Friday, August 23, 2013

Glyph Posts and Glyph Sales Compared

I've been talking the last few months about recording glyph posting, so I could compare it to the number of glyphs sold. While I've been recording the data for a long time, I have only just begun to analyse it.

In order to give everyone a sneak peak, I've produced a the graph above for just a small window of just over five days. The blue line represents glyphs posted and is divided by 10, and the red line is glyphs sold. The graph is divided into hours. So the first blue spike is me posting 738 glyphs between noon and quarter past. The first few data points for sales are 2 glyphs sold between 1pm and 2pm, 2 glyphs sold between 2pm and 3pm, nothing sold between 3pm and 4pm, 7 glyphs sold between 4pm and 5pm and so on.

Although this is a very small sample size, it is worth noting just how sharply glyph sales fall off 12 hours after posting, leading me to conclude that posting twice per day wouldn't be a waste of my time (or at least the second posting won't suffer from diminishing returns).

While only a sneak peak, it gives me the confidence to say that a good post yields sales of between 5 and 10% of glyphs posted (my posting schedule is explained bellow), and a nice visual to explain to people that only dabble in the market what to expect in terms of unsold glyph auctions.

My next steps for this analysis will include:
  • Improved graphing over a longer period of time. I will include a separate scale for glyphs posted to glyphs sold, rather than just divide the number posted by something arbitrary.
  • A metric for the average number of glyphs sold for each hour after posting, and the average yield in terms of both sales and gold value.
  • Taking suggestions from readers about what else would be useful and/or awesome.

Naturally all this data is limited by my particular server, posting schedule and competition. My main limitation with this data is that I post on two different characters (probably soon to be three in patch 5.4) and it is potentially beyond the scope of the exercise to track who sells what.* For example, the large spikes are where both characters post a full set of glyphs in quick succession, and the smaller spikes are where they post further apart.

The major influence on the yield of sales is that I post two of each glyph rather than one (or more). I post two because it is much more time efficient to post two of each glyph once per day than it is to post one of each glyph twice per day. Particularly when I generally collect my mail while I'm AFK. That brings up its own metric to explore: the number times I actually sell two of a particular type of glyph between posts, and if there's any pattern (e.g., some very popular glyphs may warrant posting in larger numbers).

As I've mentioned a bazillion times before, I never cancel glyphs, and post at my fallback when the price goes under my threshold (in order to cap my competitions ceiling on resetting).

In other news, my inability to display more than a tiny fragment of my collected data in a spreadsheet has inspired my to get my web application up and running on a newly upgraded server, so I'll be slicing my data every which way over the coming months.

* It should be very much in scope, since the posting and selling data are both recorded with character and server information, so the analysis can be split and recombined.


  1. Thanks for posting this. I've been posting once every other day and only one at a time. This post has inspired me to take a deeper look at what I'm doing. I may make a post on my new strategy once I have devised it.

  2. Hey twitchie, as always it's fascinating to look at the data.

    Slightly off topic but - It would be great if you could do a post for the noobs amongst us who would like to collect and collate our own data in such a way as to how we can go about drawing it out of the various addons/sources and histogram/subject it to statistical analysis.

    1. For the graph above, there are two sources of data. Everything related to my sales comes from TSM_Accounting. I used to recommend a website that could produce a comma separated values file for import into a spreadsheet, but since the change to the data structure, that will no longer work. However, I believe that an export function is on the cards for TSM 2 (or already exists) so I'll have a look into that and report back.

      The other source of data is me actually recording the number of glyphs I post, each time I post them.

  3. Love this! Keep it up! I'm a sucker for data analysis and statistics.

  4. I'd be interested to see how often you actually sold two of one glyph in between postings. I just post one of each each cycle because of constant camping. They will all get undercut within minutes anyway. Plus, I find it helps you fly under the radar a bit. Some competitors tend to turn extra aggressive towards guys who post multiples.