Google Display Network: Targeting, Formats, Placements and Brand Safety

by Francis Rozange | Apr 4, 2026 | Google Ads

google-display-network-ciblage-formats-1

Introduction: the ghost network of digital advertising

The Google Display Network is the channel everyone runs and almost no one defends. It moves billions of dollars in annual spend across roughly 2 million websites and 650,000 apps, yet most advertisers cannot articulate, in one sentence, what their Display campaigns are supposed to do. They run because someone clicked “Display” in the campaign type list, the impressions arrive by the hundred million, the click-through rate hovers below 0.5%, and the monthly report shows a number that looks like activity.

This guide treats Display as the strategic instrument it actually is in 2026: a top-of-funnel and remarketing channel, structurally incapable of driving direct conversions at the rates Search delivers, indispensable for the right buyer-journey stages, lethal for the wrong ones. We will walk through targeting types, the role of Display in the funnel, the divide between Responsive Display Ads and uploaded creatives, brand safety controls (including the February 2026 deprecation of the parked-domains exclusion), the realistic benchmarks (WordStream’s 2024 data shows a 0.57% median conversion rate against 3.75% on Search), the deprecation of Similar Audiences and what replaced it, the GDPR and CCPA-driven attrition that broke remarketing pools across 2024, and the awkward truth that most “Display” budget in 2026 actually flows through Performance Max.

The aim is not encyclopedic coverage. It is to give you the framework to decide, for any given account, whether Display deserves a budget line, what kind of campaign earns it, and what controls protect the brand and the spend. Sources are named inline and listed at the end.

The Display Network in 2026: what it is and what it is not

The Google Display Network is the inventory layer behind several Google Ads campaign types: standard Display campaigns, the Display half of Performance Max, Demand Gen, and (until their sunset in early 2024) Smart Display. Inventory comes from publishers running AdSense, Google Ad Manager, AdMob (mobile apps), and a handful of Google-owned surfaces (Gmail promotions tab, Discover until its 2023 absorption into Demand Gen, certain YouTube placements when bought through Display campaigns).

This breadth is what advertisers buy. It is also what they misread. Reach is not engagement, and engagement is not intent. A user reading a recipe blog is not in the same psychological state as a user typing “best CRM for SaaS startup” into Google Search. The Display impression interrupts. The Search query invites. Treating these two surfaces as if they ran on the same conversion logic is the single most expensive mistake in paid media.

Google’s own documentation, in the “About Display campaigns” help page, describes Display as suited to “build awareness and consideration” and to “drive conversions among existing customer lists or website visitors.” Note the order. Awareness first. Conversion explicitly tied to remarketing pools. The marketing literature has spent a decade trying to bend Display into a prospecting conversion channel and has spent that decade producing the same disappointing case studies.

The 2026 reality is that the GDN you control directly through standard Display campaigns is a smaller and shrinking share of total Display spend. Tinuiti’s Q4 2025 Digital Ads Benchmark Report shows that Performance Max now accounts for the majority of incremental Google Ads growth, and that Display impressions delivered through Performance Max have overtaken those bought through standalone Display campaigns for accounts above six-figure monthly budgets. That structural shift changes how you should think about Display strategy: the question is no longer “how do I run Display” but “what role does Display play across my Google Ads mix, and which campaign type buys it best.”

The funnel role: awareness and remarketing, not direct conversion

Display has one structural strength and one structural weakness, and both come from the same fact: users on Display surfaces are not searching. The strength is access. You can show a brand to a qualified audience that has not yet typed a query for your category. The weakness is intent. You are interrupting a reading or browsing session, and the click-through rate, even on the best inventory, will sit between 0.3% and 1%.

This makes Display a top-of-funnel channel for prospecting and a mid-to-bottom-funnel channel for remarketing. Both roles are legitimate. Neither role is “drive conversions on cold traffic at Search-level efficiency.”

For prospecting, the realistic objective is awareness lift, assisted conversions, and modest direct conversions on highly relevant audiences. WordStream’s 2024 Google Ads Benchmarks, drawn from over 17,000 campaigns, place the median Display conversion rate at 0.57%, against 3.75% on Search. The gap is not a bug. It is the channel.

