Wednesday, June 9, 2010

Analyzing hits: Identity in May 2010

One of the things I find interesting is the pattern of hits to my stories. I hope to discover, through statistical behaviors of readers, what is effective and what isn't. It's not exact science, but I think some reasonable conclusions can be made through some examination of data. Below is a table of hits to Identity for the month of May 2010. The total number of hits is 4603.

ActHitsHits Ratio
Prelude to In the Cold Rain1327N/A
Sounds of Night2180.72
The Cold Rain1720.99
Prelude to The Village1260.73
Valley of Magic1180.94
The Sorcerer's Den1130.96
Ritual at the Spring1040.92
Prelude to Journey to Jusenkyo1100.96
The Nerima Conclave1291.17
Night Life of the Dotonbori1431.11
The Amazon Charter1921.34
The Nine and the Three3261.70
The Last Right2230.66

The third column is calculated based on the ratio of hits from that row to the chapter of the row before (so the hits ratio for "Monday" compared to "Prelude to 'In the Cold Rain'" is 374/1327=0.28. I think this is a useful measure for a number of reasons: it shows how many people stop reading at a particular chapter, for example. In the case of Identity, having been updated three times in the month of May, I think you have to divide the story into roughly three regimes: the first being the beginning, where every new reader will pass. The second, the middle, where readers have already made a reasonable commitment to read the story. The third is the most recent set of chapters, where readers who've already read most of the story will check in for new installments. I think each regime has a different statistical behavior, in ways that the data indicate.

First, we have to consider that FFN doesn't have a table of contents page, and as such, people often go to the first chapter, reach for the drop down, and then go to the chapter they like. Thus, the first chapter has an inordinate amount of hits, and the hits ratio of the second chapter / first chapter is very small. Beyond that, however, I think we can see that the first few chapters of a long story suffer significant loss. Readers are discriminating, after all, and there's a lot of fanfiction out there and not a lot of reason to read any unknown story over any other. The first few chapters show losses of 20-25% of new readers from each chapter to the next. One thing that's interesting, however, is that "The Cold Rain" shows little loss from readers who already read "Doubt." I can speculate this shows "Doubt" to be a good cliffhanger, especially considering the loss between "The Cold Rain" and "Prelude to 'The Village,'" but again, this is just speculation.

Starting with "The Village," I think we can see that we've entered the second regime, the "middle" of the story. Losses here are only on the order of about 5%--I would suggest this means everyone who's read to this point is reasonably committed to finishing whatever's already done. Again, "Escape" has a good hits ratio, which may reflect more on the strength of "Ritual at the Spring" or perhaps indicates that "Escape" was itself very memorable and drew some return hits from people who wanted to reread it.

After that, however, I think we're squarely in the third regime, as new chapters posted (or even just recently posted in the last couple months) draw increasingly more attention, up to the final posting of May, which has fewer hits because it had only 3 days of visibility before we entered June. I'll be interested to see over the next few months how the numbers for "Journey to Jusenkyo" level off as those chapters enter the middle regime, as well as to see whether these readership losses change over time as the story becomes more saturated with the typical Ranma 1/2 readers.

No comments: