the first, relationship tension and sexual encounters drive the beat; in the second, danger alternates with brief relief. The opening matters too. The authors cite the familiar editorial advice that a book needs to hook readers early, sometimes within the first 40 pages. Strong pacing is not constant intensity; it is variation with momentum.
The emotional-curve idea makes an elusive reading experience visible without claiming that a graph can replace close reading. It encourages writers to ask where attention rises, where readers can breathe, and whether turning points arrive with enough force. The broader lesson resembles music: repetition creates expectation, while variation keeps that expectation alive. But sentiment dictionaries register emotional words, not the full complexity of irony, silence, or cultural context. A flat-looking passage may be artistically charged, and a series of plotted peaks can still feel mechanical if character and meaning are absent.
Let crisp, conversational style make familiar stories feel newly alive.
Small words shape voice. Stylometrics, the computational study of writing style, can detect patterns in common words, punctuation, sentence length, and parts of speech. In the authors' tests, just 491 frequent words and punctuation features classified bestseller status about 70% of the time. The model associated bestsellers with more contractions and questions, fewer exclamation points, and less use of words such as very. These patterns suggest prose that sounds direct and contemporary rather than heavily decorated.
Training leaves fingerprints. The researchers found that style signals often reflected writers' backgrounds. Journalism and advertising encouraged accessible, concise language, while traditional literary training was associated with different stylistic habits. J. K. Rowling's attempt to publish as Robert Galbraith did not erase her linguistic fingerprint. None of this means that one style is inherently superior. It shows that familiar plots can feel fresh when delivered through a distinctive, confident voice.
The emphasis on ordinary language counters the idea that literary distinction must announce itself through ornate sentences. Voice often emerges from repeated choices so basic that writers barely notice them. That makes computational style analysis potentially useful as a mirror: it can reveal habits that are difficult to perceive while drafting. The risk is treating correlations such as contractions or punctuation as recipes. Their effects depend on narrator, genre, and audience. A clipped voice can feel alive in one book and thin in another. The craft lies in making the linguistic choices fit the fictional consciousness.
Give protagonists needs they act on, not wishes they wait with.
Agency moves the story. The authors studied which verbs cluster around characters, using names and pronouns to connect characters with their actions. Their model classified bestseller status from character verbs with 72% accuracy. Protagonists in bestsellers were more likely to need and want things, then act on those desires. These characters also more often think, ask, reach, and decide. Characters in less successful books more often wish, suppose, or hesitate, which can leave the plot feeling passive.
Action need not mean heroism. Agency can be morally troubling. Amy in Gone Girl is compelling partly because she makes deliberate plans; Lisbeth Salander's expertise and choices drive The Girl with the Dragon Tattoo. The study also found traditional gendered patterns in actions, such as men driving and women talking more often. The authors identify those patterns, but their presence is not a reason to reproduce them unquestioningly.
A protagonist's desire gives scenes direction: readers can measure each action against what the character wants and what blocks it. This is a powerful way to diagnose a manuscript that feels inert. Agency, however, should not be confused with likability or constant activity. A character may exercise agency through refusal, concealment, or a costly decision, while deliberate inaction can itself be meaningful. The gender patterns described here also show that commercial fiction can repeat cultural expectations even when it gives its characters force. Writers can use that pattern deliberately, or make its limits part of the story.
Make the domestic noir heroine both the threat and investigator.
A new kind of girl. The authors use domestic noir for thrillers that bring danger into home, marriage, and family, spaces traditionally portrayed as safe or private. In books such as Gone Girl and The Girl on the Train, women are not merely victims waiting for rescue. They investigate, manipulate, expose, or commit violence. Their agency unsettles the domestic order even as it helps reveal what has gone wrong inside it.
Disruption creates the hook. The characters in these novels are outsiders to idealized roles such as good wife, dependable mother, or innocent girl. Their contradiction is part of their appeal: a heroine can be both the apparent problem and the person who uncovers the problem. The books often resist the reassuring closure of a conventional romance. Their unresolved endings leave readers with the continuing question of where such a woman belongs.
