Your Google Ads account shows fewer conversions than it did in spring. Spend is the same. Clicks are the same. Nothing in the campaigns changed.
If the drop started around the middle of June, there is a good chance your campaigns are fine. Google changed the rules for how ads data is collected, and many sites were set up for the old rules.
This guide explains the change, shows who is affected, and gives you a short test to find out if you lost conversions or only lost sight of them.
What Google changed on June 15#
Before June 15, 2026, two settings controlled the advertising data that Google Analytics tags sent to a linked Google Ads account:
- The Google Signals setting in Google Analytics.
- The
ad_storagesignal in Consent Mode.
Both had to allow the data. Many teams used one as a backup for the other.
Since June 15, there is one control. Google's own notice, Updates to Google Analytics Data Controls, says Consent Mode is now the single control for ads data. Google Signals only decides if Analytics data is associated with signed-in users for behavioral reporting.
In plain words: ad_storage is now the only gate between a visitor and your Google Ads account. If that signal is wrong, nothing else corrects it.
Who sees fewer conversions#
The change does not lower conversions on a site with a correct setup. It exposes setups that were already fragile. These are the four patterns we see most.
The granted update never fires#
Your banner sets ad_storage to denied by default. That is correct for visitors in the EEA, the UK, and Switzerland. When a visitor accepts, the banner must send an update that sets it to granted.
If that update is missing, late, or broken by a banner upgrade, every visitor stays denied. Google Ads receives no cookies and no click identifiers for people who agreed to be measured.
Consent Mode runs in basic mode#
In basic mode, Google tags do not load at all until the visitor accepts. Visitors who decline send nothing. Google has nothing to model from, so those conversions are gone from your reports.
In advanced mode, tags send cookieless pings for visitors who decline. Google uses those pings to model conversions. The modeled number is an estimate, but it is much closer to the truth than zero.
Tags fire before the consent default#
The consent default must be set before any Google tag runs. If your tag manager fires the Google tag first, the first hit of each session goes out with the wrong consent state. The click identifier from the ad is on that first hit. Lose it and the conversion cannot be tied to the click.
Nobody declared consent at all#
Some sites never added Consent Mode. For visitors in the EEA and the UK, Google has required it since March 2024. Without it, Google Ads does not use those visitors for conversion measurement or audiences.
The opposite case: more data than you intended#
One group sees the reverse. If you turned Google Signals off as a privacy measure, and your ad_storage default is granted, more data now flows to Google Ads than before June 15. Your numbers look fine. Your privacy notice may not be. Check this with whoever owns compliance on your team.
A 20-minute test: tracking drop or real drop#
Do this test before you change a campaign. A measurement problem and a behavior problem look the same in the Google Ads interface.
- Pick two periods. Four weeks before June 15 and the most recent four weeks.
- Pull real outcomes from your own system. Orders from your shop backend. Signups from your database. Leads from your CRM. These numbers do not depend on a tag.
- Pull Google Ads conversions for the same periods.
- Compare the two ratios.
| What you see | What it means |
|---|---|
| Real outcomes steady, Google Ads conversions down | Measurement problem. Your consent setup is hiding conversions from Google. |
| Both down by a similar amount | Real problem. Something on the landing page or checkout changed. |
| Real outcomes down, Google Ads steady | Paid traffic is fine. Look at other channels. |
If you have no independent count of what paid visitors do on your site, this is the reason to get one. A tool that counts sessions without cookies sees every paid visit, with consent or without it. Our post on the GA4 traffic gap explains how large that blind area can be.
Get a second count of your paid traffic
GrainQL measures every visit without cookies, so you can see what paid visitors do even when Google cannot. One tag, 14 days free, no card.
How to fix a measurement drop#
Work through these checks in order. Each one takes a few minutes.
- Open Tag Assistant and load your site as a new visitor. Look at the consent state on the first hit.
ad_storagemust showdeniedbefore you click the banner, for EEA and UK traffic. - Accept the banner. A consent update must appear, and
ad_storage,ad_user_data, andad_personalizationmust change togranted. - Check the order of tags. The consent default must come before the Google tag in the same page load.
- Check which mode you run. If Google tags do not load until consent, you are in basic mode. Talk to your compliance owner about advanced mode.
- Turn on enhanced conversions. Hashed first-party data from your form or checkout helps Google match conversions that cookies miss.
- Wait before you judge. Modeled conversions arrive with a delay, and modeling only activates when a country and domain have enough ad clicks. Give a fix one to two weeks.
One warning. Smart Bidding learns from the conversions it can see. If reported conversions fell by a third in June, your bidding has been learning from a smaller and biased sample for months. After you fix the setup, expect the campaigns to need time to adjust.
What to do when the drop is real#
If your own order count fell too, the consent change is not your problem. Something on the page is.
Start with what changed near the date of the drop: a release, a new banner, a new payment option, a price change. Then look at the behavior of paid visitors only. Where do they stop? Do they click something that does not respond? Does the form reject their input?
We wrote a full checklist for this case: Why Did My Conversions Drop? It starts from the moment you see the drop and ends when you have named the cause. If the suspect is a deploy, The Release That Silently Killed Our Checkout shows what that looks like in the data.
In GrainQL you can ask Kai the question directly: "Why did conversions from paid traffic fall after June 15?" Kai compares the two periods, finds the step where paid visitors stop, and shows you the sessions behind it.
The short version#
- Since June 15, 2026,
ad_storageis the only control for ads data. Google Signals no longer limits it. - A correct setup did not lose conversions. A fragile setup did.
- Compare Google Ads against your own order or signup count before you touch a campaign.
- If only the Google number fell, fix the consent setup. If both fell, fix the page.
Find where your paid clicks stop
Paste one tag and ask Kai why paid conversions moved. You get the step, the cause, and the sessions that show it.
