Google and Meta Ads are increasingly pushing towards automated campaigns, where the algorithm decides targeting, bids, and often even the creatives. For those managing significant budgets, this isn't a future problem: it's already the present.
The real risk isn't that AI will spend poorly, but that it will spend opaquely. Without clear governance structures, the budget ends up being optimized towards metrics the platform prefers, not towards those that matter for your business.
Three levers allow you to stay in control: transparent reporting that shows where the money is really going, creative guardrails to protect the brand, and campaign structures designed to preserve performance without handing everything over to the algorithm. The article explains how to apply them operationally.
The real problem: the algorithm optimizes for itself, not for you
When you turn on Performance Max on Google or Advantage+ on Meta, you are handing over decisions to an automated system that you used to make yourself: who to show the ad to, how much to pay per impression, which creative variant to use. The platform optimizes for its own conversion goals, which often match yours, but not always.
The most common warning sign: the cost per acquisition drops on the dashboard, but actual sales don't budge. This happens because the algorithm can learn to convert users who would have bought anyway, inflating the numbers without generating new demand. It is one of the classic cases analyzed when talking about advertising attribution and models that deceive the budget .
The right question isn't 'should I use automation?' — it's almost mandatory now. The question is 'how do I govern it?'
Transparent reporting: looking inside the black box
Platforms show metrics by default that make them seem more effective. To get a real picture, at least three levels of analysis are needed:
- Placement breakdown (breakdown by placement): where the ad actually appears — Search, Display, YouTube, network partners. Often the largest share of the budget ends up on poorly qualified inventory.
- Search term report (report on real queries): which keywords triggered the ads. With automated campaigns, this data is often hidden or aggregated.
- Incrementality : how many conversions would have happened without the campaign. Without this data, you're measuring correlation, not causation.
If the platform doesn't provide this data in a readable format, export and analyze it outside the system. An external spreadsheet comparing actual spend and sales is worth more than ten native dashboards.
Creative constraints: protect the brand before AI decides on its own
Performance Max and Advantage+ automatically generate creative variations, combining the headlines, descriptions, and images you provide. The risk is that the algorithm might choose combinations that perform well for clicks but don't align with your brand's tone or values.
Practical measures to take immediately:
- Provide only internally approved assets (images, texts, videos), never let AI generate content autonomously if the platform allows it.
- Explicitly exclude sensitive placement categories: polarizing news sites, game apps, inventory for minors.
- Set fixed final URLs for automated campaigns, so the algorithm cannot redirect traffic to pages other than those chosen.
- Review weekly the ad combinations that the platform has favored, and disable those that are off-brand.
It's also worth keeping an eye on the topic of bot traffic and AI eating into ad budget : some of the automated spend ends up on non-human impressions, and placement constraints also help reduce this waste.
Campaign structure: where automation can enter and where it cannot
Not all campaigns need to be automated in the same way. A hybrid structure allows you to leverage AI where it's effective and maintain manual control where the risk is higher.
Practical plan for an SME with a medium budget:
- Manual brand campaign : campaigns on your own brand name always remain under direct control, with exact keyword matching and manual bids. The algorithm must not touch them.
- Automated core campaigns : main products or services can use Smart Bidding (automatic bids based on conversion goals), but with explicitly set target CPA or ROAS, not left free.
- Performance Max in isolation : if you use it, keep it separate from other campaigns and monitor that they do not cannibalize brand traffic. This is one of the most discussed aspects in analyses on Google Ads Local Customer Optimization and Performance Max for retail .
The logic is simple: the more strategic or brand-sensitive a campaign segment is, the less automation it should have.
What to do this week, concretely
If you manage budgets on Google or Meta Ads, three immediate actions are worth more than any long strategic review:
- Download the search term report for the last 30 days and check how many irrelevant queries received budget.
- Check active placements in automated campaigns and add exclusions where needed.
- Compare the conversions reported by the platform with the data from your CRM or e-commerce: if the gap is greater than 20%, you have an attribution problem to solve before scaling your budget.
For those who want to delve deeper into the topic of advertising automation in a broader context, including emerging opportunities on new channels, it's worth reading the analysis on Amazon DSP and ChatGPT Ads for Italian B2B : the governance described here also applies to those formats.
The original in-depth analysis on this topic is available on Search Engine Journal . All the news on advertising and paid media are collected in the area Advertising, Paid Media & Video by SHM Studio.
Related articles
Discover more articles exploring similar topics, selected to offer you a more complete and stimulating perspective. Each piece of content is carefully chosen to enrich your experience.