For remarketing, Display becomes the most cost-efficient surface in the entire Google Ads stack. Users who have visited the site, viewed a product, abandoned a cart, or watched a video already carry intent. Remarketing converts these warm pools at multiples of cold-traffic Display rates, often at CPMs well below Search CPCs. This is where the channel earns its budget.

The strategic implication: Display campaigns whose primary KPI is direct cold-traffic conversion are mispriced and almost always underperform. Display campaigns whose primary KPI is awareness, assisted conversion, or remarketing efficiency consistently earn their place in the mix. Build the campaign around the right KPI from day one.

Targeting types overview: the four families

Google Display offers four broad targeting families: audiences, topics, placements, and contextual keywords. Each family addresses a different question.

Audiences answer “who is this person.” The system uses behavioral and demographic signals to build segments: in-market for specific product categories, affinity with broader interest patterns, custom segments built from keywords and URLs, demographic filters (age, gender, parental status, household income), life events (recently moved, getting married), and remarketing pools.

Topics answer “what is this page about.” You select content categories (Travel, Finance, Sports), and Google places ads on pages whose content matches.

Placements answer “where exactly does the ad appear.” You list specific websites, YouTube channels, or apps. This is the most controllable but also the most labor-intensive targeting type.

Contextual keywords answer “what words appear on this page.” Different from Search keywords: on Display, the keyword matches page content, not user query. A keyword like “lightweight running shoes” places the ad on pages discussing lightweight running shoes, regardless of who is reading.

Most campaigns combine families. The combination logic is critical: layered targeting (audience AND topic AND placement) narrows reach but tightens relevance. Stacked targeting (audience OR topic OR placement, in separate ad groups) broadens reach but loses precision. Google’s automation, particularly Optimized Targeting, blurs these boundaries by adding users beyond the manual selections when the algorithm predicts conversion likelihood. We will return to that.

Audience targeting deep dive

Audiences carry the weight of modern Display strategy. They are the channel’s answer to the death of third-party cookies and the privacy reshaping that has reduced everything else.

In-market audiences

In-market segments are users Google identifies as actively researching a purchase in a specific category. The signal mix includes search queries, comparison-site visits, review reading, shopping cart behavior, and engagement with category-specific content. Google maintains hundreds of in-market segments, from “Running Shoes” to “Enterprise Resource Planning Software” to “Wedding Planning.”

In-market audiences are the most consistently effective Display targeting type for prospecting because they front-load intent into the audience signal itself. A user in the “Running Shoes” in-market segment is not the same statistical creature as a user reading a fitness blog. The first is shopping. The second might be reading.

The catch: in-market segment quality varies. Niche B2B segments draw from thinner data pools, so the precision degrades. Consumer segments with millions of users, like “Travel” or “Auto,” carry better statistical reliability but face higher CPMs because every advertiser bids on them. Test segments at the edge of your category, not the obvious center.

Affinity audiences

Affinity segments target sustained interests rather than active intent. “Outdoor Enthusiasts,” “Foodies,” “Tech Early Adopters.” These are people whose long-term content consumption signals a lifestyle pattern, not a current shopping cycle.

Affinity is a top-of-funnel tool. CTRs run lower than in-market, and direct conversion rates are unreliable. The case for affinity is reach at low CPM for awareness campaigns where you want to plant the brand in front of a thematically aligned but pre-intent audience. Affinity also pairs well as a layered filter on top of contextual or placement targeting.

Custom segments

Custom Segments, Google’s flexible audience builder, replaced the older Custom Affinity and Custom Intent audiences in a 2022 consolidation. You feed the system a mix of keywords, URLs, and apps, and Google builds an audience of users whose behavior matches that profile. Two patterns dominate.

The competitive pattern: feed competitor URLs and brand-name keywords. The system finds users actively engaging with those brands, and you target them with a comparison message. A SaaS company can list competitor product pages, review-site URLs covering competitors, and competitor brand search terms. The audience that emerges is high-intent for the category, currently leaning toward someone else.

The categorical pattern: feed pain-point keywords and category-defining URLs. A project management vendor can list “team productivity software,” “project deadline management,” and authoritative category content URLs. Google builds an audience of users whose behavior matches the problem space, regardless of brand awareness.

