Google Search Partners: The Pre-Screen Advertisers Skip

The Google Search Partner Network pre-screen is a control most advertisers have never seen, let alone switched on — and it is not a recent arrival. Google’s own Search Partner Network announcements page files it under the heading “[April 2025] SPN pre-screen solution”, and DoubleVerify, Integral Ad Science and Zefr each announced their implementations on May 8, 2025. More than a year later, it still has no surface anywhere in the Google Ads interface.

That gap between “documented” and “adopted” is the whole story. The program does something no native Google Ads setting does: it acts on the full Search Partner Network domain list before your ads serve, rather than on the subset of sites that have already spent enough of your budget to appear in a report. It also does far less than its category name suggests, and Google says so in a single line of its own help article.

This is an evaluation, not an announcement. Below: how long the program has actually existed and how to check that yourself, the documented control deficit it patches, the structural reason after-the-fact exclusions cannot reach the long tail, two live contradictions in Google’s own documentation, and a decision matrix that includes the blunt option — switching Search Partners off.

Start with the dating, because it is checkable in under a minute. Google maintains a dated changelog for the Search Partner Network — the SPN announcements article, answer/16286960 — and the pre-screen solution appears there under a literal April 2025 heading, with outbound links to all three vendor press releases. Those releases are dated independently: DoubleVerify’s newsroom page, Integral Ad Science’s press release carrying a May 8, 2025 dateline, and Zefr’s press page all land on the same Thursday in May 2025.

Three separate companies announcing the same integration on the same day is unusually strong evidence for the date. Whatever the dedicated help article’s provenance, the product it describes had been on the market for well over a year before this post published.

The dedicated article itself, answer/17104980, carries no publication date and no last-updated stamp, and a Wayback Machine query for that URL returned zero captures at the time of writing. That is consistent with the article being newer than the program it documents — but absence from an archive is not proof of recency, so we assign it no date at all. Anyone attaching a date to that article is inventing one.

Run the absence check yourself: open the announcements article, scroll to the top of the changelog, and note the most recent heading. A current event would be sitting right there. This one is not, which is the whole point. What you are looking at is a mature control with an adoption problem — for an advertiser, a better starting position than an untested arrival.

The strongest argument for the pre-screen program is buried inside a Google help article about content suitability, answer/12764663, in a coverage matrix that maps six control types against each inventory source. It sits deep inside a long article otherwise concerned with YouTube. Read as a comparison, it is damning.

The Search & Search Partner Network row is Yes on three controls and No on three. Google Display ads, Google video partners and Mobile app are Yes on all six. That is Google’s own table, Google’s own scoring, and it is the cleanest available statement of why a third-party layer exists for this inventory source and not for the others.

One caveat before anyone over-reads the table: the Search Partner Network is not the Display Network with a different name. Different inventory, different controls, different reporting — the matrix above is the proof, not a rhetorical flourish. If you manage both, the muscle memory from Display placement management does not transfer cleanly, and that is a large part of why Search Partners drifts out of scope inside the wider account audit this sits inside.

Google’s dedicated help article describes the program as “an independent, third-party verified safety solution for the Search Partner Network (SPN)” offering “curated exclusion lists”. The mechanism is a weekly data handoff: “Google shares the full list of SPN domains with select third-party partners (DoubleVerify, Integral Ad Science, and Zefr) every week.” Three partners are named in the article. No others are.

Read the next step carefully; it compresses easily into something it does not say. The partners “package the domains they deem unsafe or unsuitable, categorized as low, medium, or high risk, into a recommended exclusion list for the advertiser.” The risk bands classify the excluded set — the domains already judged unsafe or unsuitable — not the network as a whole. A summary saying “partners band every SPN domain low, medium or high” is describing a different product.

Google lists three stated benefits: pre-screen, meaning a third party validates domains before your ads serve; independent validation, applied to the entire SPN domain list; and AI-powered classification. Availability is stated as global, for all Search Partner Network inventory within Google Ads. And enrollment is explicitly outsourced: advertisers “should work directly with their third-party partner of choice for implementation.”

There is one live disagreement between two first-party sources worth knowing before a vendor call. Google’s article describes three risk bands — low, medium and high. Integral Ad Science’s own published material describes its SPN pre-screen as based on “12 industry-aligned categories and 4 risk thresholds,” with weekly updates. Both are first-party statements from the companies operating the thing. We report both and reconcile neither: inventing a mapping between three Google bands and four IAS thresholds would be fabrication, and the difference is exactly the sort of detail to put to the vendor directly. If you want context on how these firms classify elsewhere, we have written on DoubleVerify’s media-quality methodology on other platforms and on IAS’s own media-quality benchmarks.

