What a display campaign needs before launch
The display hub describes display as a tool for reach, reminder and reactivation rather than for capturing a search someone has just typed. It also lists the problems that follow a careless launch: budget spent on mobile apps and games, ads shown far beyond the chosen audience because optimised targeting was on by default, images cropped badly in some slots, the same person seeing the ad many times a day, and conversions reported by Google Ads far above analytics. Each of those is cheaper to prevent at launch than to diagnose later.
| Launch decision | Prevents | Owned by |
|---|---|---|
| One job per campaign | Budgets, bids and reports for two jobs blended together | This tutorial, step one |
| Bid strategy matched to the job | Cheap clicks and easy conversions on low-value inventory | Bidding hub |
| Targeting route | Audiences too loose to mean anything, or too narrow to deliver | Display targeting guide |
| Optimised targeting decision | Spend drifting outside the chosen audience | Display targeting guide |
| Exclusions before launch | Apps, made-for-advertising sites and unsuitable content | Low-quality placements diagnosis |
| Frequency cap | The same person seeing the ad all day | Display targeting guide |
| Native square and landscape assets | Cropped images and clipped text | Responsive display ads guide |
| Conversion reading agreed | View-through counts mistaken for caused sales | View-through versus click-through comparison |
Before you start
- Read access to the account's conversion actions, with which are primary, and the shared exclusion lists if the account already has them.
- The brief: what the campaign is for, who it should reach, the budget and the measure it will be judged on.
- Audience material: remarketing lists and their sizes, customer lists, and any competitor or category sites a custom segment could be built from.
- Creative: images in native square and landscape versions, square and landscape logos, headlines, long headlines, descriptions and the business name, checked against the responsive display ads help page.
- An analytics view of engagement by source, so the first placement review can compare what Google Ads reports with what visitors did.
- The campaign launch QA checklist, for the checks every paid campaign shares.
The steps
Write the campaign's single job
Write one sentence naming the job: reach a defined audience at a sensible frequency, bring back people who have visited, or find new people likely to act. The marketing funnel framework helps place it: awareness, consideration or conversion. Then write the measure that matches: reach and cost per thousand impressions for awareness, conversions and cost per conversion for action, return visits and conversions for RemarketingAdvertising to previous visitors or engagers.. One campaign, one job. The targeting guide lists one campaign for prospecting and remarketing among its common mistakes, because two jobs in one campaign blend two budgets and two audiences in one report.
Choose a bid strategy that rewards it
The bidding hub separates the strategies that optimise for clicks, impressions or views, including target CPM and viewable CPM, from the Smart Bidding strategies that optimise for conversions or conversion value. Reach jobs belong with the first family; action jobs with the second, provided conversion tracking is complete and the primary action is the outcome the business wants. The low-quality placements diagnosis explains the trap: a campaign on maximise clicks buys the cheapest clicks available, which on display often means apps, and a conversion strategy with an easy primary action finds the inventory that produces easy actions. The CPM calculator turns a reach budget into impressions.
Choose the targeting route
The targeting guide divides display targeting into statements about the person and statements about the page. Audience targeting follows the person: in-market, affinity, life events, detailed demographics, custom segments and your own lists. Content targeting, or Contextual TargetingChoosing where ads appear from the content of the page, app or video being viewed rather than from data about the viewer., follows the page: topics, placements and keywords. Choose the route from the guide's layering table: affinity or topics for broad awareness, in-market or custom segments for prospecting, a segment plus a topic or placement layer where context matters, your own lists for remarketing, managed placements for brand-safe reach. Put one idea in each ad group, so the reports can show which worked, and layer only where the extra precision is worth the lost reach.
Decide on optimised targeting
Optimised targeting is on by default for new Display campaigns. The targeting guide explains that it looks beyond the selected segments for people likely to convert, treating them as a starting hint rather than a fence, and that exclusions are still respected when it is on. Leave it on for a conversion-bid campaign with solid tracking and enough conversions to learn from. Switch it off for remarketing, for a tightly briefed awareness buy and for any campaign whose point is to reach a particular audience, because the reports will otherwise describe an audience nobody planned. Record the decision in the launch notes; the diagnosis page recommends making it an explicit line, never a default.
Set exclusions before launch
Exclusions apply on top of whatever targeting is chosen, which is why the targeting guide puts them first in the order of operations. Apply the account's shared Placement ExclusionsLists of websites, apps, video channels or app categories where an advertiser's ads must not appear. list, or create one, so the campaign inherits every placement already rejected; the placement exclusions entry explains how lists are shared. Add app category exclusions where mobile apps carry most low-quality impressions, topic exclusions for subjects the brand will not sit beside, and the content exclusions and inventory type that match the brand's tolerance. Then add audience exclusions: recent converters from acquisition remarketing and existing customers from prospecting, which the guide to excluding existing customers covers across campaign types.
Set a frequency cap
A frequency cap limits how often one person sees the campaign's ads in a period. The targeting guide says a cap belongs on every remarketing campaign, because small lists with large budgets produce the high FrequencyAverage times a person saw an ad. that annoys people and wastes money. Derive the cap from the audience size, the budget and how long people take to decide, not from a habit: the reach and frequency calculator shows how quickly a list saturates at a given budget. For prospecting and awareness, a cap still protects against a few people absorbing the impressions. Set list membership durations to match the purchase cycle at the same time. If the campaign later stops spending early each day, the diagnosis for a Display or video campaign that is not spending checks whether the cap, the list or the targeting is the limit.
