Quick Answer: A steep conversion drop right after implementing Consent Mode V2 is almost always a measurement problem, not a performance problem. The three usual causes are tags that stopped firing because of a misconfigured consent setup, conversion modeling that never kicked in because your account sits below Google's data thresholds, and a cookie banner with a low granted rate. Run the three checks below before you touch a single bid.
The scariest part is not knowing which problem you have
Here's the pattern. Google tells you Consent Mode V2 is mandatory, so you implement it. Within days, your conversion count falls off a cliff. An account that reported conversions every single day suddenly reports almost none.
Now you're stuck in the worst spot in PPC. You can't tell whether your tracking broke or your ads stopped working. Those are two completely different emergencies with two completely different fixes. And if you manage client accounts, you get the bonus round of explaining a chart you don't understand yourself.
The drop can come from three different places. Guessing wrong wastes weeks. So diagnose first.
Step 1: Verify your tags fire after consent
Most Consent Mode V2 conversion drops trace back to the setup, not the mandate. Two mistakes show up constantly. The cookie banner loads but never actually gates your tags, which means your CMP was decorative before and is now sending denied signals. Or your GTM triggers fire before the visitor makes a choice, so the conversion tag runs with no consent state and gets dropped.
Open Tag Assistant in preview mode and load your landing page. Accept the banner. Confirm two things happen in order. The consent state updates first, with ad_user_data and ad_personalization moving to granted. The Google Ads conversion tag fires second. Then run it again and deny consent. With advanced Consent Mode, your tags should still send cookieless pings in the denied state. If nothing fires in either state, your setup is broken and no amount of modeling will rescue it.
Step 2: Check whether conversion modeling ever kicked in
Google's pitch was that modeling would backfill the conversions lost to denied consent. What the docs bury is that modeling has eligibility requirements, and they are steep for small accounts.
For Google Ads conversion modeling, the commonly cited bar is roughly 700 ad clicks per country across a rolling seven day window. GA4 behavioral modeling wants around 1,000 daily events from denied visitors plus a comparable base of consented users, sustained for weeks. There is no button to press. Modeling starts on its own once you qualify.
If your account sits below those lines, the modeled conversions you were promised simply do not exist. Small accounts feel consent loss at full strength. That gap between the promise and the eligibility fine print explains a huge share of the panic.
Step 3: Compare against a signal that lives outside the browser
This is the check that answers the real question. Did performance actually change?
Pull a number that no cookie can touch. Leads in your CRM. Form notifications in your inbox. Booked calls on your calendar. Compare the last two weeks against the two weeks before the consent change, side by side with reported conversions in Google Ads.
If CRM leads held steady while reported conversions cratered, your ads are fine and your measurement went blind. If both fell together, you have a real delivery problem, because smart bidding now has less signal to optimize with and may be spending worse. Either way, you finally know which fire to fight.
Raise your granted rate to shrink the whole problem
Whatever the diagnosis, a higher granted rate makes every downstream number better. Practitioners consistently report that a default banner lands near 80 percent acceptance for ad and analytics cookies, while a tuned banner can reach the mid 90s. At that point modeling only needs to cover a sliver of your traffic instead of a chunk of it. Banner wording, button design, and load timing all move the number. Test them the way you test ad copy.
Then build the backstop. Offline conversion import and enhanced conversions for leads pull your measurement out of the cookie fight entirely. First party data flows from your CRM to Google on your terms, and the next consent shakeup becomes a non-event.
If this episode showed you how fragile browser-based tracking really is, that's the actual lesson of Consent Mode V2. My free course Stop Flying Blind walks through building Google Ads measurement you can trust, so a consent banner update never blinds you again. Grab it free at Stop Flying Blind.
FAQs
Does Consent Mode V2 reduce real conversions or just reported ones?
Mostly reported ones. The people converting on your site did not stop because a banner appeared. But there is a second-order effect. Smart bidding learns from the conversions it can see, so a long stretch of thin data can degrade actual delivery. That's why Step 3 matters so much.
How long until conversion modeling starts working?
It begins automatically once your account meets the eligibility thresholds, typically within about a week of qualifying. If your click volume never clears the bar, it never starts, and you should plan around offline conversion import instead of waiting.
Should I just roll back Consent Mode V2 to get my numbers back?
No. Google has tied Consent Mode V2 to account standing, and ripping it out trades a measurement problem for a suspension risk. Fix the implementation instead. A correctly gated setup with a high granted rate recovers most of what you lost.
Does offline conversion import still work when a visitor denies consent?
The GCLID still arrives in your landing page URL regardless of consent choice, and capturing it into your form is a first party decision. Store it with the lead, upload the conversion when the deal closes, and confirm with your compliance folks how your privacy policy should describe it.