Note what the arrangement does not imply. Google shares a domain list; the classification and the resulting exclusion list are the vendor’s work. Nothing in the documentation makes the lists Google-verified, and nothing endorses one partner over the other two. You are buying a third party’s judgment, applied to Google’s data, on Google’s cadence.

Here is the honest reason to consider the program, and it is one Google never spells out — because spelling it out would undercut a feature Google shipped as an improvement.

Full placement reporting for the Search Partner Network exists. It shows the specific partner websites and impressions per site, reachable through Report Editor under “When and where ads showed” → Content suitability, and through the Google Ads API views detail_content_suitability_placement_view and group_content_suitability_placement_view. It is genuinely useful. It is also threshold-gated, in Google’s own words: “data is only included for placements that meet certain performance thresholds… A site is only listed if it has met at least one of the following conditions during your selected time range: 5 or more impressions · More than 0 clicks · More than 0 video views · More than 0 conversions.”

Follow that through. A partner site that served your campaign four impressions, no clicks and no conversions in the period never appears in the report. A site that never appears in the report can never be added to an exclusion list built from the report. So an exclusion workflow driven by placement data is, by construction, blind to the long tail — and on a network whose whole shape is a long tail, that is not an edge case. The pre-screen program operates on the full domain list instead, which is the one thing placement reporting structurally cannot do.

Be precise about what was and was not missing before. Site-level and account-level placement exclusions covering Search Partners have existed since around March 2024; the capability to exclude was there. What was missing was the site list you need in order to build those exclusions — impression-level placement reporting arrived for Performance Max in March 2024 and for Search, Shopping and App campaigns in August 2025. The pre-screen program is a different answer to the same problem: instead of a better list of where you have already spent, it gives you a judgment on where you have not spent yet.

The second practical constraint is capacity. Google publishes hard caps on placement exclusions, and they are generous but finite; the operational limits on how many rows you can add at a time are the bigger problem. The table below sets those published caps beside the only public sizing of the network we could find. It is illustrative only — the two halves come from different years and different sources, and the older figure is contested.

Do not subtract one from the other; there is no valid arithmetic between a 2026 platform limit and a 2023 third-party observation, and computing a “current” network size that way would be fabrication. The point survives without it. Whatever the true number, hand-maintaining exclusions at anything approaching that scale through an interface that accepts 5,000 rows per operation is not a workflow — it is a reason the whole category routes through a vendor.

Both of these are reader-verifiable in about thirty seconds, and both were live at the time of writing. We state them undated for exactly that reason: they may be corrected between our publishing and your reading, and either way the check takes less time than trusting us.

Contradiction one: does Google tell you where your ads ran?

Google’s Search Network overview article, answer/1722047, states that Google Ads “doesn’t provide information detailing the website where your ad was shown on the Search Network, unless you are running a Performance Max campaign.” Meanwhile the dedicated full placement reporting article, answer/17078188, and the August 2025 SPN announcement both state that site-level placement reporting is available for Search, Shopping and App campaigns. One of Google’s own pages is stale. This is an observation about the documentation, not a claim about the product — the reporting exists.

Contradiction two: how big is the network?

The Search partners definition article, answer/2616017, describes the network as extending reach “to hundreds of non-Google websites, as well as YouTube.” That phrasing — “hundreds” — is the exact wording a third-party study contested in November 2023 with a count of “more than 80,000” embed-carrying sites, and it is still on the page. We are not adjudicating the count. We are noting that the line that started the argument survived the argument unchanged.

While you are in answer/1722047, two other lines deserve a highlighter. Search partners are included by default: “When you create a campaign for the Search Network, search partners are included by default. You can opt out of these networks at any time through your campaign settings.” And Google hedges its own network’s traffic quality in its own help centre: “Ads can appear on search partner sites that display and link to products for sale. Keep in mind that clicks from these sites may not always reflect highly targeted traffic. Consider carefully whether including these networks aligns with your campaign goals.”

Two more details from the same article change how you should read partner performance. Click-through rate on search partner sites does not affect your Quality Score on Google — so partner performance is invisible to the quality signal most advertisers optimize against. And partners control rendering: they “can customize ad appearance on their sites regarding placement, size, formatting, and visual style”, and can host image and video ads. An ad on a search partner site need not look like a Google text ad at all. If you want a baseline for what unmixed Search performance looks like first, start from what normal Search performance looks like before partner traffic is mixed in.

The pre-screen program did not appear from nowhere, and the sequence explains what it was built to answer. A November 2023 third-party study put Search Partner transparency on the industry agenda; Google rejected the study on the record and, within months, traded an opt-out for reporting; reporting then widened; and somewhere in the middle, a third-party pre-screen layer appeared with essentially no fanfare.

