Approved Is Not Serving: Limited Ad Serving Goes Wide

Google’s Limited Ad Serving policy — the mechanism that throttles impressions for advertisers it judges unqualified rather than disapproving their ads — is being extended to all of Google Ads. The Advertising Policies Help Center published the update on August 5, 2026, and it moves the policy beyond Search and YouTube to cover Gmail, Play Store and Discover inventory as well.

The operational consequence is a sentence most paid media teams have never had to say out loud: an approved ad and a serving ad are no longer the same thing. Nothing in this policy rejects creative. Google’s own page is explicit that individual ads are not disapproved. What changes is how many impressions an account is permitted to buy, decided at the account level, against criteria Google names but does not quantify.

This guide covers what the August update actually says, why its 2028 completion date is older than the announcement it appears in, the three-year arc that got the policy here, and the account audit worth running this week. Our earlier post on the June expansion of Limited Ad Serving to Search covers the qualification mechanics in depth — this one is about the wider scope and how to read the runway.

The change is documented on a dated article in Google’s Advertising Policies Help Center titled “Update to Limited Ad Serving Policy (August 2026),” posted August 5, 2026. Its opening line states that in August 2026 Google will update its Limited Ad Serving policy to cover all Google Ads, and that the page carries a new layout of the policy plus a list of best practices for reference. Search Engine Land reported the expansion on August 7 and PPC Land covered it the same week, both citing the same Google page.

Before this update, the publicly documented scope was Search and YouTube. The August page adds Gmail, Play Store and Discover to the named surfaces. Search keeps its own detailed best-practices section; the other four surfaces share a thinner one that comes down to completing advertiser verification, staying policy compliant, and continuing to build campaigns and creatives that accumulate positive user interactions.

Most coverage of the August expansion presents the 2028 completion target as a new element — a two-year runway attached to the all-of-Google-Ads move. It is not. Google’s June 12, 2026 update page, the one covering the Search-only expansion, carries the same sentence structure and the same year for a materially narrower scope.

Read the two pages side by side and the pattern is obvious: this is rollout boilerplate. Google set a 2028 horizon for a Search-only change two months before it set a 2028 horizon for everything. Nothing about the August announcement started a clock; the clock was already running, and the announcement widened what sits under it.

That distinction matters for planning. If you read 2028 as new, the natural conclusion is that you have two years of grace before anything bites. The accurate read is the opposite — enforcement on Search has been rolling since June under the same completion date, so the gradual phase is already in progress and the August update simply enlarged the inventory it applies to. Treat 2028 as the far edge of a process that has already started, not as the date the process begins.

Limited Ad Serving is not a 2026 invention. Search Engine Land covered its launch on August 31, 2023, when the mechanism was much narrower: throttling advertisers bidding on brand terms with an unclear relationship to that brand, during what Google described as a get-to-know-you period. Every source we reviewed covers only its own moment in this history, so we assembled the four documented stages in one place.

Two housekeeping notes on this table. PPC Land’s August 2026 piece dates the policy’s origin to November 2023; we use August 2023 because Search Engine Land published contemporaneous launch coverage on August 31 of that year, and a dated article written at launch is the stronger evidence. And Google’s general Limited ad serving overview page still described Search and YouTube only at the time of writing — the August-dated update article is the authoritative statement of current scope until the overview catches up.

Laid out this way, the direction of travel is unambiguous. Each stage widens either the surfaces covered or the scenarios that trigger a limit, and none has ever narrowed. Whatever your read on 2028 as a date, the trend line is the more useful planning input: impression eligibility is steadily becoming an account-level property rather than a per-ad one, across an inventory footprint that now spans most of where Google sells advertising.

The most commonly misread part of this policy is what it does to an account. It is not a disapproval mechanism. Google’s August page states plainly that individual ads will not be disapproved, and that the policy applies to a certain set of ad-serving scenarios in which only qualified advertisers can serve without impression limits.

That decoupling is what makes the situation hard to detect from the usual dashboards. Policy status stays green. Ad strength stays where it was. Disapproval counts stay at zero. What moves is delivery — an impression ceiling that no column in the interface labels as such. An account under a limit does not look broken. It looks quiet.

