You find out where your readers come from by looking at the referring page of the first view in each reading visit, which is what a story acquisition panel records and sorts into a small number of channel buckets such as Reddit, search, Instagram, an on-platform browse page, or a direct link with no referrer at all. On IlorisNovel that panel sits in your story's analytics under the heading How readers found this story, it counts the opening page of each tracked reading visit rather than every chapter you serve, and it excludes your own reads and known crawlers, so what you are looking at is the closest honest answer to the question of which channel is actually doing the work.
Where do web fiction readers actually come from?
They come from a handful of channels in proportions that almost never match what the author assumes, which is why the question is worth measuring rather than reasoning about. In my experience talking to authors, the guess is nearly always some version of "my subreddit post did most of it", and the numbers just as often say that on-platform browsing quietly out-produced the post by a wide margin, or that a link somebody pasted in a Discord six weeks ago is still sending people every week. The reason the guess goes wrong is structural rather than careless: you are vividly aware of the promotion you did, because you did it, and completely blind to the reader who found you through a genre listing at two in the morning. Measurement fixes the asymmetry by giving equal weight to the channels you never think about.
The other reason to care is that promotion time is the scarcest thing a web fiction author has. If you are writing four chapters a week, whatever hours remain are the only hours you get for audience-building, and spending them on the channel that produces the least is the most common quiet mistake I see. My feeling is that most authors would double the return on their promotion effort simply by looking at the numbers once a month and reallocating, without doing anything cleverer than that. This is the same logic that makes an email list worth building early, since both are about knowing and controlling the path between you and the reader instead of hoping.
How the "How readers found this story" panel works
The panel records the referring page of the first view in each reading visit and files it into exactly one channel bucket. When somebody opens your story to read, the server looks at where the request came from, classifies it, and adds one to a counter for that channel on that day. The classification is deliberately coarse. A referrer from any reddit.com address becomes Reddit, an instagram.com referrer becomes Instagram, tiktok.com becomes TikTok, x.com and twitter.com and t.co all collapse into X / Twitter, facebook.com becomes Facebook, and Google, Bing, or DuckDuckGo become Search engine. Anywhere else external becomes External site. If the referrer is a page on IlorisNovel itself, it resolves to the home page, the browse page, or a generic other-Iloris-page bucket depending on which page sent the reader. And if no referrer arrives at all, the visit is filed as Direct link.
Three details change how you should read the result. The first is that the counter fires on the arrival, not on every chapter, so a reader who opens your story from Reddit and then reads eight chapters in a row registers as one Reddit arrival rather than eight; moving from chapter to chapter inside the reader is recognized as a continuation of the same visit and deliberately not counted again. The second is that your own reads of your own story are excluded, which matters more than it sounds, because authors reread their own chapters constantly and would otherwise dominate their own numbers. The third is that known crawlers and link-preview bots are filtered out by user agent before anything is recorded, so the count is not inflated every time somebody pastes your link into a chat app that fetches a preview card. The panel sits in your story's analytics with a seven day and a thirty day toggle, and each channel is shown as a raw visit count with a rounded percentage of the total.
What each bucket actually means
Every bucket is a claim about the referring page, not about the reader, and reading them correctly means knowing what each claim leaves out. Reddit, Instagram, TikTok, Facebook, and X mean the reader's browser reported one of those domains as the page it came from, which reliably catches desktop and web traffic and unreliably catches traffic from inside a mobile app, since apps frequently strip referrers on the way out. Search engine means Google, Bing, or DuckDuckGo sent them, and because search engines pass the referring domain but not the query, you learn that search worked without learning what they searched. External site is the catch-all for any other domain, which in practice tends to mean a forum, a Discord web link, an aggregator, or a blog. Browse page and Iloris home page mean the reader was already on the platform and found you through discovery rather than through anything you promoted, which is the number most authors underestimate.
Direct link is the bucket that misleads people, so it deserves its own sentence: it does not mean the reader typed your URL from memory, it means no referring page was reported at all. That covers links opened from private messages, links tapped inside apps that do not pass a referrer, links pasted by hand, and a fair amount of ordinary mobile traffic, which is why direct is usually the largest or second-largest bucket for a small story and why I think it is better read as a proxy for private sharing than as a mystery. If you want a fuller treatment of that specific problem, I wrote about why so much reader traffic shows up as a direct link separately, including the one tagging trick that recovers some of it.