Domestic noir turns a familiar setting into a pressure test for social roles. A home is not simply a backdrop when the story asks who has power inside it, whose account is believed, and what violence is concealed by respectability. The heroine's moral ambiguity also makes identification more complicated than admiration: readers can follow her agency without endorsing every act. These novels are specific to a moment and market, though, and should not be treated as proof that audiences universally prefer darker women. Their importance lies in the tensions they make visible, not in a checklist of traits to imitate.
Read The Circle's 100% score as pattern recognition, not a formula.
One manuscript combined the signals. The model's top candidate was Dave Eggers's The Circle, a novel about a powerful technology company and the erosion of privacy. Its report showed about 21% modern technology, 4% workplace, and 3% human closeness, alongside a familiar emotional structure, bestseller-like style, and a highly active female protagonist. The model assigned it a 100% likelihood score, and the book did appear on bestseller lists.
The score was not a recipe. The authors had not read the novel before the model selected it, and its result showed how multiple features can reinforce one another. They explicitly reject the idea that the findings offer a guaranteed way to write a hit. A writer cannot simply add the right theme, punctuation, and character verbs. The report is best understood as a diagnostic portrait of patterns associated with success in the studied books, not a substitute for imagination or editorial judgment.
The Circle is a memorable case because the model's successful prediction sits inside a novel skeptical of technological measurement. That irony makes the result vivid, but it does not prove that the algorithm understands the book's warning or its artistic value. A high score can summarize how a manuscript resembles past successes while leaving questions of interpretation, ethics, and taste open. The strongest use of such a tool is collaborative: it can point to patterns for human readers to inspect, challenge, or revise. Its apparent precision should not disguise how much judgment went into defining the categories.
Use computers to inspect fiction, not to replace human authors.
Analysis is not authorship. The book distinguishes a machine that extracts patterns from a machine that creates a novel. Algorithms can classify books, track emotional language, and assemble sentences from existing text. The authors describe computer-generated fiction as often repetitive or dependent on borrowed material, with characters who lack distinct inner lives. The fact that a computer can imitate some textual patterns does not show that it has lived experience or understands what those patterns mean.
Human judgment remains central. The researchers present computation as a tool for seeing what readers and critics might miss at scale, not as a replacement for creative work. Their own interpretation of the findings depends on reading, publishing experience, and discussion of specific books. They conclude that writers still supply the human knowledge and imaginative choices that give fiction its emotional force.
The distinction between pattern recognition and lived understanding is essential to a sober account of creative technology. A system can detect recurring structures without knowing why a particular loss, betrayal, or moment of intimacy matters to a person. At the same time, the boundary is not a reason to dismiss computational tools. They can reveal habits, test assumptions, and widen the range of works critics take seriously. The productive question is not whether a machine is secretly human, but which parts of literary work can be measured, which resist measurement, and how the two kinds of insight can inform each other.
Analysis
The book's most important contribution is not a formula for hits but a change in what counts as evidence about popular fiction. It treats sales not as a verdict on literary worth, but as a cultural signal that can be studied alongside style, theme, character, and reader response. This makes commercial fiction available for serious analysis without requiring critics to pretend that sales equal quality. The method also changes the scale of literary attention: patterns across thousands of texts can become visible where individual close reading would struggle.
The central methodological caution is that the model learns to predict a historically defined category. Its target is appearance on the New York Times list, not universal popularity, lasting influence, critical merit, or a reader's private attachment. The corpus is also a designed sample, and the book acknowledges that marketing, adaptation, timing, reputation, and unusual cultural events affect outcomes. Accordingly, the reported accuracy demonstrates useful classification within the study, not a universal law of fiction. A pattern can be predictive without being causal. Human closeness may accompany successful books, but the analysis does not establish that adding more intimate scenes will make a manuscript sell.
The model's strongest cultural implication may be its challenge to literary hierarchy. If commercial and prize-winning novels share measurable craft, then dismissing popular fiction as mere formula hides the work that makes millions of readers care. Yet the account also raises a question about feedback loops. If publishers used such models to select manuscripts, the features learned from yesterday's bestsellers could narrow tomorrow's range, rewarding familiarity while claiming neutrality. The authors imagine the model widening access, since it can assess an unknown writer without knowing their name. That possibility is real, but only if human editors use prediction as one source of evidence rather than as a gatekeeping score.
The best way to read the findings is as an invitation to notice. Ask what a character wants, what emotional pattern keeps a story moving, and how its themes reinforce or complicate one another. The algorithm can point to regularities. Readers still decide what those regularities mean.
