Analytics

‹ Go back home

This page tracks the amount of users that visit each site, but I managed to do it in a way that is more performant and privacy focused than Google Analytics using the secret technique of removing half of the features. Just kidding, but it is unique in the way that it uses zero cookies while still ensuring that a visitor is unique without being able to identify them.

How it works

Each person that visits my site brings with them some information that can be used to make a unique identifier for them. The server code combines all of this information into a single hash using SHA256. Specifically, the information is:

This means that we can count views across pages without being able to track a specific user's history. If a page has already seen a hash, then that is a repeat visitor, but that visitor will have a completely new hash if they decide to visit another page! All that appears to me is a list of hashes that visited a route. There is no way for me to extract anything meaningful from the hashes. This is what it looks like to me:

{
    "/": ["a/svXjSyMgxnSFw1vSYRJrau4Beraq/UfGzfPsp1Bg4", ...]
    "/catalog/": ["ZS9VAWJDvxufG76kbVdJ74ktvjlORt6dZqsarPC0r1c", ...]
    "/poems/mango": ["jwTocO0X61lx1VMMLc9XG+7eU7+5hq1EcQrot3NgM18", ...]
}

There's a lot going on in there.

Implementation

Before the information gets hashed, a random base 36 number gets thrown into the hash as well. This number is regenerated every 24 hours by the server. This ensures that visits are counted by day. For example, a person viewing a specific poem across two days will count as two views for that poem, allowing for some rudimentary session tracking.

This is what the hash generator looks like. If pageSpecific is false, then the hash is used to track visits to the site, which will still be completely different from hashes used for specific pageviews.

const generateUniqueId = (req, pageSpecific) => {
    const ip = req.ip;
    const ua = req.get("user-agent");
    const date = new Date().getDate().toString();
    let uniqueID = randomString + ip + ua + date;
    if (pageSpecific) {
        // add the page to the hash
        const page = req.originalUrl;
        uniqueID += page;
    }
    return crypto.createHash("sha256").update(uniqueID).digest("base64");
};

Future Work

I think there's a lot of value in this type of cookie-less analytics. Even though there's some good stuff here, there's a lot missing. For example, bounce rate and more detailed session tracking would be really cool to see implemented.

I think Navigator.sendBeacon() would work really well in conjunction with window.unload to provide specific timings of a session. The other obvious features that are missing are geographical data from visitors and user retention analysis. That last one probably can't be done within the constraints of this analytics framework.

Right now, the analytics module is cleverly implemented as express middleware so that a programmer can simply use app.use(analyticsMiddleware) at the beginning of their server code to have simple plug-and-play analytics on all routes of their site.

Who knows? I might continue working on it and publish it as an npm module. If I do, I'll update this page.