If your Demand Gen campaign is not running, spending slowly or experiencing a sudden drop in traffic, it could be due to a technical issue.
Google Ads AI requires specific parameters to enter auctions effectively.
This article will help you to troubleshoot the issues with Demand Gen campaigns. Follow the steps below to identify and resolve issues to improve your campaign performance.
On this page
Troubleshooting steps
Step 1: Initial launch and system delays (24–48 hours):
If the campaign is newly launched, or is a previously paused campaign that has been re-activated, make sure that you verify the below requirements:
- Wait for propagation: It generally takes up to 24 hours (though often under 12 hours) for new campaigns or major changes to be processed and synced across Google's serving infrastructure before impressions appear in the UI.
- Check for creative disapprovals: Ensure that your video, image and text assets are in approved status. It is important to note that Demand Gen maintains a stricter creative quality threshold compared to standard video or Performance Max campaigns. Verify your asset status as follows:
- If an ad is flagged as 'Low quality' or has a policy rejection, use the Self-Service Appeal (SSA) tool directly in the Google Ads UI to appeal the decision.
- Check for audience incompatibilities and too broad or too narrow targeting: To ensure that your campaign serves, avoid using first-party data, custom segments or lookalike segments, if your content falls under a sensitive category, as personalised advertising policies restrict these audience types.
- Audit product feed eligibility (if applicable): If you've attached a Google Merchant Center (GMC) feed, a feed misconfiguration can completely halt campaign serving:
- You must have at least four approved, distinct and in-stock products to serve. If stock drops below four, the feed will stop showing. Google recommends having at least 50 approved products.
- Make sure that products are selected from at least four different group IDs (e.g. not just the same shoe in four different sizes).
- Ensure that your country-level campaign targeting matches the country location of your GMC products.
- Other reasons for campaigns not running include:
- Incompatible bid strategies: Incompatible targeting, such as Discover and Apple devices.
- Google signals not enabled in Google Analytics: When targeting a Google Analytics list, if the advertiser has not turned on or enabled Google signals, then the list won't serve along with the ad group.
- Video asset not public: If an advertiser adds a video asset to their campaign targeting YouTube inventory but the video asset is marked as private or unlisted, then the ad won't serve because the video ad is not public.
Step 2: Addressing bid and budget limitations
The primary reason for low delivery in over 80% of Demand Gen campaigns is budget constraints, which restrict the AI's bidding capacity. If budgets are too low or bids are overly conservative, the campaign may fail to serve.
- Ease aggressive bidding: If you are using Target CPA (tCPA) or Target ROAS (tROAS) bidding strategies and have noticed no spend, your targets may be too restrictive for these mid-funnel formats.
- Action: Increase tCPA or decrease tROAS by 20–30% to improve auction competitiveness. Aim for targets roughly 30% looser than Search or Performance Max benchmarks, or match your Meta/social clickthrough targets.
- Adhere to the '10x rule' for budgets: Ensure that the campaign has sufficient budget to achieve the 50+ conversions needed to clear the learning phase and avoid a 'cold start’.
- Target CPA (tCPA)/Target ROAS (tROAS): Your daily budget should be at least 10 times your tCPA (e.g. a $15 tCPA requires a minimum $150 daily budget).
- Maximise conversions: Maintain a minimum daily budget of $100.
Step 3: Navigating audience and targeting limitations
Highly restrictive targeting parameters often limit the reach of Smart Bidding algorithms, resulting in diminished traffic or a total cessation of spend.
- Activate optimised targeting (OT): For conversion-oriented campaigns with narrow segments – such as small first-party lists – serving may be limited. Enable optimised targeting at the ad group level to permit Google’s AI to identify high-intent users beyond your initial inputs.
- Address lookalike and NCA goal conflicts: Avoid combining new customer acquisition (NCA) goals with lookalike segments in a single campaign. Due to divergent algorithmic logic, layering these features can drastically reduce reach.
- Action: Select only one of these strategies.
- Opt-in to 'Unknown' demographics: If you have unticked 'Unknown' age or gender categories, it excludes a significant portion of signed-out or untrackable mobile/iOS traffic. Keep 'Unknown' demographics enabled. If strict age/gender compliance is required, use the UI setting 'only show ads to people within my age and gender specifications' inside the optimised targeting module instead of excluding them manually.
Step 4: Monitoring recent modifications and the learning phase
- Review recent campaign edits: Significant adjustments within the last 72 hours, such as reducing budgets, increasing ROAS targets or comprehensive creative refreshes, will reset the campaign’s learning phase.
- This reset typically leads to performance fluctuations and temporary delivery drops as the system recalibrates.
- Action: Allow 5–7 days (or until 50 conversions are reached) for stabilisation before making further assessments. Ensure that future bid or budget changes remain within a +/-15% range to minimise disruption.
