Cookieless analytics, explained in plain words
How Tally counts visits without storing anything in the browser, and what you give up in return.
On this page
In Tally 2.1 we added cookieless mode: a way to count visits and funnels without storing anything in the visitor's browser. Many customers asked how it can work, and what they lose by using it. This post explains it in plain words.
How normal tracking works
Most analytics tools, including Tally in its default mode, store a random ID in the visitor's browser, usually in a cookie. The next time the person visits, the tool reads the ID and knows it is the same browser. That is how it can tell a new visitor from a returning one, and connect the steps of a journey across days.
In many countries, storing that ID requires the visitor's consent. That is why so many websites show a banner.
How cookieless mode works
In cookieless mode, Tally stores nothing in the browser. Instead, when a request arrives, our server combines a few pieces of information, such as the website, the visitor's general network information and the browser type, with a secret value that changes every day. From these, it creates a short code.
The same visitor gets the same code for the rest of that day, so Tally can connect the steps of one visit and count unique visitors per day. The next day, the secret changes, and so does the code. Yesterday's visitor cannot be connected to today's.
We count visits, not people. That is the whole idea.
Luis Ortega
What you keep, and what you lose
| Normal mode | Cookieless mode | |
|---|---|---|
| Visits and page views | ✓ | ✓ |
| Funnels within a day | ✓ | ✓ |
| Unique visitors per day | ✓ | ✓ |
| Retention over weeks | ✓ | ✗ |
| Journeys across several days | ✓ | ✗ |
For a marketing website, cookieless mode usually answers every question that matters. For a product where people sign in and return over weeks, you will want normal mode inside the product, with consent.
Is it accurate?
It is close. In our tests, daily unique visitors in cookieless mode were within 3% of normal mode. Two people on the same office network with the same browser may be counted as one; that is the main source of error.
