Google Ads against Analytics: where the differences come from

Every reporting tool is built differently and therefore reports differently. Differences between data sources come with the territory. The sting is in the roles: an ad platform reports on its own performance, which is not a neutral position.

An analytics tool is usually the neutral source. It only looks at where the visitor came from, provided that information is passed correctly. That is where it tends to go wrong.

Treat about 10% difference as normal. The truth sits somewhere in between.

Two ways: through autotagging from Google Ads, which appends the gclid parameter, or through manual UTMs.

Analytics reads that gclid from your landing page URL to identify traffic as Google Ads. Lose the parameter and you lose the attribution with it.

• Invalid clicks. Google Ads filters bot and competitor clicks out of your report and refunds them. Analytics counts the resulting sessions regardless.
• Bookmarks carrying the gclid. If someone bookmarks a page with the parameter, Analytics reads every later visit as Ads traffic. Google Ads counts no click.

• Clicks are not sessions. Two clicks within the same 30 minute session without closing the browser is one session in Analytics and two clicks in Ads.
• Autotagging is off. Without it and without UTMs, the traffic lands under organic or direct instead of Google CPC.
• The accounts are not linked. And if they are but you see no click or cost data, check that importing that data is actually switched on.
• Redirects strip your tagging. Sending visitors from HTTP to HTTPS, or through any other hop, can drop the parameter on the way.
• Ad blockers and disabled JavaScript. Analytics never fires while Ads has already counted the click.
• The tracking code doesn't load on the landing page. It sounds basic. It happens more than you'd think.
• The visitor leaves before the tag loads. The gclid never reaches Analytics and the click is invisible. A slow site costs you data here directly.

• Different attribution. Ads credits the last Ads click by default. Analytics uses last non-direct click in nearly every report. Arrive through an ad, return through a newsletter, and Ads claims the conversion while Analytics credits the newsletter.
• Click time against conversion time. Ads books a conversion on the date of the click, not the purchase. Buy on 25 April after clicking on the 22nd and Ads shows the 22nd while Analytics shows the 25th.
• Reporting speed. Ads lands about 3 hours later, Analytics about 9. Look too early and you are comparing different things.
• MCC-level conversion tracking. With several Ads accounts pointing at the same site, a conversion can be counted more than once.
• How often a conversion counts. Analytics counts one per session. Ads has no sessions and can count the same conversion per click. That is a setting on the conversion action.
• Cross-device. Ads measures conversions across devices. Analytics does not surface those by default.
• View-through conversions. Conversions after an impression without a click. They do not exist in Analytics.
• Phone calls. Calls through your call extensions that last long enough count as conversions. Analytics sees none of it.

Mostly imported transactions. Ads deduplicates on order ID so each transaction counts once. If several transactions with the same ID reach Analytics, they can be counted more than once there.

• Wait at least one conversion window before comparing. Two weeks back is usually enough.
• Use a generous period for meaningful numbers, at least two weeks to a month.
• Set the same lookback window on both sources.
• Report on both clicks and sessions, not one of the two.
• Use the same attribution model on both sides.

• Link your Ads and Analytics accounts.
• Turn autotagging on and drop manual tagging.
• Check the tag fires on every page.
• Keep redirects to a minimum.
• Make the site fast, because every second costs you clicks that never become sessions.
• Measure at MCC level if you run several Ads accounts.
• Invest in call tracking if phone calls count as conversions.