Ask ten performance marketers about the Google Display Network and you will get two answers. Half will tell you it is where budgets go to die: cheap impressions, accidental clicks, and a report full of mobile game apps nobody has heard of. The other half quietly run profitable display campaigns every month and would rather you did not find out how.
Both groups are describing the same product. The difference is almost never the platform and almost always the setup. Display is unforgiving of defaults, and Google's defaults are tuned for spend, not for your margin. This guide covers what the network is, which targeting still earns its keep, the exclusion work that stops the bleeding, and how to judge whether display is contributing anything at all.
What the Google Display Network Actually Is
The Google Display Network is Google's inventory of non-search ad space: roughly two million websites, plus YouTube, Gmail, and hundreds of thousands of mobile apps. By Google's own figure it reaches over 90% of internet users worldwide. Most of that inventory belongs to publishers who monetise through AdSense or Ad Manager, and Google acts as the broker between them and advertisers.
That framing matters because it explains the failure mode. On Search, you are buying against a query — a stated intent. On display, you are buying against a person and a context, neither of which the user has volunteered. Nobody browsing a recipe site has asked to see your B2B software ad. You are interrupting, and interruption has a much lower baseline conversion rate than answering a question.
This is not an argument against display advertising. It is an argument for buying it differently. Search campaigns reward bidding aggressively on high-intent terms. Display campaigns reward ruthless negative work, tight audience definitions, and patience with a longer measurement window. Treat the two the same and display will lose every time.
It also helps to know how the auction underneath works. Google Ads display buying sits on top of the same real-time bidding infrastructure that powers the wider programmatic ecosystem — our explainer on what RTB marketing is covers how those millisecond auctions price an impression. The practical takeaway: your display CPM is set by whoever else wants that same eyeball at that same moment, which is why a poorly targeted campaign can look cheap and still be expensive per outcome.
The Three Ways Display Campaigns Usually Fail
Before touching settings, it is worth naming the failure patterns, because most underperforming display accounts have all three.
Untouched placements. The campaign runs on automatic placements, spends 40% of its budget on made-for-advertising sites and in-app banner units, and nobody looks at the placement report for six months. The clicks are real. The intent behind them is not.
Search Network with Display Select, or the opt-in nobody noticed. Plenty of accounts have display spend inside campaigns their owners believe are pure search. The performance gets averaged into the search numbers, display looks fine, and the search campaign looks worse than it is. Segment by network before you conclude anything about either.
Last-click measurement on an upper-funnel channel. Display rarely wins the last click. Judged on last-click conversions alone it will always look like the worst channel in the account, including on days when it is genuinely driving incremental demand. This is a measurement problem masquerading as a performance problem, and it is the one that gets good campaigns switched off.
Fix the first two with settings. The third needs a different reporting approach, covered further down.
Targeting: What Still Earns Its Keep
Google has steadily folded manual display targeting into automated systems, but the controls that matter are still there if you build the campaign yourself rather than accepting a smart default.
Remarketing is still the highest-return display audience
If you run one display campaign, make it remarketing. You are advertising to people who have already visited, which restores the intent signal that display otherwise lacks. Segment by behaviour rather than lumping all visitors together: cart abandoners, pricing-page viewers, and repeat blog readers deserve different creative and different bids. Our guide to building Google Ads remarketing lists goes through the list logic and membership durations in detail.
Two rules keep remarketing from becoming an annoyance tax. Cap frequency — three to five impressions per user per day is plenty, and past that you are mostly paying to irritate people who were going to convert anyway. And set membership durations that match your actual sales cycle, not the 540-day maximum. A 30-day consideration window does not need a 180-day list.
Custom segments beat broad interest categories
Custom segments let you build an audience from the terms people search, the sites they browse, and the apps they use. In practice they outperform Google's packaged affinity and in-market categories, which are broad enough to include large numbers of people who will never buy. Feed a custom segment with competitor domains, high-intent search phrases, and the review sites your buyers actually read.
Similar segments and optimised targeting: useful, with a leash
Optimised targeting will expand beyond the audience you specified when it thinks it can find conversions. On a well-instrumented account with a healthy conversion volume that can work. On a new campaign with ten conversions a month it will burn budget learning. Leave it off until the campaign has a stable conversion signal, then test it as a deliberate experiment rather than a permanent setting.
Topic and placement targeting for context
Targeting specific topics or a handpicked list of placements gives you contextual control, which matters in regulated or reputation-sensitive categories. It scales poorly, but as a small campaign alongside remarketing it can be a reliable performer.
Placement Exclusions: The Unglamorous Work That Saves the Money
This is where display campaigns are won. Every serious display advertiser runs an exclusion routine, and most struggling ones do not.
Start with account-level exclusions before launch:
- Mobile apps and game categories. Unless you are running an app install campaign, exclude the
mobileappcategory::69500app category. Accidental in-app clicks are the single largest source of junk traffic in most display accounts. - Parked domains and error pages. Available as a content exclusion in campaign settings. Turn it on.
- Sensitive content categories. Tragedy and conflict, sexually suggestive content, profanity, and so on. Brand safety aside, these placements rarely convert.
- Made-for-advertising sites. These do not come as a tidy checkbox. You find them in the placement report: sites with generic names, enormous impression volume, near-zero conversion rate, and a bounce rate that looks like a typo.
Then make placement review a recurring job, not a one-off. Every two weeks, pull the placement report, sort by cost descending, and look at the top 50. Any placement that has spent more than roughly three times your target cost per acquisition with zero conversions goes on the exclusion list. Any placement converting well goes on a separate list you may want to target directly later.