Search Engine Land’s coverage of Custom Segments performance indicates these audiences typically deliver within 10 to 15% of in-market segment performance for the right use cases, with the advantage that the advertiser controls the input signals directly. They are a natural fit for niche B2B categories where Google’s pre-built in-market segments are too broad or too thin.

Demographic targeting

Demographic targeting filters by age, gender, parental status, and household income. The data quality is uneven. Age and gender are reasonably reliable for consumer brands; household income is a top-decile-tier estimate that varies by country and is unavailable in many markets. Parental status is signal-derived and noisy.

Demographics work best as exclusionary filters or secondary layers, not primary targeting. A luxury watch brand can filter out the bottom income tiers to concentrate spend on the top deciles. A pediatric healthcare company can layer “parents 25-44” on top of an in-market segment for higher relevance. Building a campaign on demographics alone is rarely enough.

Life events

Life events target users Google identifies as undergoing major transitions: moving home, graduating from college, getting married, having a baby, retiring. The segments are narrow by design and limited in scope (Google offers a small fixed list), but the conversion windows around life events are dense. A furniture retailer targeting “Recently Moved” or a financial services brand targeting “Recently Graduated” can capture concentrated demand.

The signal source is mostly behavioral pattern matching against historical cohorts. Privacy regulation has narrowed availability in some regions, particularly the EU under GDPR.

Remarketing audiences

Remarketing audiences are users who have engaged with the brand: site visitors, app users, video viewers, customer-list uploads through Customer Match. The conversion logic is straightforward: users who already know the brand convert at multiples of cold-traffic rates.

The 2024 audience attrition crisis (covered later) has made remarketing pools smaller and more volatile, but they remain the most efficient layer of Display spend. Standard remarketing serves generic ads to past visitors; dynamic remarketing, which pulls product feed data into the creative, lifts conversion rates substantially for ecommerce by showing the exact products the user viewed.

Topic and placement targeting

Topic targeting was once a primary Display lever. In 2026 it is mostly a contextual safety net, useful as a layer on top of audience targeting or as a brand-safety filter to keep ads on relevant content. Topics work by Google’s content classification of pages and apps, organized into a hierarchy (Travel > Adventure Travel > Hiking, for example). Selecting a topic adds eligibility to all pages classified under it.

The strategic value of topic targeting has declined as Google’s audience signals have improved. A user reading a hiking blog is less interesting, in conversion terms, than a user in the “Outdoor Equipment” in-market segment, regardless of what page that user is currently reading. Topics still matter for brand context: a financial advisor brand may insist its ads appear on financial content, regardless of who is reading.

Placement targeting is the most controllable family. You list URLs, YouTube channels, or app bundle IDs, and Google delivers ads only on those properties. Managed placements were once the gold standard for premium Display buying. They still are, for advertisers willing to do the research.

The workflow: pull placement performance reports from existing Display campaigns, identify the URLs and channels that drove conversions or strong engagement, build a list, and run a tightly placement-targeted campaign on that list. The result is concentrated spend on inventory that has already proved itself for the brand. Optmyzr’s analysis of placement-level data consistently shows the top 10% of placements delivering the bulk of conversions, with a long tail of zero-converting impressions absorbing budget if not actively excluded.

Contextual keywords work similarly to topics, but with finer granularity. You provide keywords; Google places ads on pages where the content matches. The match logic is content-based, not query-based. A keyword like “marathon training plan” places ads on pages about marathon training, irrespective of whether the user is researching marathons or just stumbled onto the page from a link. Contextual keywords are useful as a brand-safety mechanism (the ad appears only on relevant content) and as a fallback when audience signals are thin.

Responsive Display Ads vs uploaded creatives

Display has converged on two creative formats: Responsive Display Ads (RDAs), the default and overwhelmingly dominant format, and uploaded image ads, the legacy format with a narrower but still legitimate use case.

RDAs work by asset assembly. The advertiser uploads up to 15 images (in three aspect ratios: square 1:1, landscape 1.91:1, and optionally portrait 4:5), up to 5 logos, up to 5 short headlines (30 characters), one long headline (90 characters), up to 5 descriptions (90 characters), a business name, and a final URL. Google’s machine learning then assembles these assets into hundreds of possible ad combinations and serves them across the network, optimizing combinations against the campaign goal.

