What Is Display Advertising? Formats, Platforms, Costs, and Measurement

Separate format, inventory, campaign, and audience

A banner or responsive asset describes creative. GDN is inventory; Demand Gen is a campaign structure. Retargeting describes audience selection using previous behavior and also exists outside display. Video and social advertising can overlap in format and placement rather than forming mutually exclusive categories.

Searchers and people using websites or apps can be at different buying stages. Avoid assuming search serves only immediate buyers and display only unaware audiences. Specify what the person needs to understand. Compare alternatives with the online advertising guide and paid search guide.

YouTube's old overlay ads on videos ended on April 6, 2023. Do not use them as an unqualified current display example. This does not mean every advertisement near or below a video has ended. Consult the official YouTube format guidance for the actual placement being discussed.

Check current Google and LINE Yahoo transitions

Google began a phased migration-tool rollout in June 2026. GDN remains available, including GDN-only selection in Demand Gen; other inventory can include YouTube, Discover, Gmail, and Maps in beta. Future creation restrictions and automatic migration remain announced later steps, not proof that all accounts have moved.

Migration changes feature support: manual CPC, viewable-impression bidding, and pay-for-conversions are not carried over as equivalent options. Some demographic settings become signals rather than hard constraints. HTML5 has announced future availability. Migration cannot be reversed, and earlier same-day spending is not recognized by the new campaign. Check the official migration and feature table before acting.

In Japan, April 2026 integration introduced LINE Yahoo Ads Display Ads and the renamed LINE Yahoo Ads Search Ads. Legacy LINE Ads delivery is scheduled to stop around late October 2026, with full termination around the end of March 2027. Those are planned dates, not completed shutdowns. The source is LINE Yahoo's April 2 announcement; check current account notices before scheduling work.

Do not choose a platform from unsupported claims that it inherently reaches a particular age or profession. Require current market, audience, and delivery evidence. A familiar historic product name is not proof that the service is still offered.

Choose creative and outcomes from the objective

Define what the message should help the reader understand and do. Awareness needs appropriate contact and visibility; consideration needs relevant information; acquisition needs accepted outcomes and quality. One immediate CPA target cannot adequately evaluate every role.

Review images, headlines, descriptions, logo, CTA, and destination together. Check asset specifications, disclosure, claims, pricing, permissions, and operation. Preview small screens and sound-off video where relevant. Do not insert unsupported customer counts or promised results to make the asset appear more persuasive.

Optimized delivery across several assets is not automatically a randomized A/B test. For a causal comparison, use the experiment design and interpretation guide and verify formal assignment and analysis conditions.

Separate bid objectives from payment events

Target CPA seeks an average outcome cost. It is neither an individual conversion-cost cap nor a pay-only-on-conversion contract. Demand Gen billing varies: GDN and YouTube use CPM, Gmail uses the first expansion or specified initial action, and Discover depends on format and settings. Do not assume universal click billing or per-second video billing. Consult the official billing and bidding FAQ.

Budget for production, management labor, landing page work, and measurement as well as media. Use contribution and close rate to determine allowable acquisition costs; label media-only CPA separately from broader costs. The CPA guide provides worked examples.

Manual frequency caps are not available in every campaign. Demand Gen's FAQ states that manual impression-frequency control is not supported. Check the actual service, then compare reach, impressions, frequency, and outcomes over the same period. Avoid a universal claim that three or five exposures is optimal.

Calculate viewability and costs with a worked example

Google Active View uses at least 50% of display-ad area visible for one second, or 50% of a playing video's area for two seconds. Large display ads of at least 242,500 pixels have a separate 30%-for-one-second condition. See the official Active View explanation. Meeting a visibility condition does not prove attention or understanding.

Fictional 30-day campaign: media spend is JPY 100,000, with 100,000 impressions, 80,000 measurable impressions, 60,000 viewable impressions, reach of 20,000, 500 clicks, and ten accepted click-through outcomes. These are arithmetic inputs, not platform benchmarks.

The viewable-rate denominator is measurable impressions, not all served impressions. An unmeasurable impression is neither confirmed viewed nor confirmed unviewed. Demand Gen and some other reports require format segmentation for suitable viewability interpretation; different viewing experiences should not be compared as though identical.

The spend-to-viewable ratio above is a deliberately defined reporting calculation. It does not prove vCPM billing or necessarily match a platform column with a different cost scope. Average frequency of five over 30 days does not mean every individual saw exactly five ads or that five was optimal. Like CTR, the rate needs a named denominator, period, and population.

Use delivery and migration checks before and after changes

1. Outcomes: define acceptance, purchase, quality, and maturation.
2. Creative and destination: verify logo, message, terms, disclosure, links, layout, and operation.
3. Audience and exclusions: check region, attributes, expansion, placements, and suitability at the applicable account, campaign, or ad-group scope.
4. Costs: verify bid goal, payment event, budget, and migration-day spending exposure.
5. Migration: record eligibility, unavailable features, reversibility, approval state, owner, and timing.
6. Measurement: test success, failure, reloads, duplicates, production host, consent, and cross-domain transitions.
7. After delivery: reconcile placements, assets, costs, outcomes, and quality; record differences and uncertainty.

Exclusion support depends on feature and migration context. Official migration tables and general FAQs can address different scopes, so do not assume an old campaign's exclusions retain identical effect in the new structure. Check the current documentation and actual controls. A before-and-after change involving several settings also cannot identify one setting's causal effect.

Distinguish attributed outcomes from incrementality

Retargeting includes people who already had interest. A favorable CPA or ROAS therefore does not by itself prove incremental impact. Separate click-through, view-through, and engaged-view windows, and check overlapping claims. Adding view-through outcomes to the worked example would not preserve its click-through conversion-rate definition.

Increased impressions or branded searches may also reflect other changes. Asking how many outcomes would not have occurred without the advertising requires an additional evaluation design, such as a suitable controlled experiment. Distinguish observations from hypotheses and avoid decisive claims from a small number of outcomes.

To organize initiative owners, plans, media and production costs, and monthly budget-versus-actual reviews, explore Xtrategy's features. Preserve delivery and migration decisions alongside the work completed and the costs reviewed.