3 ms·

Comparing absolute change in Alexa rank seems fundamentally broken to me. It weights massively to early companies. Snipshot's rank rose by nearly 80,000 - that

by pacaro 14y ago

Comparing absolute change in Alexa rank seems fundamentally broken to me. It weights massively to early companies. Snipshot's rank rose by nearly 80,000 - that's undeniably good, but 37 of the companies on the list had a previous rank of less than 80,000... I would suggest comparing the delta of the log of rank, this gives a more interesting (to me) metric... For example this bumps Newsblur up from a "meh" 23 to 1! And AnyPerk from an exciting 6 down to a not so cool (unless you like catches) 22 (not picking on AnyPerk, it's just the most demoted of the original top 10) Under this change of methodology the biggest winner is WorkFlowy from 61->24, biggest loser's are Circle (39-67); FundersClub (40-68); and Cloudant (44-72)

beambot 14y ago
Presumably, the sites on Alexa follow a power law (based on traffic), so deltas on log-rank would indeed make a lot of sense. Of course... the data itself is quite interesting. Mad props to Danielle for compiling all of it -- this makes running our own quick calcs that much easier.
- dmor 14y ago
  I considered it but was on the fence so I wasn't going to use log this first time, but it sounds like that really might be the best approach. Give me a minute and I'll make a spreadsheet you guys can check out to see if it makes more sense. UPDATE: Here you go! https://docs.google.com/spreadsheet/pub?key=0ApqWF3CqjgjSdGMxbi1qYkdCcXVLY0lmVnk4R1FVV0E&output=html https://docs.google.com/spreadsheet/pub?key=0ApqWF3CqjgjSdGM...
  - teej 14y ago
    App store rank to download volume follows a power curve almost exactly, I think it's safe to assume similar here. I was going to make the same recommendation as beambot.
  - conesus 14y ago
    NewsBlur's #1 on the spreadsheet, which is crazy talk to me. I've been too busy scaling to even look at my analytics.
    - dmor 14y ago
      Well this is the kind of graph I dream about :) http://www.alexa.com/siteinfo/newsblur.com# http://www.alexa.com/siteinfo/newsblur.com#
chacham15 14y ago
I think that its both. Current traffic is like position and the delta is velocity. To know who is bigger now, the velocity is relatively unimportant. To try and guess who is going to be bigger next year, current position is less important.

by pacaro 14y ago

beambot 14y ago

Presumably, the sites on Alexa follow a power law (based on traffic), so deltas on log-rank would indeed make a lot of sense. Of course... the data itself is quite interesting. Mad props to Danielle for compiling all of it -- this makes running our own quick calcs that much easier.

dmor 14y ago
I considered it but was on the fence so I wasn't going to use log this first time, but it sounds like that really might be the best approach. Give me a minute and I'll make a spreadsheet you guys can check out to see if it makes more sense. UPDATE: Here you go! https://docs.google.com/spreadsheet/pub?key=0ApqWF3CqjgjSdGMxbi1qYkdCcXVLY0lmVnk4R1FVV0E&output=html https://docs.google.com/spreadsheet/pub?key=0ApqWF3CqjgjSdGM...
- teej 14y ago
  App store rank to download volume follows a power curve almost exactly, I think it's safe to assume similar here. I was going to make the same recommendation as beambot.
- conesus 14y ago
  NewsBlur's #1 on the spreadsheet, which is crazy talk to me. I've been too busy scaling to even look at my analytics.
  - dmor 14y ago
    Well this is the kind of graph I dream about :) http://www.alexa.com/siteinfo/newsblur.com# http://www.alexa.com/siteinfo/newsblur.com#

chacham15 14y ago

I think that its both. Current traffic is like position and the delta is velocity. To know who is bigger now, the velocity is relatively unimportant. To try and guess who is going to be bigger next year, current position is less important.