AI in Meta Ads and Google Ads: the new era of media planning and optimization

Machine learning has been part of digital advertising for a long time. Google and Meta have used predictive models for years to estimate the likelihood of clicks, conversions, purchases, views, and many other actions. What is new is that these systems can interpret more and more signals, evaluate more combinations, and make real-time decisions at a scale that is impossible to replicate manually.

Previously, a media buyer could create multiple ad sets, separate audiences, choose placements, control bids, and move budgets between campaigns. Many of those tasks can now be performed or assisted by artificial intelligence. The goal should no longer be to compete with the algorithm by trying to manually make thousands of decisions a computer can evaluate in seconds.

The media professional's role is increasingly shifting toward defining objectives, setting budgets, structuring conversions, supplying good signals, establishing controls, interpreting results, and deciding what we want the technology to try to achieve. The foundations of Google AI Essentials focus precisely on preparing the measurement, first-party data, and objectives that power automation.

The Meta Advantage+ family of solutions represents much of this transition within Meta Ads. Advantage+ automates various components related to audiences, budgets, placements, bidding, and delivery, using machine learning models to find conversion opportunities within Meta's ecosystem.

For years, a typical Facebook and Instagram Ads strategy might involve creating multiple manual audiences based on interests, ages, behaviors, lookalikes, remarketing, and different targeting combinations. That level of control may still be necessary in certain scenarios, but Meta's current systems are designed to work with a much broader range of signals and find opportunities we might not have identified manually.

The result is a shift in focus. Instead of concentrating exclusively on building twenty different segments, it is increasingly important to ask whether the objective is properly defined, whether we are sending good conversion signals, whether there is enough budget for the system to learn, whether we are measuring the right action, and whether we are giving the algorithm enough flexibility to find opportunities.

One of the most interesting developments within Meta's advertising system is Andromeda. It is not a button inside Ads Manager or a new campaign type an advertiser can activate. It is part of the Artificial Intelligence infrastructure Meta uses to decide which ads should initially be considered before determining which ones will ultimately be shown to a person.

Meta describes Andromeda as its proprietary machine learning system for ads retrieval. The retrieval stage is the first filter within the ad recommendation system: it narrows tens of millions of possible ads down to a few thousand relevant candidates. More complex ranking models then analyze those candidates and determine which ones are most likely to generate value for the user and the advertiser.

Andromeda uses deep neural networks, hierarchical indexing, and specialized hardware to handle a much larger number of possibilities and detect more complex relationships between people, products, services, and ads. From a marketing standpoint, the implication is straightforward: Meta is increasing its ability to select, from an enormous number of possible ads, those that may be most relevant to a specific person at a particular moment.

Andromeda does not require advertisers to activate a specific setting, but it does change how we should think about strategy. Meta explains that one of the challenges this system aims to solve is precisely the exponential growth in advertising options generated by Advantage+ and other automation tools.

This means the platform is equipped to evaluate a greater variety of ads, audiences, products, and signals. From an agency's perspective, the work is starting to shift from telling the algorithm exactly who should see each ad to giving it good objectives, conversions, budgets, signals, and constraints so it can identify the best opportunities.

This does not eliminate audience strategy. It transforms it. Professionals still need to understand who the consumer is, what they need, what result we want, and what information we should provide to the system so automation works in the business's favor.

One of the main concerns when we talk about automated campaigns is the idea that the platform will simply spend money without oversight. In reality, advertisers continue to define budgets, objectives, limits, locations, exclusions, and various campaign settings.

What changes is the number of granular decisions the platform can make within those parameters. Meta can identify opportunities across different audiences and placements, while Google can adjust bids in each auction based on the likelihood of conversion and the estimated value of that conversion.

The conversation about digital advertising and media buying is no longer just about how much we want to spend. We also have to define what result that budget should produce, how much we are willing to pay for a conversion, how much that conversion is worth, whether we want to maximize volume or value, and what CPA or ROAS limits make sense for the business.