Review Summary
Reviews of The Bestseller Code are mixed. Many find the premise fascinating - using algorithms to analyze bestselling novels - but some criticize the execution as repetitive or lacking depth. Readers appreciate insights into what makes a bestseller, like strong characters and accessible language, but some wanted more detailed data. The book is seen as potentially useful for aspiring authors, though its focus on adult fiction is noted as a limitation. Overall, it's viewed as an intriguing look at the science behind bestsellers, even if not revolutionary.
People Also Read
Glossary
bestseller-ometer
Novel success classifierThe authors' name for a computer model that analyzes novels and estimates whether they are likely to appear on the New York Times bestseller list. It combines textual features such as themes, style, emotional patterns, and character actions. Its reported accuracy applies to the research collection and classification methods described in the book.
topic modeling
Textual theme discoveryA computational method that identifies recurring groups of words in a collection of texts and estimates how much each group appears in each book. The method uses the surrounding words to infer a topic, so a word such as bank can be associated with finance in one context and a river in another. The researchers used topic proportions to compare novels.
Human Closeness
Intimate human connectionThe authors' label for a recurring topic involving people communicating in moments of emotional connection, shared intimacy, chemistry, or bonding. It is not restricted to romantic love and can appear in different genres. In their comparisons, this topic was especially useful in distinguishing bestsellers from other novels.
sentiment analysis
Emotional language trackingA natural language processing method that identifies positive and negative emotional language in text. The authors used sentiment scores across a novel to plot changes in emotional tone, producing curves that approximate the story's highs, lows, and pacing. These curves represent patterns in language, not a complete or definitive measurement of a reader's feelings.
stylometrics
Computational style analysisThe quantitative study of writing style through features such as word frequencies, punctuation, sentence length, and grammatical patterns. Stylometric methods can compare writers or classify texts, sometimes identifying a writer's recurring linguistic habits even under a pseudonym. In this book, the authors use stylistic features to investigate patterns associated with bestseller status.
linguistic DNA
Recurring authorial language habitsThe authors' metaphor for relatively stable, often unconscious patterns in a writer's language, including choices of common words, punctuation, and grammatical structures. These habits can persist across different subjects or pen names and can be detected through stylometric analysis. The phrase is an analogy, not a claim that language and biological genes work identically.
domestic noir
Thrillers inside domestic lifeA type of thriller that brings danger, suspicion, or crime into settings associated with private life, especially homes, marriages, and families. In the authors' account, its women protagonists disrupt idealized roles and may act as investigators, victims, or agents of violence. The genre makes domestic relationships part of the danger rather than a refuge from it.
FAQ
What's "The Bestseller Code" about?
- Exploration of Bestsellers: "The Bestseller Code" by Jodie Archer and Matthew L. Jockers explores the anatomy of bestselling novels using data analysis and algorithms.
- Text Mining and Algorithms: The book delves into how text mining and machine learning can predict which novels will become bestsellers by analyzing thousands of features in thousands of books.
- Patterns in Bestsellers: It identifies patterns and features common to bestselling books, challenging the notion that bestseller success is purely random or based on marketing.
- Cultural Insights: The book also provides insights into cultural trends and reader preferences that influence the success of novels.
Why should I read "The Bestseller Code"?
- Unique Perspective: It offers a unique perspective on the publishing industry by combining literary analysis with data science.
- Understanding Success: Readers gain an understanding of what makes a novel successful beyond traditional literary critique.
- Practical Insights: Aspiring authors and publishers can find practical insights into crafting and selecting manuscripts with bestseller potential.
- Cultural Reflection: The book reflects on contemporary culture and reading habits, making it relevant for anyone interested in literature and media.
What are the key takeaways of "The Bestseller Code"?
- Not Random: Bestsellers are not random; they share common features that can be identified and analyzed.
- Key Features: Successful novels often have a balance of themes, plot structures, and styles that resonate with a wide audience.
- Human Closeness Theme: The theme of human closeness is a significant predictor of a novel's success.
- Algorithmic Prediction: Algorithms can predict bestseller potential with a high degree of accuracy, challenging traditional publishing wisdom.
How does "The Bestseller Code" define a bestseller?