The table below is our compilation. Every row carries its confirmation status, because half of what circulates about this network is third-party research and half is platform documentation, and they do not carry the same weight.

The November 2023 study needs careful handling, because its headline numbers travel far more widely than its caveats. As reported by AdExchanger, Adalytics said it observed “more than 80,000” sites carrying the Google custom-search or search-partner embed code — a count produced by its own methodology, in November 2023, against Google documentation describing “hundreds” of sites. The same reporting carried two further estimates: that the Search Partners program generated “roughly $10.5 billion annually” — an estimate derived from a third-party search-revenue figure, not a disclosed number — and that brand-unsafe inventory ran at “between 6% and 8%”, where the denominator is share of overall supply in the program, not share of spend, not share of impressions, and not share of advertisers affected. More than 20,000 of the sites the study identified were parked domains, which Google removed as an SPN ad surface effective February 10, 2026.

The pre-screen is one of four controls, not a standalone decision. What an advertiser actually has to choose is which of the four to run — and for a meaningful number of accounts, the honest answer is the fourth one.

One number will be put to you in this decision, and it needs its denominator attached every single time. In December 2025 Google published that advertisers “where Search Partner Network spend was at least 5% of their ad spend, can typically experience a 11% uplift in conversions when using volume-focused Smart Bidding strategies (Target CPA or Maximize Conversions), specifically in campaigns that were not constrained by budget,” with a parallel “7% uplift in conversion value” claim for value-focused strategies under the same conditions. That is a Google-observed aggregate for a narrow subgroup — advertisers already spending at least 5% of budget on SPN, in campaigns not budget-constrained, on named bid strategies. It is not “SPN lifts conversions 11%”, and restating it that way would be wrong on three separate axes.

There is also a tension between two things Google publishes: those uplift claims and the pre-screen program point in opposite directions. Google’s pre-screen documentation says the solution is designed for brand suitability and not to drive performance optimization. So the uplift figures cannot be used to justify pre-screening, and pre-screening cannot be expected to deliver them. If a proposal uses one to sell the other, that is the moment to slow the conversation down.

The forward-looking consideration matters more than either. This program is a control that fills a gap — and the gap is closing from the other side. Google’s temporary Performance Max search-partner opt-out closed on March 1, 2024, with impression-level placement reporting arriving three days later in its place; since then PMax’s Partners setting has begun letting some advertisers switch Search Partners off. If that capability becomes generally available, the value of paying a third party to curate exclusions for inventory you can simply decline drops sharply. Price the vendor engagement as something with a shelf life, not as permanent infrastructure — and if PMax is where most of your partner exposure sits, read it alongside how Performance Max distributes spend across networks before committing budget either way.

Four limits, all of which follow from the documentation.

It is not a performance product. Google’s own note forecloses that reading. Pre-screening removes inventory; removing inventory is as likely to reduce volume as to improve efficiency, and no primary source quantifies either effect.

The “industry standards” being verified against have no standards body. Google’s doc says partners classify sites against industry standards. IAS cites “industry-aligned categories”. Zefr cites industry standard safety and suitability standards. The reference framework all three point at is GARM’s Brand Safety Floor and Suitability Framework — and the organization behind it was discontinued in 2024. The taxonomy remains in wide industry use; the body that governed it does not exist. “Independently verified against industry standards” therefore means something weaker than it sounds.

Nobody can tell you how many advertisers run it. No adoption figure, customer count, share-of-advertisers number or price is published by Google or by any of the three partners, and we found none. When we say adoption looks low, that is an inference from the enrollment mechanism — no interface surface, no notification, an email address inside a help article — and not a measured fact. Treat any adoption claim you are shown, including ours, as unevidenced.

It does not change who classifies. Google supplies the domain list weekly; the judgment, the categories and the resulting list belong to the vendor. That is the point of an independent layer, but it also means the exclusion list is not Google-verified and two partners can hand you different lists for the same account. Ask each vendor to run the same brand profile and compare outputs before signing.

One further piece of context that reframes the whole discussion. Court documents in the United States Department of Justice case against Google, as quoted by PPC Land, describe the underlying mechanism: Google “currently offers an ads syndication product known as AdSense for Search or AFS, which syndicates Google search text ads from Google.com advertisers who do not opt out to Google’s ‘Search Partner Network,’ i.e., third-party publishers”. We have not fetched the filings ourselves, so that is PPC Land’s reporting of court documents rather than our reading of them. The framing is useful anyway: the Search Partner Network is a publisher monetization product that advertisers are enrolled into by default, and the controls discussed in this post are all attempts to make an advertiser-side decision inside a publisher-side product. If your account governance is being tightened generally, this belongs on the same list as recent Google Ads account-governance changes and the corporate-email pilot changing who can touch account settings.