Performance Max represents one of the most comprehensive forms of automation within Google Ads. It is a goal-based campaign type that provides access to inventory across Search, YouTube, Display, Discover, Gmail, and Maps from a single campaign. Google uses audience signals, conversion goals, budgets, assets, and Smart Bidding to decide where opportunities exist to generate conversions or value.

Performance Max does not mean all traditional campaigns should disappear. Google describes it as a tool that can complement keyword-based Search campaigns, especially when the goal is to find additional conversion opportunities across different channels.

The quality of the results will depend largely on the signals we provide. If all conversions are configured with the same value, Google will optimize on that basis. If a low-quality call and a high-value sale count exactly the same, the platform cannot know the difference unless we communicate it through data and settings.

AI optimizes what we configure, not necessarily what we had in mind.

Behind many Google campaigns is Smart Bidding, which uses machine learning to automatically adjust bids based on the likelihood of conversion or conversion value in each auction. This allows the system to consider signals that would be virtually impossible to process manually at that speed.

The strategic importance lies in correctly defining the result we want to achieve. Depending on the business, we can work with objectives such as Maximize Conversions, Maximize Conversion Value, Target CPA, or Target ROAS. Automation can execute the tactical decision of how much to bid in a specific auction; the agency and the advertiser remain responsible for determining what financial outcome we want to pursue.

Google is taking this approach a step further with AI Max for Search campaigns. It is important to clarify that AI Max is not a new standalone campaign type, but an optimization layer activated on existing search campaigns. Google describes it as a suite of features that uses real-time signals to optimize matching, targeting, and delivery.

One of its most important components is the ability to find relevant searches beyond traditional keywords. Google AI Max can combine broad match with asset-based technologies and keywordless systems to identify queries the advertiser may not have anticipated manually.

This becomes especially important as searches grow longer, more conversational, and more complex. A person no longer necessarily types “advertising agency Puerto Rico.” They might search for “agency that can manage Google Ads and social media to increase inquiries for my business.” In that scenario, understanding intent may be more important than relying exclusively on literal matches, a shift that also connects with the evolution of SEO and AI Search.

Automation does not necessarily mean giving up control or transparency. Google has incorporated features into AI Max related to brand controls, locations of interest, URL inclusions and exclusions, search term reporting, and asset reporting.

It can also use Final URL Expansion, where Google selects a website page it considers more suitable for a particular search, always subject to the settings and exclusions defined by the advertiser. This directly connects media and web development.

If landing pages are outdated, poorly organized, or do not accurately represent the business's services, the platform will have weaker options to choose from. Advertising automation works better when the digital infrastructure supporting it is also well structured.

Google has also incorporated Gemini-powered conversational experiences into Google Ads. These tools can help analyze campaigns, answer questions, suggest optimizations, identify problems, and assist with various setup and creation tasks.

Google describes Ask Advisor as a conversational experience built with Gemini capabilities that can interpret account-specific questions, analyze performance, help identify problems, and propose changes that require advertiser approval. An advertising platform is no longer just an interface with tables and metrics; it is beginning to incorporate assistants capable of interpreting some of that information through natural language.

The responsibility to review, interpret, and decide remains human.

This is probably one of the most important principles of this entire transformation. Platforms do not read minds. They learn from the data and objectives we provide. If we want to optimize for quality leads, we need to define what a quality lead means; if we want sales, we have to measure those sales correctly; and if we want to maximize value, we need to identify which conversions generate the most value for the business.

This makes elements like Google Analytics, conversion tracking, enhanced conversions, Meta Pixel, Conversions API, CRM, and first-party data even more important. An intelligent algorithm cannot make up for a poor measurement implementation.

We can have extremely sophisticated technology perfectly optimizing the wrong objective.

A campaign that uses Artificial Intelligence should not be turned on and forgotten. Budgets change, promotions change, competition changes, landing pages change, and business objectives change too.

Platforms need data and time to learn, but they also need supervision. The difference lies in the type of supervision: less and less micromanagement of small decisions, and more and more analysis of strategy, objectives, signals, profitability, and the quality of results.

