Consent mode modelling needs 700 clicks or you get nothing

Consent mode modelling is sold as the thing that recovers the conversions you lose when visitors decline. It only runs above 700 ad clicks over seven days, per country and per domain. Below that, Google models nothing, and the conversions from people who declined are simply gone.

Seven hundred clicks a week in Australia alone is a real number. A lot of mid-market accounts here sit at two or three hundred. Which means a chunk of the consent mode implementations sold in this market deliver exactly zero modelled conversions and nobody checks.

The thresholds, precisely

For Google Ads conversion modelling under consent mode: at least 700 ad clicks over a seven day period, counted per country and per domain, alongside enough consented traffic to build the model from.

For GA4 behavioural modelling, the requirements are different again: at least 1,000 events per day with analytics_storage set to denied for at least seven days, and at least 1,000 daily users sending events with analytics_storage granted on at least seven of the previous 28 days.

Two separate systems, two separate bars, and clearing one does not clear the other.

Note what per country and per domain does to a business operating across the Tasman or running separate domains for separate brands. If you split 600 weekly clicks across an Australian and a New Zealand domain, neither qualifies. Consolidated onto one domain with country targeting, you might.

What to do if you are under the bar

The single highest-leverage move is to stop losing consent in the first place, and in Australia that conversation starts somewhere unexpected.

Australian privacy law does not impose the same European-style consent requirement that blocking banners were designed for. A significant number of Australian sites run one anyway, inherited from a global template or added by an agency being cautious. If that is you, you may have cut your own measurement data to comply with a regulation that does not apply to your business.

That is a question for your legal counsel, not for a marketing blog, and the answer might legitimately still be to keep the banner. Global brands, health-adjacent categories and anyone processing sensitive information have reasons. But it should be a decision someone made on purpose.

If you do run a banner, its design is now a measurement variable. A banner where accepting takes one click and declining takes three yields a very different consent rate to one where both are equal, and the ethics of that gap are worth thinking about properly rather than defaulting into.

The things that work regardless of volume

These four do not depend on modelling thresholds at all, and together they recover more than modelling would.

Enhanced conversions. Hash the customer data you already collect at the point of conversion, and send it with the conversion. It matches against signed-in Google users and recovers conversions the browser could not report. It works at any volume.

First-party tag serving through Google Tag Gateway. Your tags load from your own domain, which sidesteps pattern-based blocking and lets identifiers be set server-side, outside Safari's seven day cap on JavaScript-set cookies. A day of work, no snippet change.

Offline conversion imports. If your sale closes on a phone call or a site visit, send the closed revenue back to Google against the original click identifier. This is more valuable than any amount of modelling because it changes what the bidder optimises towards: actual customers rather than enquiry forms.

Basic consent mode as the honest fallback. If you cannot clear the modelling bar, advanced consent mode's main benefit does not apply to you. Implement consent correctly, measure what you legitimately can, and spend the saved effort on the three items above.

Check whether modelling is running

In Google Ads, look at your conversion actions and check whether modelled conversions are being reported at all. If your account is under the threshold, the modelled portion will be absent rather than small, and the interface will not explain why.

In GA4, behavioural modelling appears as a data quality note on reports where it is applied. No note, no modelling.

Do this before anyone presents you with a consent mode business case. It takes five minutes and it determines whether the case is real.

The honest position

Consent mode modelling is a good feature for large accounts and an irrelevance for small ones, and it is marketed identically to both.

If you are running 700-plus weekly clicks per country, implement advanced consent mode properly and take the recovery. If you are not, you are not being punished, you just have a different set of levers, and those levers happen to be cheaper and more durable than modelling anyway.

The thing worth internalising is broader than one threshold. Almost every measurement feature Google ships has a volume floor written into documentation that nobody reads, and mid-market Australian accounts sit under a lot of them. Data-driven attribution needs 400 conversions on the action and 20,000 on the property. Behavioural modelling needs its thousand-a-day. Conversion modelling needs its 700.

Before you adopt any of it, find the floor and check where you stand.

Knowing which features you do not qualify for is worth more than switching on the ones you cannot use.