The advantages: scale, automation, and inventory coverage. RDAs serve into ad slots of any size and shape across the network without manual size production. The system tests combinations continuously and tilts delivery toward winners. For most advertisers without a dedicated creative team, RDAs deliver better aggregate performance than fixed-size uploaded creatives.

The tradeoffs: less creative control, no guarantee of how any specific combination will look, asset rejection issues at scale, and the algorithm’s tendency to over-favor a small subset of assets once it identifies winners (which can produce creative fatigue if the asset library is small).

Uploaded image ads remain available in standard sizes (300×250, 728×90, 336×280, 320×50, 970×90, 300×600, and others). The use case in 2026 is narrow but real: brands that need pixel-perfect creative control for premium positioning, seasonal campaigns with hand-crafted designs, regulated industries (pharma, finance) where creative review is strict, or A/B tests where you want to isolate the impact of a specific creative against an RDA baseline.

The practical recommendation: lead with RDAs, supply a rich asset library (at least 5 headlines, 5 descriptions, 5 images per aspect ratio), monitor the asset performance ratings (Low / Good / Best) Google surfaces in the interface, and replace Low-rated assets aggressively. Reserve uploaded ads for the brand-control use cases where they earn their cost.

Brand safety controls: 2026 reality

Brand safety on Display is the discipline of preventing ads from appearing alongside content that damages the brand. The controls have evolved over the past five years from coarse on/off settings into a layered system. They are also moving targets: Google has added, removed, and reorganized controls every twelve to eighteen months, and 2026 brought significant changes.

Content exclusions

Content exclusions sit at the campaign or account level and let advertisers opt out of categories of inventory based on the content’s nature. The categories include sensitive content groupings (tragedy and conflict, sensitive social issues, profanity, sexually suggestive content), content type categories (live streaming videos, embedded YouTube videos, games), and digital content labels (G, PG, T, MA, equivalent to film ratings).

The most consequential 2026 change: Google removed the “parked domains” exclusion category on 2026-02-10. Parked domains, sites that exist only to host ads with minimal or no original content, had long been a low-quality inventory source. The dedicated exclusion category let advertisers opt out wholesale. Its removal pushes brand safety responsibility onto other controls, principally placement exclusion lists and content quality filters via third-party verification.

The “Sensitive content” exclusion category remains the workhorse for most advertisers. The “Mature content” exclusion is essential for any brand that does not want adult-adjacent inventory. Digital content label exclusions let you cap inventory at G, PG, or T ratings. Google’s content suitability documentation details the exact category definitions.

Placement exclusions and exclusion lists

Placement exclusions are the most direct brand safety lever. You list specific URLs, YouTube channels, or apps where the ad must not appear. The list applies at the account level (via shared exclusion lists) or campaign level.

The discipline that distinguishes serious accounts: weekly review of placement reports, with aggressive exclusion of anything that does not fit the brand. Made-for-advertising sites, MFA in industry shorthand, are the major 2024-2026 brand safety problem. The IAB and the Association of National Advertisers’ 2023 Programmatic Supply Chain Transparency Study documented that approximately 15% of programmatic display spend was reaching MFA sites, sites engineered specifically to capture advertising impressions with low-quality, often AI-generated content. MFA sites do not show up as obvious in placement reports because they often carry plausible-sounding URLs. Active list maintenance, supplemented by third-party MFA exclusion lists from vendors like DoubleVerify and Integral Ad Science, is the practical defense.

Shared exclusion lists, accessible at the account or MCC level, are the operational tool: build a master list, apply it across all campaigns, update it weekly. Single-campaign exclusions are a maintenance nightmare in any account with more than a handful of campaigns.

Inventory types

Inventory types group publisher properties by Google’s own brand suitability tier. “Standard inventory” is the default and includes most ad-supported content meeting Google’s advertiser-friendly content guidelines. “Limited inventory” excludes a wider range of potentially sensitive content but reduces reach; “Expanded inventory” loosens the filter and increases reach at the cost of variance.

