The Anatomy of an NSFW AI Character Card
A character card is a small stack of fields, and the order you read them in decides how fast you can tell whether a character is worth your evening. Take one entry from the trending list, the NSFW AI characters card showing 18.2M views beside 4.7K favourites: 4,700 divided by 18,200,000 gives 0.0258%, which is the first bar in the chart below.
Key takeaways
- Read a card in a fixed order: display name, creator signature, tag strip, view counter, favourite counter.
- A card at 18.2M views and 4.7K favourites turns views into favourites at 0.0258%, and the cards on the page sit in a band of roughly 0.015% to 0.029%.
- Tags such as Unfiltered, Submissive, Fictional, OC or Female describe how a character is framed, not how a conversation will run.
- The listing refreshes every Thursday, so a card you noted last week may be ranked differently today.
A fixed reading order beats browsing
Start at the display name, then move through the four remaining fields in order. A card gives you a name, a creator signature introduced by the word "by", a short strip of tags, and two counters printed next to each other. That is the whole entry. The Crushon trending page prints all five in a fixed layout, so everything you want to know about a character's popularity, authorship and framing sits in those five positions.
The creator signature is the field most readers skip, and it is the one that saves the most time. A signature ties the card to a body of work, so if you already liked one character from that creator, the odds of liking the next one go up. If you have never heard of the creator, the signature tells you the card is someone's original work rather than a generic listing. Following a signature through the character search page turns a single card into a shelf.
Tags come next, and they are descriptive rather than predictive. Unfiltered tells you the character can be used without the filtered layer applied. Submissive, Fictional, OC and Female describe the framing the creator chose. None of those tags promise a specific tone in your own conversation, because tone comes from what you type as much as from the card.
What the two counters measure
The view counter is a measure of attention and the favourite counter is a measure of intent. Views move whenever anyone opens the character, including people who bounce in a few seconds. Favourites move only when someone decides the character is worth keeping. That difference is why the ratio, not either number on its own, is the useful reading.
| Field | What it carries | How to read it |
|---|---|---|
| Display name | The character's identity, usually short and memorable | Use it as the search term if you want variants of the same idea |
| Creator signature | "by …" followed by the account that published the card | Treat it as a quality hint: the same creator usually keeps a consistent style |
| Tag strip | Unfiltered, Submissive, Fictional, OC, Female and similar labels | Read as framing and content type, never as a promise about your chat |
| View counter | Total opens, counted in millions (18.2M, 19.1M, 24.3M on the same page) | Broad attention only; inflated by quick visits that go nowhere |
| Favourite counter | Deliberate saves, counted in thousands (4.7K, 2.8K, 7.1K) | The better signal of repeat use; read it against the view count, not alone |
| Position in the list | A rank that is recalculated every Thursday | Compare cards on the same day; a rank from last week is stale |
Notice how the last row changes the whole exercise. Two counters read on different days are not comparable, because the list is refreshed weekly. If you want to judge movement, note the counters on a Thursday and again the following Thursday, and compare like with like.
Worked example: turning counters into a ratio
Take the card at 18.2M views with 4.7K favourites. Write the favourite count out in full, 4,700, and the view count out in full, 18,200,000. Divide the first by the second: 4,700 / 18,200,000 = 0.0002582. Multiply by 100 to express it as a percentage, and you get 0.0258%, the first bar in the chart.
Flip the same division around and the number becomes easier to feel: 18,200,000 / 4,700 = about 3,870, so roughly one viewer in every 3,870 chose to save that character. Run the same arithmetic on two other cards on the page and the spread widens. A card with 7.1K favourites against 24.3M views gives 7,100 / 24,300,000 = 0.0292%. A card with 2.8K against 19.1M gives 2,800 / 19,100,000 = 0.0147%. The three readings in the chart, 0.0258%, 0.0464% and 0.0155%, sit inside that same order of magnitude, which is the point: on a list this busy, none of these cards is being ignored, and the differences between them are small enough that a ratio difference should nudge your choice rather than make it.
Where a card stops being informative
A card cannot tell you whether the character's replies will match your taste, and no counter will fix that. What the counters can do is help you decide what to open first. The full list sits at https://crushon.ai/trends/nsfw_ai, and reading it as twenty entries with five fields each is faster than opening characters at random.
Two habits make the reading honest. First, check the counter pair together, because a high view count with a thin favourite count usually means a card that attracts curiosity and does not hold it. Second, re-check on a Thursday, since the ordering has moved by then and any note you wrote last week is describing a different list. If you want the longer explanations behind these mechanics, the AI chat blog keeps a running set of walkthroughs.
One more caution: a card is a description, not a script. A creator writes the framing, the tags and the opening premise, and the conversation itself is text you and the character build together. Reading the card well gets you to a better starting point faster; it does not pre-write the session for you.
Frequently asked questions
Is a higher view count a sign of a better character?
Not on its own. Views count every open, including brief ones, so a card can rank high on curiosity alone. Read the favourite counter against the view counter: two cards on the same page at 0.0155% and 0.0464% differ by three times in how often viewers save them.
Why do the counters on a card change between visits?
Views and favourites accumulate continuously, and the ordering itself is recalculated every Thursday, so the same character can appear higher, lower or not at all in the following week's list.
Which field should I check first if I am short of time?
The creator signature. A name you recognise from a card you already kept is the quickest filter available, and it costs one glance rather than a full read of the tag strip.
Do the tags apply to every conversation with that character?
No. Tags describe the framing the creator gave the character. The direction of an individual chat depends on what you bring to it, within the platform's content and community guidelines.
A twenty-second habit worth keeping
Read the name, the creator, the tags, then the two counters in that order, and do the division once for the card you are considering: favourites divided by views, multiplied by 100. Ten seconds of arithmetic tells you whether the crowd stuck around.
Then decide with the tag strip in hand, open the character, and judge the actual conversation rather than the listing. If the chat does not land, the ratio did its job by getting you there quickly, and the next card is one scroll away.
Data and sources: counters displayed on the trending character cards and the 20-character weekly list, read from the platform on 22 Sept 2026[1]. Ratios are the author's own arithmetic on those displayed counters. This article is written for adults 18+.