PPC Land characterised the shift as a move from query-level risk assessment to account-level trust evaluation. That is the outlet’s own framing rather than Google’s, but it captures the operational reality well: the unit being judged is the advertiser, not the keyword or the creative. It also explains why the usual remediation reflex — pause the offending ad, rewrite the headline, resubmit — has nothing to grab hold of here.

Google names seven factors that feed the qualification decision: account attributes, user activity and reports, account maturity, ad format usage, history of policy compliance, advertiser industry, and advertiser verification status. That list is unchanged between the June and August updates — the mechanics are covered in depth in our Search rollout guide, so the useful addition here is a different question: which of these can an advertiser actually observe?

The gap between naming a factor and quantifying it is not a technicality. Google’s own wording on user reports says that when users have persistently and disproportionately reported that an advertiser’s content, products or behavior do not meet their expectations, it may consider that advertiser unqualified and limit its impressions. Neither “persistently” nor “disproportionately” is defined anywhere on the page. Search Engine Journal made the same observation about the June Search-only update in June 2026, noting the policy identifies no reporting thresholds, qualification scores or warning systems — and since the seven-factor list is identical in the August version, that critique has simply travelled to a much larger share of Google’s inventory.

So the honest framing for a client conversation is that this is not a diagnosable condition. You cannot pull a report, compare it to a bar, and declare an account safe. What you can do is separate the factors you have direct control over from the ones you can only influence indirectly, and spend your effort on the first group. The table below is our own mapping of the seven factors against that test.

Google’s stated notification path is narrow. The August page says that unqualified advertisers who have a meaningful proportion of impressions in scope of this policy will receive an in-account notification. Read the qualifier carefully: the trigger for being told is a meaningful proportion of in-scope impressions, which is itself unquantified. An account throttled on a smaller slice of its delivery may simply not be notified.

Where a limit is applied, the remedy Google names is the Limited Ad Serving Appeals Form, linked from the policy page. What the page does not offer is any commitment on how long resolution takes.

For agencies that means the escalation script needs rewriting before it is needed. There is no disapproval to point at, no ticket reference to chase, and no published turnaround to quote to a client. The only honest status update is that an appeal has been filed and the account remains under review. Pairing that with a documented remediation log — verification completed on this date, these creatives retired, these landing pages corrected — is what turns an indefinite wait into a defensible one. Our wider platform-policy roundup tracks how the other major networks are moving on the same axis.

Because there is no diagnostic to run, the sequence below is ordered by controllability rather than by likelihood — start with what you can definitively close out, then work outward into the things you can only influence. None of this guarantees an account stays unthrottled; it removes the reasons that are within reach.

One structural item belongs on the list too. Because enforcement is evaluated at account level, PPC Land argued that agencies running many client accounts on shared creative templates or a common branding approach carry correlated risk across the portfolio — a pattern flagged in one account has structural siblings elsewhere. That is the outlet’s analysis rather than a Google statement, but it is the right question to ask of your own book: how many accounts would inherit a problem if one template turned out to be the trigger? Google’s corporate email requirement for agency access points the same direction — identity and account provenance are becoming delivery inputs, not admin chores. If you want that audit run across a portfolio rather than an account, that is the shape of our paid media engagements.

Google has published no figure for how many accounts are affected, no revenue impact and no advertiser count — not on the policy pages, and not through the trade coverage. Anyone circulating a number for this is estimating. The only quantitative facts available are the seven named factors, the zero thresholds behind them, and the dates.

What can be said with confidence is directional, drawn from the arc in section 03. Each stage of this policy has widened scope and none has narrowed it; the qualification criteria have stayed constant while the inventory they govern has grown; and the completion language has been reused verbatim across two updates two months apart. The reasonable expectation for the gradual phase is therefore more of the same shape — additional surfaces or scenarios folded in under the existing framework, rather than a new mechanism or a published scoring system.

There is a measurement consequence worth planning for as well. If impression eligibility can vary at account level without any interface signal, then year-over-year and account-to-account benchmarks quietly acquire a hidden variable. Impression share, auction insights and budget pacing all assume the ceiling is set by budget and competition. Building the possibility of a policy-imposed ceiling into how you read those reports — and saying so in the commentary — is more honest than attributing every dip to seasonality or competitive pressure. That reporting discipline is part of how we run analytics engagements for paid media clients.