Build responsive display ads
Build the ads from assets that work in every combination, as the responsive display ads guide explains: native square and landscape images rather than one shape cropped by the system, the subject in the centre and no text baked into the image, square and landscape logos on plain backgrounds, headlines that each stand alone, descriptions that hold nothing essential, and the business name people recognise. The help page is the reference for ratios, sizes and counts. Write different assets for remarketing and prospecting, because the viewer and the job differ. Before launch, check every ad in the preview at the smallest and largest sizes and in a native layout, and review any automatic enhancements, switching off what does not fit the brand.
Check how conversions will be counted
Display is seen far more than it is clicked, and Google Ads reports view-through conversionsA conversion credited to an ad that was shown but not clicked, within a set view window., conversions after an impression without a click, within a window the advertiser sets, separately from the main conversions column. Decide before launch how they will be read. The comparison of view-through and click-through conversions recommends reporting click-through conversions as the default, view-through on a separate line, and an incrementality test to decide what the views are worth. Confirm the primary action is the outcome the campaign should bid on and the rest are secondary, and expect analytics to show fewer conversions than Google Ads, because an unclicked impression leaves no session.
Launch and review placements in the first week
Enable the campaign and read the placement report daily for the first days, because the worst waste appears early and is cheapest to remove then, as the low-quality placements diagnosis advises. Sort by cost, then by impressions with no engagement, then by any name that is unfamiliar, and set the top rows against analytics engagement. Exclude what fails, adding it to the shared list so every campaign benefits. Read the audience report too: if most spend sits under optimised targeting rather than the chosen segments, the campaign is no longer the one that was planned. Check the delivered frequency against the cap.
Set a review cadence
After the first fortnight, move the reviews to a fixed rhythm: the diagnosis page suggests weekly for large budgets and monthly for small ones. Each review follows the same order: placements by cost, the audience report, frequency, the asset report, then results against the measure written in step one. Refresh creative on a schedule rather than only when the asset report complains, because remarketing audiences see the same ads many times. Give the review a named owner and a calendar slot, and plan an incrementality test once the campaign has run long enough to make one worthwhile; the display hub describes conversion lift studies for eligible accounts and geographic holdouts.
The decisions in steps one to four depend on each other. The table shows the usual combinations for the three common jobs.
| Job | Bid strategy family | Targeting route | Optimised targeting | Judge on |
|---|---|---|---|---|
| Reach and awareness | Impression-based, such as target CPM | Affinity or topics, one per ad group | Off unless bidding on conversions | Reach, frequency, cost per thousand impressions |
| Prospecting for action | Smart Bidding on the real outcome | In-market or custom segments | On, with tracking in place and exclusions set | Click-through conversions, view-through on its own line |
| Remarketing | Smart Bidding on the real outcome | Your own lists, converters excluded | Off | Conversions under a frequency cap; holdout for incrementality |
An account that later outgrows Google Ads display, because the plan needs publisher deals or inventory beyond Google's network, can follow the tutorial on moving display campaigns from Google Ads to DV360, which rebuilds these same decisions as line items, Floodlight lists and sized creatives rather than copying them.
A lighting shop's first display campaign (illustrative)
Lanternrook Lighting launches a remarketing campaign
Lanternrook Lighting, an invented online lighting retailer, launches a display remarketing campaign to people who viewed a product page in the last 30 days but did not buy, about 26,000 people, with a budget of £40 a day on maximise conversions. Purchasers in the last 180 days are excluded, optimised targeting is off, the account's shared exclusion list is applied, and the frequency cap is set at three impressions per person per day. Ads use product photography in native square and landscape versions, with headlines about free returns and next-day delivery.
| First fortnight | Days 1 to 7 | Days 8 to 14 |
|---|---|---|
| Spend | £280 | £280 |
| Impressions | 61,000 | 58,000 |
| Share of spend on mobile game apps | 16% | 4% |
| Average frequency per person per week | 6.1 | 4.7 |
| Click-through conversions | 9 | 14 |
| View-through conversions (separate line) | 22 | 19 |
The day-three placement review finds 31% of the first three days' spend (about £37 of £120) on puzzle and casual game apps sending visitors who leave instantly. The team excludes those app categories and 40 named sites, adding them to the shared list, and lowers the cap to two a day because the weekly frequency is higher than planned. Only 4% of the £160 spent on days four to seven still reaches game apps, which brings the first week's share down to 16%. In the second week click-through conversions rise from 9 to 14 at the same spend, a cost of £20 each against £31. View-through conversions are reported beside them but not added to the total, and a regional holdout is planned for the following quarter.
Common mistakes
- Two jobs in one campaign. Prospecting and remarketing blended, and neither can be judged.
- Maximise clicks on display for a sales goal. The cheapest clicks are in apps, and few of them buy.
- Optimised targeting left on by default. The audience report then shows most spend outside the chosen segments.
- Exclusions added after the first invoice. The waste happens in the first days and is cheapest to prevent.
- No frequency cap on remarketing. A small list and a steady budget mean the same person all day.
- One image in one shape. Every other slot shows a crop that removed the point.
- View-through conversions added to the total. The campaign looks better than any holdout will confirm.
- No placement review in the first week. The worst placements are found months later in a budget review.