Most advertisers should sit on Standard. Limited makes sense for kid-focused brands, conservative B2B, regulated categories. Expanded is appropriate only for direct-response advertisers willing to accept context variance for reach.

Third-party verification

Integral Ad Science, DoubleVerify, and Moat (now part of Oracle) integrate with Google Ads to provide independent verification of ad delivery against viewability, brand safety, and ad fraud thresholds. For accounts above significant monthly Display spend, third-party verification is the standard hygiene control. The vendors maintain their own MFA exclusion lists, content classification systems, and bot-traffic detection that operate beyond what Google’s native controls offer.

Display Network benchmarks: the numbers

The single most important fact about Display performance is the gap between Display and Search. WordStream’s 2024 Google Ads Benchmarks, the most recent industry-wide dataset, draws from over 17,000 campaigns and reports the following medians.

For Display Network campaigns: average click-through rate of 0.46%, average cost per click of $0.63, median conversion rate of 0.57%, median cost per action of $90.80. Industry variance is substantial. Real Estate runs near 1.08% CTR; Apparel near 0.45%; B2B Services near 0.22%. CTR is higher on mobile (about 37% above desktop in WordStream’s data), reflecting both higher mobile inventory volume and the increased prominence of mobile ad units relative to surrounding content.

For Search Network campaigns, the same dataset reports a median conversion rate of 3.75% and a median CPA of $66.69 across industries. The conversion rate ratio of 3.75% to 0.57% is the structural truth of Display: at the median, Display converts at roughly 15% of the rate Search does. The CPA gap is smaller because Display CPCs are far lower, but the underlying conversion mechanics are not the same channel.

This is why “Display drives conversions” framing fails. Display drives some conversions, mostly on warm audiences (remarketing, customer match) and on highly relevant in-market segments. Cold-traffic prospecting on Display is a awareness and assist play, not a direct conversion play. Plan accordingly.

Cross-checking against Tinuiti’s Q4 2025 Benchmark Report confirms the order of magnitude. Display CPMs sit between $1 and $4 for most categories; CPCs run between $0.40 and $1.50; conversion rates land between 0.3% and 1% for cold prospecting and between 1% and 5% for remarketing pools. Numbers will drift, but the structure is stable.

Profitable Display: where the channel earns its budget

Display becomes profitable when matched to the right job. The order of priority, from highest expected return to most uncertain:

1. Standard remarketing. Site visitors who did not convert are the warmest audience available. Standard remarketing campaigns serve them generic brand or offer creatives across the Display Network. Conversion rates run several multiples of cold prospecting; CPMs are low because the audience is small and Google’s competition for it is more limited. Frequency capping is essential to avoid burnout (3 to 5 impressions per user per day for direct response, lower for awareness).

2. Dynamic remarketing for ecommerce. Ecommerce accounts with a Merchant Center feed should run dynamic remarketing as the default Display strategy. The creative auto-populates with the products the user viewed, so the ad serves relevance without manual production. Conversion rates lift further than standard remarketing because of product-level personalization.

3. Customer Match remarketing. Upload customer email lists and target Display ads to users matched in Google’s signed-in user base. Match rates run between 30% and 70% depending on list quality. Use cases: cross-sell campaigns, win-back campaigns for lapsed customers, premium-tier upgrade campaigns for existing accounts.

4. In-market prospecting on the right segments. For prospecting, in-market audiences carry the most reliable performance. Test segments at the edge of your category rather than the obvious center to find lower-CPM pools with adequate intent. Layer with topic or placement targeting for additional relevance.

5. Custom Segments for B2B and niche categories. Where in-market segments are too thin or too broad, Custom Segments built from competitor URLs and category keywords give the advertiser direct control over the audience signal. Performance approaches in-market quality for the right setup.

6. Awareness campaigns with affinity or topic targeting. The lowest-direct-conversion play, but legitimate when the campaign KPI is reach, brand awareness lift, or assisted conversion. Use affinity or topic targeting at low CPMs to plant the brand in front of thematically aligned audiences.

The campaigns that fail are the ones with cold-traffic prospecting on broad audiences with conversion KPIs and no remarketing layer. The math does not work, the channel fights the goal, and the budget bleeds.

