|
All of these new-fangled analytics sites seem to be stuck in the same dumb page-views-are-everything model as GA — just watching bird patterns, basically. Things haven't changed much since the bad old days of WebTrends and Analog. There's no profiling based on facts known to the app (user's age, gender, ethnicity, occupation) or behaviour (posts a lot of long texts, posts a lot of photos, is mostly reading/lurking, etc.), or Bayesian probabilistic analysis of new users. There is so much we want to know about our users, but no real turnkey tools to anlyaze the data with. Instead, everyone seems to be cobbling together their own toolchain with Hadoop and what not. Seems like an important piece of the analytics stack, yet I don't see anyone doing it. Is this something that anyone has started offering? (I'm thinking SaaS here, not enterprise stuff.) Analytics tools such as GA are also as terrible at tracking people across domains as these tools have ever been. We have a site which has single sign-on across many, many sites with different domains; we could have used the GA stuff to rewrite links in order to transfer the GA session ID across the domains -- but it's extremely awkward and gets very complicated when you need to implement it consistently both in the server logic (eg., redirects and emails) and the client logic (links, iframes, AJAX, etc.). It seems really silly when the app already knows the identity of the user and can track it consistently across domains. |