The last thing worth internalizing is that the panel counts visits rather than distinct humans. One devoted reader arriving from Reddit on five different days contributes five to the Reddit bucket, and there is no deduplication across days by design, because the underlying store is a daily counter per channel rather than a log of individual visitors. That makes the percentages an honest answer to "which channel produced reading sessions this month" and a wrong answer to "how many different people did Reddit send me", and confusing the two will make a small loyal channel look bigger than it is.
How to actually act on what the panel tells you
The useful move is to compare a thirty day window against the previous one and change exactly one thing, rather than to stare at a single snapshot and try to redesign your whole promotion strategy. If Browse page and Iloris home page dominate, your discovery on the platform is working and your cover, title, and blurb are doing the heavy lifting, which means the work most worth doing is probably on-platform presentation rather than more posting elsewhere. If Reddit or another social bucket dominates but your chapter-to-chapter retention is weak, you have an audience arriving into a story that is not holding them, and no additional promotion will fix that; the fix lives in the opening chapters. If everything is flat and direct is most of it, you are being shared privately and the sensible next step is to make that sharing easier and to capture those readers somewhere you control.
I think the discipline that matters most here is refusing to over-read small numbers. A story doing forty arrivals a month will show wild percentage swings from a single good day, and treating that as a trend leads authors to abandon channels that were fine. Wait for the thirty day window, look for a change large enough that you would have noticed it without the chart, and when you do run a promotion, run it in isolation so you can actually attribute the bump. This connects directly to the economics: the reason to know your channel mix at all is that it is the front of the funnel that ends in how web fiction authors get paid, and every honest number about where readers come from is a number about where your future income comes from too. Running a platform, the pattern I see most is that authors who look at this once a month make better decisions than authors who look at it obsessively or never, and the gap between those two groups is larger than the gap between any two promotion tactics.
Web fiction has an unusual relationship with analytics, because the major platforms have historically given authors very little, and as of this writing most give you a view count and a follower count and leave you to infer the rest. That absence trained a whole generation of serial authors to run on instinct, which works right up until the moment you have to choose between two ways of spending a limited Saturday. The panel is not clever and does not need to be. It answers one question, coarsely, honestly, and in a form you can check in fifteen seconds, and my advice is to treat it as exactly that: a monthly sanity check on a story you are otherwise too close to see clearly.
Common questions about web fiction traffic sources
Where do web fiction readers come from?
For most small and mid-sized web fiction authors, readers arrive from a mix of on-platform discovery, a handful of social posts, search, and direct links pasted into chats and forums. The proportions vary enormously by author and are almost never what the author guesses, which is the whole reason to measure rather than assume. An acquisition panel answers the question for your specific story by recording the referring source of the first page of each reading visit and grouping those into channel buckets you can compare week to week.
How does the How readers found this story panel work?
When someone opens your story to read, the server looks at the referring page for that visit and sorts it into one bucket, such as Reddit, Search engine, Instagram, Browse page, or Direct link. It counts the opening page of a visit rather than every chapter, so a reader who works through six chapters in one sitting registers as one arrival rather than six. Your own reads are excluded, and known crawlers and link-preview bots are filtered out before anything is counted.
Why does so much of my traffic show up as a direct link?
Because a direct link simply means no referring page was sent with the request, which happens far more often than most authors expect. Mobile apps, private messages, in-app browsers, and readers who paste a URL by hand all commonly arrive with no referrer attached, so real word-of-mouth traffic tends to pile up in the direct bucket. It is usually a sign that people are sharing your story privately rather than a sign that nothing is working.
Does the acquisition panel count views or people?
It counts visits, not unique people. A reader who comes back from Reddit on three different days adds three arrivals to the Reddit bucket, because the panel deliberately stores a daily counter per channel instead of tracking individuals across sessions. That makes the percentages a good measure of which channel produces reading sessions and a poor measure of how many distinct humans a channel sent, and the distinction matters when you are deciding where to spend your promotion time.
How often should a web fiction author check their traffic sources?
Roughly once a month for the shape of things, and once right after anything unusual, such as a promotion, a cross-post, or a chapter that got picked up somewhere. Checking daily invites you to read noise as signal, since small stories produce small numbers and a single busy afternoon can swing a percentage. The useful comparison is a thirty day window against the previous one, because that is long enough for a pattern to be real and short enough to still be actionable.
by Jacob Tam · August 6, 2026
I run IlorisNovel, a platform where web fiction writers keep 93% of what they earn and can see, in plain language, which channels are actually bringing readers to their stories.