Artificial Intelligence is making media, analytics, and strategy even more interconnected. Media professionals should no longer limit themselves to looking at CTR, CPC, or CPM. They also need to understand conversions, lead quality, revenue, Cost per Acquisition, Return on Ad Spend, margins, customer value, and attribution.

The question is no longer just how much the click cost, but how much value the investment produced. This shift is especially important because automated systems need clear financial objectives. If a business can assign different values to different conversion types, the platform can use that information to optimize in a way that better aligns with its actual priorities.

Meta and Google are different ecosystems. Meta focuses heavily on discovery, content, behavior, and social environments, while Google continues to hold a unique position around intent, Search, YouTube, and its broad ecosystem of properties.

However, both platforms are evolving in a similar direction: more Artificial Intelligence within targeting, bidding, budgets, delivery, ad selection, and optimization. Meta uses Advantage+ and internal systems like Andromeda. Google uses Performance Max, AI Max, Smart Bidding, and Gemini-powered tools.

The names and interfaces change, but the logic is similar: give the system enough information and flexibility to find opportunities that would be difficult to identify manually at the same speed and scale.

Creative generation is also part of this evolution, but from our perspective, it should be viewed as a complementary capability rather than the center of the strategy. Meta and Google continue to incorporate features capable of generating, adapting, or extending advertising assets. These tools can be useful for developing variations, adapting formats, exploring alternatives, and accelerating certain parts of production.

However, just because a platform can generate an image does not mean it should automatically replace professional art direction, design, photography, video, or copywriting. A brand needs identity, judgment, consistency, cultural context, and creative direction that accurately represents what it wants to communicate.

At Alterno, we continue to believe in working with designers, photographers, videographers, editors, and other creative professionals. We see generative AI as an additional tool within the process, not as an excuse to eliminate creative judgment.

For years, a significant part of digital media operations involved manually handling targeting, bids, placements, budget shifts, combination testing, and daily adjustments. Many of those tasks are gradually being automated. The work does not disappear; it moves to a different level.

An agency increasingly needs to focus on understanding the true business objective, which conversion we want to optimize, how much that conversion is worth, what budget makes sense, what our CPA or ROAS limits are, what data we are sending to the platform, and how to interpret the results we receive. We also have to decide which parts we want to automate and which require human control.

That work is more strategic than simply moving budgets around every day.

The platform can make thousands of tactical decisions, but it should not define the marketing problem on its own. Automation works best when it starts with a clear strategy and well-informed human decisions:

• It can decide where to show an ad, but someone must define the marketing problem.
• It can optimize for a conversion, but someone has to decide which conversion matters.
• It can move budgets, but someone must determine how much we are willing to invest.
• It can find new audiences, but someone needs to understand whether they make sense for the business.
• It can generate an image, but someone must determine whether it accurately represents the brand.
• It can produce results, but someone has to interpret them.

Artificial Intelligence can make better tactical decisions when we make good strategic decisions.

At Alterno, we have been running digital media campaigns since long before the word AI appeared in every marketing presentation. We have watched Google Ads, Facebook Ads, Meta Ads, YouTube, programmatic advertising, retargeting, attribution, and analytics evolve. Now we are entering another stage, and our approach is not to compete with automation, but to learn how to use it correctly.

That means combining strategy, objectives, budgets, technology, measurement, optimization, and human judgment. We use the platforms' Artificial Intelligence tools when they can help identify opportunities, allocate investment, find audiences, optimize campaigns, or analyze results.

At the same time, we continue to believe that strategy, critical thinking, and professional creativity remain human responsibilities. A platform can decide how to better allocate a budget, but first someone has to decide why we are investing that budget and what result truly matters to the business.

Perhaps that is the most important transformation AI is bringing to digital advertising. It is not simply about machines doing more tasks, but about how they push us to focus more on what requires judgment: defining the problem, setting the objective, measuring the right things, and interpreting results to make better decisions.