Accounts that do this consistently typically see display cost per conversion fall substantially over the first two or three months, simply because the budget stops going to inventory that was never going to work. It is dull, repeatable maintenance, and it is the closest thing display has to a guaranteed win.
Creative That Actually Works on Display
Responsive display ads are now the default format: you upload assets, Google assembles combinations. The automation is fine. The inputs are usually the problem.
Give the system enough to work with — multiple landscape and square images, several short headlines, a long headline, and a few description variants — and make each asset genuinely different rather than five rewordings of the same sentence. Google can only test what you give it, and asset variety is what lets it find the combination that works on a given placement.
Two things consistently separate display creative that performs from creative that does not. First, one idea per ad. A display unit is seen peripherally for well under a second, so a single legible message beats a feature list every time. Second, an offer or a reason to act now. "Learn more" is not a reason. A named discount, a free audit, a specific piece of research, or a comparison worth clicking is.
Refresh creative on a schedule. Display fatigue shows up as a climbing CPM against a falling click-through rate within the same audience. If those two lines diverge for two consecutive weeks, your creative is tired regardless of what it did at launch.
Bidding, Budget, and What Display Should Cost
Display inventory is cheap on a CPM basis and expensive on a cost-per-outcome basis if you get the first two sections wrong. Before you set a budget, work out what an impression is worth to you rather than what it costs. Our CPM calculator converts between impressions, spend, and effective CPM so you can sanity-check a media plan against the reach you actually need.
For bidding, the sequence that works for most accounts:
- Start on manual CPC or maximise clicks with a bid cap while you build exclusion lists and gather conversion data. Automated bidding on a noisy, unfiltered account optimises toward noise.
- Switch to target CPA once you have a stable conversion signal — as a rough guide, thirty or more conversions in thirty days. Set the initial target near your recent actual cost per acquisition, not at your aspirational number, or the campaign will simply stop serving.
- Move to target ROAS only when revenue values are passing back reliably. That depends entirely on your tracking being correct; the conversion tracking guide covers the setup mistakes that quietly corrupt value-based bidding.
On budget, display should be funded from a growth or prospecting line, not taken out of a search budget that is still capped by impression share. If search is leaving profitable volume on the table, fund search first. Display earns its place once search is saturated.
Before committing, run the economics. The ROAS calculator tells you what return a given spend needs to clear, and the CAC calculator shows what you can afford to pay for a customer once you account for the full acquisition cost. If display cannot plausibly hit those numbers at realistic conversion rates, the honest answer is not to run it.
Measuring Display Honestly
Last-click attribution will tell you display does nothing. Attributing every view-through conversion to display will tell you it does everything. Both are wrong, and the second is how agencies justify budgets that should have been cut.
A more defensible approach uses three signals together:
Data-driven attribution in Google Ads to see display's assisted contribution rather than only its closing role. It is imperfect and self-reported, but it beats last-click for a channel that operates upstream of the final conversion.
View-through conversions, reported separately and discounted. Never add them to click conversions in the same column. Report them alongside, with an explicit haircut, and treat them as a directional signal rather than revenue.
A geo or audience holdout, which is the only real evidence. Pick a region, turn display off there for four to six weeks, and compare total conversions against a matched region where it kept running. If overall conversions do not move, the display spend was not incremental — no attribution model will change that. It is more work than reading a dashboard, and the only method that answers the question you care about.
The measurement window matters too. Display touches the user earlier, so a seven-day window will systematically understate it. Match the window to your genuine sales cycle and keep it consistent when you compare periods.
A Sensible First 30 Days
If you are starting display from scratch, this sequence avoids most of the expensive lessons:
Week 1. Build a remarketing campaign segmented by behaviour. Apply account-level exclusions for apps, parked domains, and sensitive content. Set frequency caps. Run manual CPC with a conservative bid.
Week 2. Pull the placement report. Exclude the obvious waste. Add a second campaign using a custom segment built from competitor domains and high-intent search terms, kept on a separate, smaller budget so it cannot cannibalise the remarketing spend.
Week 3. Review creative performance by asset. Replace the weakest headline and image. Check network segmentation to confirm nothing is leaking between search and display.
Week 4. Repeat the placement review. If conversion volume supports it, move remarketing to target CPA. Plan the holdout test that will tell you, six weeks from now, whether any of this was incremental.
Display rewards the operator who shows up every two weeks to do the boring work. That is the whole secret. Targeting options and bidding strategies matter far less than whether someone is reading the placement report and acting on it. For where impressions are worth buying across every channel, not just Google's, our ad placement guide applies the same discipline more widely.
FAQ
Is the Google Display Network worth it in 2026?
For remarketing, almost always yes — you are re-engaging people who have already shown intent, at a fraction of search CPCs. For cold prospecting, it depends on whether you can define a genuinely narrow audience and commit to ongoing placement exclusions. If nobody owns that maintenance work, the budget is better spent on search or paid social.
What is a good conversion rate on the Google Display Network?
Meaningfully lower than search, and comparing the two is misleading. Remarketing display typically converts several times better than prospecting display, so judge each campaign against its own purpose and against your target cost per acquisition rather than against a published benchmark.
How is the Google Display Network different from Performance Max?
Display campaigns give you campaign-level control over audiences, placements, and exclusions. Performance Max spans display alongside search, Shopping, YouTube, Gmail, and Maps inventory with far less visibility and control. Run standard display when you need to see and steer where the money goes; Performance Max trades that control for reach and automation.
Should I use automatic or managed placements?
Start automatic with a strong exclusion list, because managed placements alone rarely deliver enough volume to learn from. As the placement report identifies consistent winners, build a managed campaign around those specific sites and fund it separately. Most mature accounts end up running both.