Google Released a New Analytics This Month

What was announced

Last week Google introduced a new version of Analytics, built on the property type that had been in beta as App plus Web.

The headline changes: a data model based on events rather than pageviews and sessions, measurement across a website and an app in one property, and a set of machine learning features for prediction and insight.

It is available now as a new property type, created alongside whatever you already have.

What has not been announced is any end date for the existing version, which continues to work and continues to be what most sites run.

The event model, in plain terms

The substantive change, and the one that takes adjusting to.

The existing version counts pageviews and groups them into sessions, with everything else layered on top as goals and events.

The new one counts events, and a pageview is simply one kind of event among many.

Which means a form submission, a phone tap, a video play and a page load are all the same kind of thing, described differently.

That is more flexible and it is also less immediately familiar, because the reports do not open onto the numbers people are used to reading.

What it means for a small business

• Nothing urgent. The existing version still works and is still supported.
• Data does not transfer. A new property starts empty.
• The reports differ, and some familiar ones are absent.
• Setup is a small job, roughly an hour.
• It is free, as before.

The second is the only genuinely time-sensitive point. Historical data does not move across, so a property created next year begins next year.

Why not to rush

Worth stating, because new releases attract urgency that is rarely warranted.

The reporting in the new version is less mature than what it sits beside, and several things people rely on are either absent or arranged differently.

Anybody who has built a monthly report around the existing version will find it does not reproduce directly.

And a small business with modest traffic is not being held back by its analytics, which is worth remembering before spending a weekend on this.

Nothing about the announcement requires anybody to act this month.

Why not to ignore it either

The counterweight, and it comes down to one thing.

Data does not backfill. Whatever date a new property is created, that is where its history starts.

Which means a business that sets one up now and leaves it alone has a year of comparison data available next October, and one that waits does not.

The cost of setting it up and ignoring it is an hour. The cost of needing a year of history and not having it is a year.

That asymmetry is the whole argument, and it does not depend on any prediction about what happens next.

A worked example

A business that had run the existing version since 2016 and had a settled monthly report.

They created a new property, added the tag alongside the existing one, confirmed both were collecting, and then went back to using the old reports.

Total time: about ninety minutes, most of it spent confirming the tags did not conflict.

Nothing changed about how they measured anything, and the new property quietly accumulated data.

Their view was that it was insurance rather than a project, which is a reasonable description of what this is for most small businesses.

The counter-case

Where more attention is warranted now.

A business with an app as well as a website, where measuring both in one property is a genuine capability rather than a technicality.

An online shop with meaningful volume, where the event model suits the way purchases and steps are actually structured.

Anybody spending substantially on advertising, where the reporting connections matter and are worth understanding early.

And anybody starting fresh, with no existing setup and no history to protect, who may as well begin on the newer version.

What is genuinely new

Beyond the data model, three things worth knowing about.

Cross-platform measurement in one property, which was previously awkward and is the clearest improvement.

Predictive metrics, which estimate the likelihood of somebody purchasing or leaving, and which need substantial data volume to be meaningful. A small local business will not have enough for these to say anything.

And a stated intention to function with less reliance on identifiers, which is a response to a changing privacy landscape rather than a feature as such.

For a five-page site with a contact form, none of those changes what you would do tomorrow.

The name and the confusion

A small practical point that will cause trouble for a while.

Both versions are called Google Analytics, and the interface for creating a property now offers a choice that is not obvious to anybody unfamiliar with it.

Which means somebody setting up analytics for the first time may end up on either, without realising there was a decision.

And any guide, plugin or article written before this month describes the older one, which is most of the material available.

Worth knowing when following instructions from anywhere: check which version they were written for, because the steps differ and the reports differ.

What we do not know yet

Worth being explicit about, since a good deal of commentary this week is filling the gap with assumption.

Whether the existing version will continue indefinitely, be retired at some point, or be maintained alongside the new one has not been stated.

Whether the reporting will reach parity with what people currently rely on, and how quickly, is unknown.

And whether the predictive features will be useful at small scale is untested, though the data requirements suggest not.

Which argues for the position above: set one up, keep what you have, and let the situation clarify rather than reorganising around a version that is a week old.

Anybody telling you confidently what happens next is guessing, including the people selling migration services.

What to actually do

1. Create a new property alongside your existing one.
2. Add the tag, without removing the old one.
3. Confirm both are collecting.
4. Set up form and phone tap events if you have half an hour.
5. Keep using your existing reports.
6. Look at the new one in six months.

The fifth is deliberate. Switching how you measure and how you read the results at the same time makes any change in the numbers uninterpretable afterwards.

What the existing version is actually telling you is covered in the first five numbers worth looking at.

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.