The Similar Audiences deprecation and what replaced it

Similar Audiences was Google’s lookalike-targeting feature: feed it a seed audience (a remarketing list or customer list), and Google built an audience of users with similar behavioral patterns. It was the workhorse of Display prospecting for nearly a decade.

Google deprecated Similar Audiences across Google Ads on 2023-08-01 (existing audiences were generated until 2023-05-01 and removed from targeting on 2023-08-01). The deprecation was driven by privacy regulation: third-party cookie restrictions, ITP and equivalent browser-level limits, and the broader signal degradation that hollowed out lookalike modeling.

Two replacements absorbed the use case.

Optimized Targeting (in standard Display campaigns, formerly called Audience Expansion) automatically extends targeting beyond the manually selected audiences and signals to users the algorithm predicts will convert. The advertiser sets audience signals as starting points; Google’s machine learning expands eligibility to similar users in real time, optimizing against the conversion goal. Google’s Optimized Targeting documentation describes the mechanism. In practice, Optimized Targeting is on by default for many campaign objectives and behaves as a continuous, real-time replacement for static Similar Audiences.

Customer Match took on the role of the seed mechanism. Upload first-party customer data, and Google’s machine learning uses it as a signal both for direct targeting (customers who match) and as a base for audience expansion via Optimized Targeting. The shift from cookie-based lookalikes to first-party-data-seeded expansion is the broader 2024-2026 industry direction; Customer Match is the operational tool inside Google Ads.

For advertisers, the practical effect: Similar Audiences targeting in the manual sense no longer exists, but the underlying capability (find users like my best customers) is now distributed across Optimized Targeting and Customer Match. The user experience moved from “select Similar Audience” to “supply audience signal + let optimization expand.” The strategic implication is that first-party data investment (CRM hygiene, Customer Match list quality, conversion tracking integrity) became more valuable, not less, as the lookalike feature disappeared.

The 2024 GDPR and CCPA audience attrition

2024 broke remarketing pools across Google Ads, and most accounts have not fully recovered. The cause was a combination of regulatory enforcement, browser-level changes, and Google’s own consent infrastructure rollout.

In the EU, the 2024 enforcement of GDPR consent requirements, combined with Google’s Consent Mode v2 mandate (effective 2024-03-06 for advertisers serving EEA users), required explicit user consent for ads personalization and remarketing. Sites without compliant consent banners saw remarketing pools shrink dramatically. Google’s published guidance indicated that advertisers running Consent Mode v2 with default modeling would partially recover lost data through behavioral modeling, but the recovery was incomplete and varied by country.

In California, CCPA and the broader CPRA (effective 2023-01-01, with enforcement intensifying through 2024) added similar opt-out rights and required Global Privacy Control signal recognition. Other US states (Virginia, Colorado, Connecticut, Utah) added their own laws with overlapping requirements through 2023-2024.

The aggregate effect on Display:

Remarketing pool sizes in EEA markets dropped by 20% to 50% for many accounts during 2024, depending on consent banner implementation quality.
In-market and affinity audience signal quality degraded as third-party cookie deprecation in Chrome (delayed but advancing) reduced cross-site behavioral data.
Customer Match match rates fell in markets with stronger consent enforcement.
The reliability of cold-traffic Display prospecting on behavioral audiences declined further.

The strategic consequence: first-party data became the most valuable input to Display campaigns. Customer Match, properly maintained, is now disproportionately important relative to behavioral targeting. Server-side conversion tracking via Google Ads Enhanced Conversions and the broader migration to first-party data infrastructure are not optional hygiene; they are the operational layer that keeps Display campaigns from collapsing.

Display vs Performance Max in 2026

The strategic question that determines most Display budget allocation in 2026 is not “how do I optimize Display campaigns” but “should I run Display campaigns at all, or should this budget flow through Performance Max?”

Performance Max is Google’s automated, multi-channel campaign type. A single Performance Max campaign serves across Search, Shopping, Display, YouTube, Discover, and Gmail, with Google’s machine learning allocating budget across channels and inventory in real time. The advertiser supplies asset groups (creative assets, audience signals, conversion goals); Google handles the rest.

