Linking Google Business Profile to GA4: The Seven Metrics

Google added a Business Profile link to the GA4 admin on 8 June 2026, under Product links. Once the link is made, Analytics builds a reporting collection for the profile without any further configuration and it appears in the Reports menu straight away.

Seven metrics come across: interactions, calls, bookings, directions, website clicks, messages and menus. Six of those describe something that happened on the listing, before anyone reached the website at all.

What the metrics are actually counting

This is the part worth being precise about, because the collection sits inside GA4 and everything else inside GA4 counts sessions and events on a website.

Calls, directions, messages, bookings and menus are actions taken on the listing itself. Somebody found the profile in search or maps, pressed something, and never visited the site. Those actions have real commercial value – a call to a restaurant is worth more than a pageview – and none of them can be joined to anything on the site, because the person never arrived.

Website clicks is the only one that points across. And even there, the two sides count differently.

The gap between clicks and sessions

website clicks     counted by Google as the click leaves
                   the listing

sessions           counted by the tag once the page has
                   loaded and consent has resolved

between them       loads abandoned before the tag ran
                   consent refusals, where measurement stops
                   redirects that lose the referrer
                   bots and prefetches on one side only

  the difference is not an error. It is the first time both
  ends of that step have been visible in the same property

Anyone treating that difference as a discrepancy to be fixed will spend a long time on it. It is a measurement, and a fairly interesting one: it puts a number on how many people who intended to visit the site did not end up being counted as visitors.

Three limits, and what each rules out

The metrics are confined to the native standard reports. They cannot be pulled into an exploration, cannot be used in a comparison, and cannot have a filter applied. So the collection is something to read rather than something to analyse, and any deeper question has to be answered in the Business Profile interface instead.

Several profiles linked to one property are aggregated into a single dataset, with no dimension to separate them. For a business with one location this is irrelevant. For a chain it is decisive: the combined figure answers no operational question, because the useful version of every question is per branch.

History is a rolling six months. Which removes year-on-year comparison entirely, and for a local business that is the comparison that matters, since seasonality drives most of the variation and a month-on-month reading confuses season with performance.

What to do with it

For a single location the link is worth making and costs a few minutes. The collection then shows profile actions alongside site behaviour in one place, which is genuinely more convenient than two interfaces.

For multiple locations, the sensible approach is a separate property per location if the properties are separate anyway, and otherwise to keep reading the Business Profile interface, which does distinguish locations.

And in either case, the six-month window is the thing to plan around. A monthly export of the seven metrics into a sheet costs almost nothing now and is the only way to have a year of history in six months’ time. The alternative is discovering next spring that last spring is gone.