Cookieless analytics

The phrase "no cookies" is also used by tools that store an identifier in localStorage, which is the same thing in a different drawer. This page shows how to tell them apart without taking anyone word for it, and states plainly what measurement without a persistent identifier cannot do.

Free plan with no time limit. No card. Setup takes a minute.

The law does not talk about cookies

Article 5(3) of the ePrivacy Directive talks about storing information on terminal equipment and gaining access to information already stored there. It names no technology. Cookies, localStorage, sessionStorage, IndexedDB and a cache entry are all covered equally, because all of them are writing to someone else device.

How to check it yourself, in thirty seconds

You do not have to trust any vendor, us included. Open your site, open the browser developer tools and go to the storage tab. You will see everything the page has written, grouped by kind.

  1. Open the site in a private window so you start from empty storage.
  2. Press F12 and go to the Application tab in Chrome or Storage in Firefox.
  3. Look through Cookies, Local Storage and Session Storage for your domain.
  4. Click through a few pages and check again, because some tools write only on the second pageview.

If after all that there is nothing in storage that your own application did not put there, the analytics tool really does write nothing. If there is an identifier, which drawer it sits in makes no difference.

Recognising a visitor without storing anything

With no identifier on the device, it has to be produced on the server, and from material the browser sends to every site anyway. We compute it like this: a cryptographic hash of the site id, the truncated network address and the browser header, signed with a random salt for that day.

  • the salt changes at UTC midnight and its row is deleted after two days, so yesterday pseudonym cannot be matched to today one, not even by us
  • the site id goes into the hash, so the same person on two measured sites gets two different pseudonyms
  • the address is truncated before use: IPv4 loses its last octet, IPv6 is cut to /48
  • we store eight bytes of the result and nothing else

What you will not see because of it

This is a price, not a missing feature, and every tool without a persistent identifier pays it. Worth knowing before you switch rather than after the first report.

Tabelle seitlich scrollen, um die übrigen Spalten zu sehen.

What is absentWhy
New versus returning visitorsThe same person has a different pseudonym tomorrow
Cohorts and retentionThey require following a person across days
Cross device journeysThe pseudonym comes from the browser and the network
Funnels spread over daysThe funnel window is one visit
Multi touch attributionIt requires joining visits by the same person

Funnels within a single visit work normally, because the session identifier lives as long as the visit. What disappears is only what needs a memory longer than a day.

There is a side effect in the other direction too: the visitor count usually comes out lower than in a cookie based tool, because returning people are counted twice and several people behind one connection merge into one. It is an estimate that errs in a known direction, and that is how the dashboard describes it.

Questions that come up most often

Is cookieless measurement less accurate?
Inaccurate in a different way. It loses the new versus returning split, but it sees traffic a tool behind a consent banner will never see, because half the people never click accept. Which inaccuracy hurts more depends on why you are looking at the numbers.
Does Do Not Track change anything here?
Yes. The script checks that header and, when it is on, sends nothing at all: no request, not merely no data. Content blockers work too, because the script is an ordinary file from our domain and hides nothing.
What about visitors with JavaScript disabled?
You get an optional image tag to paste inside noscript, and those visits are recorded too: path, browser, system and country. They will not record the referrer, so they land in direct traffic, and we say so plainly instead of leaving a silent gap.
Can I compare your numbers with Google Analytics?
Run both side by side for a month and compare directions and proportions, not values. Agreement on values would not be good news here: it would mean one of the tools counts something other than it says.

Check it on your own site

Paste one line, open the storage tab in your browser and see that nothing was added. The free plan covers one site and has no traffic ceiling.

enpldees