First, read the placement report by spend and conversions
Open the placement report for the period the problem covers and sort it by cost. For each placement in the top of the list, note four things: impressions, clicks, conversions, and what the analytics property says about the visitors it sent, such as engaged sessions or pages per visit. A placement can look busy in Google Ads and be empty in analytics. Then group the waste into kinds: mobile apps, especially games; sites that exist to carry ads; parked or unfamiliar domains; content that does not suit the brand. Each kind points to a different control.
Next, open the audience report for the same period and note how much of the spend went to optimised targeting rather than to the segments that were chosen. Finally, check the campaign's bid strategy and the conversion actions it optimises for. With those three readings, placements, audiences and bidding, the table below usually settles which cause applies. The targeting options themselves are catalogued in the display targeting guide; this page uses them as levers.
Diagnosis table
Diagnosis
| Symptom | Check | Likely cause | Fix |
|---|---|---|---|
| Most spend sits under optimised targeting, on placements unrelated to the chosen segments | Audience report: share of spend outside chosen segments | Optimised targeting widening delivery beyond the chosen segments | Turn it off for audience-defined campaigns, or keep it with tight exclusions |
| Ads follow a good audience onto poor pages and apps | Targeting tab: audience only, no content layer; placement report | Audience-only targeting following people onto any page | Add exclusions; for context-sensitive campaigns, add a content layer |
| Waste in obvious categories that a basic list would have blocked | Account and campaign exclusion lists; content exclusion settings | No placement, topic or content exclusions | Build account-level exclusion lists and set content exclusions |
| Cheap clicks or easy conversions concentrated on low-value placements | Bid strategy; which conversion actions are primary | Automated bidding chasing cheap clicks or conversions on low-value inventory | Optimise towards the real outcome; stop bidding on clicks for direct response |
| High click-through on apps and native slots with no engagement after the click | Placement report against analytics engagement; ad format by placement | Responsive formats served where accidental taps are common | Exclude the app categories; use uploaded ads for the slots that matter |
| Nobody can say where the budget went last month | Date the placement report was last reviewed; exclusion list history | Placement report never reviewed | Schedule reviews; feed every review into the exclusion list |
Fix, verify and prevent each cause
Expansion beyond the chosen segments
Optimised targeting is on by default for new Display campaigns and treats the chosen segments as a hint rather than a boundary, looking for people it predicts will convert. In a campaign with weak conversion data, or one bidding on clicks, it often finds them on cheap inventory.
Fix: decide what the campaign is for; for remarketing, a briefed audience or any campaign defined by who it reaches, switch optimised targeting off, as the display targeting guide recommends; for a conversion campaign with sound tracking, keep it on but add category and placement exclusions. The targeting help page describes how expansion and exclusions interact.
Verify: the audience report's share of spend outside the chosen segments falls, and the placement report's top rows change.
Prevent: make the optimised targeting setting an explicit line in the launch checklist, never a default.
An audience that follows people anywhere
Audience targeting is eligible wherever the person goes on the Google Display Network (GDNGoogle's network of display ad placements.), so a well-chosen in-market segment can still end up beside a game, a puzzle app or a content farm the person happened to visit. The audience was right; the pages were not.
Fix: add the exclusions that remove the worst places, and for campaigns where context matters, run an ad group that layers a topic or managed placements over the audience, accepting less reach for cleaner pages, as the layering table in the display targeting guide sets out.
Verify: the share of spend on excluded kinds of inventory falls, and conversions per pound spent hold or improve.
Prevent: keep one idea per ad group so the placement and audience reports can show which combination produced the waste.
No exclusions in place
A campaign launched without exclusions inherits the whole network. Mobile app categories, sensitive content, parked domains and sites that exist to carry advertising are all eligible until something says otherwise.
Fix: create account-level exclusion lists so every campaign inherits them, add app categories where apps carry most of the waste, set the content exclusions and inventory type that match the brand's tolerance, and add topic exclusions for subjects the brand will not sit beside; the display hub calls this placement hygiene and treats it as the most valuable routine in the channel.
Verify: newly launched campaigns show the exclusions applied, and the first week's placement report contains none of the excluded kinds.