For accounts that have adopted Performance Max, a substantial share of Display impressions flow through Performance Max rather than standalone Display campaigns. Tinuiti’s benchmark data indicates Performance Max now accounts for the majority of incremental Google Ads growth above six-figure monthly spend levels. The Display inventory served through Performance Max is the same underlying inventory; the difference is who controls allocation.

The case for Performance Max as the Display vehicle:

Cross-channel optimization with Search and Shopping inside the same campaign.
Google’s most sophisticated machine learning for asset combination and audience signal use.
Reduced operational overhead.
For ecommerce with a Merchant Center feed, integrated product-level optimization.

The case for standalone Display campaigns:

Direct control over targeting, placements, and brand safety.
Transparency on Display-specific performance metrics.
Pure remarketing or pure awareness use cases where mixing with Search and Shopping is undesirable.
Brand-safety-strict accounts that need granular placement and exclusion control.

The practical 2026 architecture for most accounts: run Performance Max as the primary cross-channel performance vehicle, and run standalone Display campaigns for specific, isolatable use cases (pure remarketing, awareness with explicit reach KPIs, brand-safety-strict campaigns). Search Engine Journal’s coverage of Performance Max best practices documents this layered structure as the emerging norm in mature accounts.

Common mistakes

Treating Display as a direct-conversion channel for cold traffic. The math does not support it. Median Display conversion rates run at 0.57%, against 3.75% for Search. Cold-traffic Display prospecting with a CPA goal is structurally fighting the channel.

Skipping placement reports. Without weekly placement review, ads run on MFA sites, irrelevant content, and parked domains (or what is left of them). The 80/20 of Display optimization is in the placement exclusion list.

No frequency capping. Without caps, the same user sees the same ad twenty times in a day. CTR drops, CPMs rise, brand perception suffers. Caps of 3 to 5 impressions per day per user are a default, not a luxury.

Thin asset libraries on RDAs. Two headlines and one image is not an asset library. Google’s machine learning needs variety to optimize. Five of each, minimum, with regular replacement of Low-rated assets.

Mixing remarketing and prospecting in the same campaign. Conversion rates and CPMs differ enormously. Mixing them obscures both, distorts bidding optimization, and prevents independent budget control. Separate campaigns.

Ignoring brand safety until something breaks. Reactive brand safety, exclusion lists built only after an incident, is expensive. The MFA problem in particular requires proactive third-party exclusion lists.

Running Performance Max and Display campaigns without role separation. If both campaigns target the same audience for the same goal, they cannibalize and the data becomes unreadable. Either run Performance Max as the cross-channel performance vehicle and use Display only for specific remarketing or awareness slots, or commit to standalone Display with clear isolation from Performance Max scope.

Holding on to Similar Audiences workflows. Similar Audiences are gone. Optimized Targeting and Customer Match replaced them. Accounts still running campaigns built on Similar Audiences logic are working from a deprecated mental model.

Conclusion: Display in its right place

The Google Display Network in 2026 is not the same channel it was in 2018. Privacy regulation has hollowed out behavioral targeting. Performance Max has absorbed the bulk of automated Display spend. Similar Audiences is gone. The parked domains exclusion category is gone (as of 2026-02-10). Made-for-advertising sites have become the dominant brand safety threat. Conversion rates on cold prospecting sit at roughly one-fifth of Search, and that gap is structural.

None of this makes Display obsolete. It makes Display specific. The channel earns its budget on remarketing pools, on customer match, on tightly relevant in-market and Custom Segments, on awareness campaigns with explicit reach KPIs. It loses budget on cold-traffic prospecting with conversion KPIs and on campaigns run without active brand safety discipline.

The advertisers who succeed with Display in 2026 do three things consistently. They match the campaign type to the job (remarketing, awareness, or in-market prospecting, never the same campaign). They invest in first-party data (Customer Match, server-side conversion tracking, consent infrastructure). They maintain placement exclusion lists weekly and treat brand safety as a discipline, not a setting. Everything else is variation on those three themes.

Treat Display as the precision instrument it is. Stop asking it to do Search’s job. Use Performance Max as the cross-channel performance layer when it fits. Keep standalone Display for the use cases where direct control matters. The channel will pay back the discipline.