Analytics Without Cookies

Why this is suddenly a question

Nothing was banned. Three separate pressures arrived at once and the combined effect is that cookie-based analytics no longer sees what it used to.

Consent requirements mean a proportion of visitors are never measured at all, because they declined or ignored the banner.

Browsers have been shortening how long analytics cookies survive, so a visitor returning after a few weeks is frequently counted as a new person.

And the wider shift in how platforms handle identifiers has made cross-site measurement less reliable in general.

The result is not that your analytics is broken. It is that the numbers are less complete than they were and the gap is not evenly distributed.

What the cookie is actually doing

Less than people assume, and knowing this makes the alternatives easier to judge.

Conventional analytics sets a small identifier in the browser so that a second page view can be recognised as the same person as the first.

That identifier is what stitches individual requests into a session, and sessions into a returning visitor.

It is not what records the page, the referrer, the device, or the country. All of that arrives with the request itself and needs no cookie at all.

What genuinely breaks without one

• New against returning visitors, which becomes an estimate.
• Multi-visit journeys, where somebody researches then buys a week later.
• Session stitching across a long gap.
• Individual user paths through the site.

The first two matter for considered purchases. For a business whose visitors decide in one sitting, losing them costs almost nothing.

The alternatives, and their trade-offs

Three broad approaches, and they are not equivalent.

Privacy-focused analytics tools count visits without persistent identifiers, usually by deriving a short-lived signature from the request that is discarded daily. They give you pages, sources, devices and countries, and they deliberately cannot follow anybody across days.

Log-based measurement reads what your web server already records. Nothing is added to the page at all, so nothing can be blocked or declined, and the trade is that logs are noisy and need filtering.

Server-side collection moves the measuring from the browser to your own server, which improves reliability and does not by itself remove the privacy questions, since you may still be identifying people.

The consent point people miss

Worth stating plainly because it changes the calculation.

Where analytics requires consent and a large share of visitors decline, you are not choosing between complete data and reduced data.

You are choosing between reduced data with a banner and reduced data without one.

A measurement approach that collects no personal identifiers may not require consent in the same way, which can mean it sees a larger proportion of your visitors than the conventional tool it replaced.

Whether that applies to your situation depends on where you operate and what you collect, and this is a general description rather than legal advice.

A worked example

A regional services firm had a consent banner, and their analytics recorded roughly six in ten visits.

Comparing against their server logs, they could see the shortfall clearly but not who was in it.

They added a cookieless tool alongside the existing one for two months.

The cookieless figures were consistently higher, because it did not sit behind the banner, and the pattern of pages and sources was very close to the consented sample.

They kept the cookieless tool as the primary source, removed the analytics cookie, and simplified the banner accordingly.

They lost returning-visitor figures, which nobody had used for anything, and gained a page report that covered nearly everybody.

What you keep

More than the loss suggests, and it is most of what a small business actually acts on.

How many people came, which pages they read, where they arrived from, what device they used, and roughly where they are.

Which pages lose people, and which lead to an enquiry.

And crucially the conversions themselves, since a form submission or a phone tap is an event you can count directly without needing to know who did it.

Be honest about accuracy

No approach here produces a complete count and pretending otherwise leads to bad comparisons.

Cookieless tools undercount somebody who visits from two devices, since they cannot know it is the same person.

Log-based counting overcounts unless bots are filtered carefully, and bot traffic is a substantial share of raw requests.

Modelled figures, where a platform estimates what it could not observe, are useful for trends and should never be treated as counted events.

The practical answer is to pick one method, keep it consistent, and compare against yourself over time rather than against an absolute.

Do not run two and average them

A common instinct and a bad one.

Two tools will disagree, sometimes by thirty percent or more, because they define a visit differently, filter bots differently, and are blocked to different degrees.

Neither is wrong and the average of them is meaningless.

Running two in parallel for a short period is genuinely useful, but only to understand the offset between them, after which you choose one as your source of truth and read the other as a cross-check.

The counter-case

Cookie-based analytics is not obsolete and switching for its own sake is a cost with no return.

A business running paid advertising needs conversion data flowing back to the platform, and the platform's measurement is part of how the campaign is optimised. Removing it to be tidy will make the advertising work less well.

An online shop with a long consideration period genuinely needs to see a visitor return, because the first visit and the purchase are different sessions and attributing them matters.

And a business already collecting consent properly, with a good acceptance rate, has less to gain than one whose banner is being declined by half its visitors.

The question is what decisions you make from the data, not whether the method is fashionable.

What to do

1. Check what share of visits your analytics actually records.
2. List the decisions you have made from it in the last year.
3. Check whether any needed returning-visitor data.
4. Trial a cookieless tool alongside for two months.
5. Compare the shape, not the totals.
6. Pick one source of truth and keep it.

Step two settles most of these decisions, and it usually produces a shorter list than expected.

The banner itself is covered in a cookie banner that actually does something.

Frequently asked questions

West Coast Media Solutions Inc. provides web design, web development, hosting, digital marketing, and business consulting to organisations across Canada, drawing on more than twenty-five years in the field.