Genre Signals from 241 Competitor Materials: Why Romance, Horror, and Fantasy Are Most Common

Sample disclosure: The 241 items are publicly available materials recorded during competitor research on 2026-09-22. They are not a random sample of the entire market, nor do they represent 241 distinct works. Both automated and manual labels may contain errors, so this article discusses only directional signals.

Romance, horror, and fantasy appear repeatedly in the public sample, not only because they are popular, but also because each has states that readily lend themselves to interaction: romance has relationships, horror has risks, and fantasy has world rules. Whether a genre can produce a good work still depends on whether creators turn those states into choices with information and costs.

Whether to confess something or keep a date can quickly change a character's attitude, and users can easily understand the feedback. The problem is that many works reduce relationships to a single affection score, making “always choose the pleasing answer” the optimal strategy. Better design distinguishes intimacy, trust, respect, and boundaries.

Fleeing, hiding, investigating, and rescuing others all carry clear risks, and QTEs and timed choices are easy to justify. But horror is also prone to relying on deaths without warning, repeated replays, and jump scares. Interaction should let players learn the rules of a threat, rather than guess what the author is thinking.

Factions, abilities, objects, and rules can support states that persist across multiple chapters. The trade-off is higher costs for worldbuilding and assets, while AI generation can easily introduce inconsistencies in clothing, settings, and creatures. Starting with one understandable rule is more reliable than building a vast world first.

Retain the source URL, collection time, page type, primary and secondary genres, interaction mechanics, and annotation confidence for each item; have two people review a random subset and report their agreement rate; deduplicate multiple pages belonging to the same work. Without these steps, genre proportions are merely a pretty chart.

“Few samples” may indicate an opportunity, but it may also mean weak demand, difficult production, or poor platform fit. An underserved genre must pass three checks at once: whether users have a clear emotional need, whether the genre offers natural choices, and whether the team can produce assets consistently.

For example, workplace investigations, family caregiving, disaster collaboration, and lighthearted comedic misunderstandings may be less common than popular genres, but all can offer states involving information, responsibility, and relationships. Test them with a three-minute prototype first, rather than immediately greenlighting a long-form work because there are few competitors.

The value of these 241 materials is not in predicting the next hit, but in forcing teams to turn “I think this is popular” into genre hypotheses with sources, consistent definitions, and reproducible checks.

A single work may have a cover page, episode pages, character cards, and pages in multiple languages, while the same material may also be posted repeatedly across different channels. Before compiling statistics, assign each record a work ID, page type, and source platform, then decide whether to deduplicate based on the research question. Genre labels should allow primary and secondary levels; a horror work should not be relabeled entirely as romance just because it contains a romantic relationship.

Have two annotators independently assess a subset of the sample, and record their agreement rate and reasons for disagreement. Automated classification can improve speed, but low-confidence cases, mixed genres, and short descriptions still require manual review. When reporting proportions, also provide the denominator, missing values, and snapshot date, without presenting a convenience sample as market share for the entire market.

Actionable states in romance may include trust, boundaries, commitment, and information asymmetry; in horror, threat locations, time, tools, and understanding of rules; in fantasy, factions, abilities, costs, and world rules. A topic-selection meeting can determine whether a genre truly suits interaction only after clearly defining player verbs, visible states, and the costs of choices.

For example, a romance prototype need not start with multiple romanceable characters; it can test a single choice between “protecting a secret or being honest about a risk.” A horror prototype can use one learnable monster rule to verify whether failure is fair. A fantasy prototype can introduce just one ability and its cost, then observe whether players can predict the consequences. A three-minute prototype gets closer to product evidence than continuing to collect more popular genre labels.

Less frequent genres require simultaneous validation of demand in searches or communities, natural interaction verbs, the ability to deliver assets, and fit with platform distribution. Few competitors may mean users are underserved, but it may also reflect filming difficulties, content review risks, or weak replay value. Write down explanations both for and against each candidate, then validate them through interviews with a small sample and behavior in prototypes.

The next round of research should retain the same classification rules and add a time dimension, comparing genre releases, completions, and ongoing updates rather than looking only at cumulative inventory. Upgrade a directional signal into a content strategy only when both data from multiple periods and user behavior support it.

When publishing charts, show both absolute counts and proportions, and list unknown and mixed genres separately. Avoid fine-grained rankings for categories with small samples, and check whether source coverage or annotation rules have changed before interpreting changes across periods. Charts should help readers understand limitations, rather than erase them.

• 51PAPAYA Series
• CastLoop
• 66RPG