Prevent: attach the shared lists to campaign templates so nobody launches without them.
A bid strategy rewarded for the wrong thing
Bidding does what it is told. A campaign on maximise clicks buys the cheapest clicks available, which on Display often means apps and low-attention sites. A campaign on a conversion strategy whose primary action is something easy, such as a page view or a newsletter tick box, finds the inventory that produces those easy actions. The smart bidding strategies guide notes that the model treats every included action as equally worth chasing unless values distinguish them.
Fix: for direct response, bid on the conversion that matters, with secondary actions set as secondary, and add values where outcomes differ in worth; the Smart Bidding help page describes what the strategies optimise.
Verify: cost per click (CPCAverage cost of each click.) may rise while cost per real conversion falls, and the placement mix shifts towards sites that send engaged visitors.
Prevent: review the primary conversion actions whenever a Display campaign is created or its goal changes.
Responsive formats in slots where taps are accidental
Responsive display ads reshape themselves to fit almost any slot, including native and in-app placements that fixed-size ads cannot reach. That breadth is the format's strength and, in some apps, its weakness: an ad near a game's controls collects taps nobody meant. The report shows a high click-through rate (CTRShare of impressions that result in a click.) and an analytics record of visitors who leave instantly.
Fix: exclude the app categories and specific apps responsible, and for the slots that matter most, run uploaded image ads in the sizes you choose alongside the responsive ad, as the responsive display ads guide describes; the responsive display ads help page documents how the format is assembled.
Verify: click-through falls on the affected inventory while engaged sessions per click rise.
Prevent: judge creative by what happens after the click, never by click-through alone.
A report nobody opens
Many low-quality placement problems are not caused by any single setting but by time: a campaign that ran for months with nobody looking at where it ran. Inventory changes, new apps appear, and the exclusion list never grows.
Fix: read the placement report now, export the poor placements into the shared exclusion list, and schedule the next review, following the reading order in the display targeting guide: cost, then impressions without engagement, then unfamiliar names.
Verify: the exclusion list shows a dated addition from each review, and the share of spend on placements with no engaged visits falls review by review.
Prevent: give the review a named owner and a recurring calendar slot.
Prevention: a placement review cadence
Placement quality is maintained, not configured once. The routine has three tempos. At launch, every Display campaign starts with the shared exclusion lists, content exclusions set to the brand's standard, an explicit decision on optimised targeting, and a bid strategy pointed at a real outcome. In the first days, someone reads the placement report daily, because the worst waste appears early and is cheapest to remove then. After the first fortnight, the review moves to a fixed rhythm, weekly for large budgets and monthly for small ones. The display campaign launch tutorial builds a new campaign in that order, from its single job and bid strategy to the first-week placement review.
Each review follows the same order. Sort placements by cost and check the top rows against analytics engagement. Look at impressions with no engagement and at names nobody recognises. Check the audience report's share of spend outside the chosen segments. Add what fails to the account-level list, not to one campaign, so every campaign learns from every review. Record the date and the number of additions, so the history shows whether the list is still growing fast, which suggests an upstream setting is wrong, or slowing, which suggests the campaign is clean. When a review keeps finding the same kind of waste, fix the setting that admits it rather than excluding one placement at a time.
A home insurance broker finds its budget in puzzle games
A home insurance broker runs a prospecting Display campaign on £4,000 a month, targeting an in-market segment for home insurance with optimised targeting on and maximise clicks as the bid strategy. The campaign reports a 1.9% click-through rate and a £0.21 cost per click, but only 6 quote requests a month. The placement report shows 64% of spend on mobile apps, mostly puzzle and casual games, with an average engaged session rate of 3% in analytics. The team excludes the games app category and 140 named placements at account level, turns optimised targeting off, sets content exclusions to the standard inventory tier, and switches to maximise conversions on the quote-request action. In the following month cost per click rises to £0.68, click-through falls to 0.4%, and quote requests rise to 31.
Common mistakes
Building exclusions one campaign at a time. Leaving optimised targeting on for a remarketing campaign. Judging placements by click-through. Running maximise clicks on a direct-response Display campaign. Reviewing the report once at launch and never again. Excluding placement after placement while the setting that admits them stays unchanged.
