How advanced consent mode works
In the advanced implementation, Google tags load on every page view, including for people who have not yet answered the banner. Before consent, they behave in a stripped-back way: no cookies are read or written, no identifiers are attached, and what leaves the browser is a minimal, anonymous ping saying an event of some kind occurred. When the visitor accepts, the tags switch to their normal behaviour for the rest of the session.
Those pings are the whole point. They give Google a shape to work with — how many unmeasured visits and conversions occurred alongside the measured ones — and from that it estimates the activity consent hid. Without them there would be nothing to estimate from, which is the difference between this and the basic implementation.
The estimates come back into your reports as modelled conversions, blended with the observed ones rather than shown separately in most views.
Why advanced consent mode matters
It recovers reporting that would otherwise vanish, and it keeps automated bidding better fed. Smart bidding learns from conversion signals, so a campaign in a market with heavy refusal rates is bidding on a thinner picture in the basic setup than in this one.
It also carries a real trade-off, and it should be stated plainly rather than skipped. Something does leave the browser before the visitor has agreed to anything, even though it carries no identifier. Some legal advisers are comfortable with that and some are not, and views among European regulators are not uniform. This is a decision to take with advice for your markets, not a setting to enable because it produces nicer numbers.
Where advanced consent mode goes wrong
The most common technical fault is a default consent state that never runs, or runs after the Google tag. If the tag loads without seeing a denied default, it behaves as though consent were granted, which is both a compliance problem and a reporting one.
The most common interpretation fault is treating modelled figures as observed ones. Modelled conversions cannot be broken down to individual users, will not match your own records row by row, and can be restated when Google changes its methods. They belong in a total, not in a list you intend to follow up.
Third, modelling is not guaranteed. Google applies volume conditions before it will model at all, and accounts with modest daily click and conversion volumes — which describes a great many Nepali advertisers — may see no modelling and therefore none of the benefit that justified the choice.
How to act on it
Verify the sequence first. Use Google Tag Assistant and the browser’s developer tools to confirm the denied default fires before any Google tag, that no advertising or analytics cookie is written before a choice, and that the state updates the instant someone accepts. Test refusal as carefully as acceptance; it is the path nobody checks.
Then label your reporting honestly. Tell clients which numbers include modelled figures, and keep a parallel count of enquiries and orders from your own systems so there is always a source that owes nothing to a browser. If the consent architecture, the tags and the reporting need to be built as one piece, that is ordinary analytics and tracking work rather than a specialist project.
