1. 18

  2. 4

    I’m not sure I like the accuracy comparison methodology. Google Analytics was considered a gold standard, which is not true – the gold standard is your Nginx/Apache/etc server log. Many people have ad blocker that prevents visitor tracking from working correctly. For example, I don’t see a problem that Microanalytics more or less consistently measured 12% more visitors than GA – I read it as 12% of the visitors (possibly) have an ad blocker that blocks GA but haven’t learned about Microanalytics yet.

    Also, I am missing at least one option that would rely on server logs directly. I think in the past it was considered bad to use server logs because (a) you can track more in JS – now we are trying to collect only what’s necessary, and (b) it’s harder to detect bots in server logs – well, many bots use headless browsers these days, perfectly tricking the JS counter code to count them as non-bots. In fact, server log analysis may be more correct these days due to ad blocking. Looks like Pirsch, suggested in this thread, supports collecting stats from Caddy directly.

    1. 3

      the gold standard is your Nginx/Apache/etc server log

      That’s filled with bots like you said. It also doesn’t see people who get pages out of local cache. There is no gold standard.

    2. 3

      Another great one that could have been included is pirsch. I recently switched from Ackee to them and have been really satisfied.

      1. 1

        Oh I like that this one is implemented in Go. I find hosting things that can ship as a compiled binary a lot easier than ones that don’t.

      2. 1

        Article misses the https://metrika.yandex.ru

        Yandex.Metrika has two killer features: heat click map & webvisor.

        1. 1

          Three: add an unparalleled respect for user privacy to the mix! /s