- Sales and Longevity: A bestseller is defined by its sales figures and its ability to remain on bestseller lists for extended periods.
- Reader Engagement: It engages a broad audience, often transcending typical genre boundaries.
- Cultural Impact: Bestsellers often reflect or influence cultural trends and conversations.
- Algorithmic Criteria: The book uses specific algorithmic criteria to identify features common to bestsellers.
What methods do Archer and Jockers use in "The Bestseller Code"?
- Text Mining: They use text mining to analyze the language, themes, and structures of thousands of novels.
- Machine Learning: Machine learning algorithms are employed to identify patterns and predict bestseller potential.
- Feature Extraction: The authors extract thousands of features from novels, including word usage, plot structure, and thematic elements.
- Cross-Validation: They use cross-validation techniques to test the accuracy of their predictions.
What role does "human closeness" play in bestsellers according to "The Bestseller Code"?
- Central Theme: Human closeness is identified as a central theme in many bestsellers, contributing significantly to their success.
- Emotional Connection: This theme often involves emotional connections between characters, which resonate with readers.
- Predictive Power: The presence of human closeness in a novel is a strong predictor of its likelihood to become a bestseller.
- Subversion in Plot: Some bestsellers subvert this theme, using it to create tension and conflict, which also appeals to readers.
How do Archer and Jockers address the concept of plot in "The Bestseller Code"?
- Plot Shapes: They identify seven fundamental plot shapes that are common in bestselling novels.
- Emotional Curves: The book discusses how emotional highs and lows in a plot can engage readers and drive a novel's success.
- Three-Act Structure: A clear three-act structure is often present in bestsellers, providing a satisfying narrative arc.
- Regular Rhythm: Successful plots often have a regular rhythm, with well-timed emotional beats that keep readers engaged.
What insights does "The Bestseller Code" provide about writing style?
- Everyday Language: Bestsellers often use everyday language that is accessible and relatable to a wide audience.
- Contractions and Simplicity: The use of contractions and simple sentence structures is common in bestselling novels.
- Gendered Style: The book explores how certain stylistic elements are more common in novels by female authors, which often align with bestseller characteristics.
- Journalistic Influence: Many successful debut authors have backgrounds in journalism, influencing their concise and engaging writing style.
How does "The Bestseller Code" explain the success of "girl" novels?
- Character Agency: The book highlights the strong agency of female characters in "girl" novels, which drives their success.
- Dark Feminine: These novels often feature a dark or complex feminine character, challenging traditional roles and expectations.
- Plot and Theme: The plots of these novels often subvert traditional themes, creating tension and intrigue.
- Cultural Relevance: The success of "girl" novels reflects contemporary cultural conversations about gender and identity.
What are some of the best quotes from "The Bestseller Code" and what do they mean?
- "Bestsellers are not random": This quote encapsulates the book's central thesis that bestsellers share identifiable features and are not purely the result of chance.
- "Human closeness is a significant predictor": This highlights the importance of emotional connections in novels, which resonate with readers and contribute to a book's success.
- "Algorithms can predict bestseller potential": This emphasizes the role of data science in understanding and predicting literary success, challenging traditional publishing practices.
- "The market is not as unknowable as others suggest": This suggests that with the right tools and analysis, the success of novels can be better understood and anticipated.
How can aspiring authors benefit from "The Bestseller Code"?
- Crafting Manuscripts: Authors can use insights from the book to craft manuscripts with elements that align with bestseller characteristics.
- Understanding Trends: The book helps authors understand current trends and reader preferences, informing their writing choices.
- Improving Style: By analyzing successful writing styles, authors can refine their own style to appeal to a broader audience.
- Predicting Success: Authors can use the book's findings to assess the potential success of their work before submission.
What impact does "The Bestseller Code" have on the publishing industry?
- Data-Driven Decisions: The book encourages publishers to use data analysis in their decision-making processes, potentially reducing the risk of publishing flops.
- Challenging Traditions: It challenges traditional notions of literary success, suggesting that data can complement editorial judgment.
- Identifying Trends: Publishers can use the book's insights to identify emerging trends and capitalize on them.
- Broadening Access: The book suggests that data-driven approaches can help discover new authors and diversify the publishing landscape.
Download PDF
Download EPUB
.epub digital book format is ideal for reading ebooks on phones, tablets, and e-readers.