Check out part 1 linked below ⬇️

Key Concepts

• Sequential Learning: Meta’s system for determining the optimal ad sequence for individual users to maximize conversion rates.
• GEM (Generative Engine for Meta): Meta’s proprietary Large Language Model (LLM) used for ad delivery, predicting user actions, and optimizing ad serving.
• Creative Diversity: The practice of utilizing a wide range of ad formats, messaging, and subjects to cater to individual user preferences.
• Budget Liquidity: Allocating budget at the campaign level to allow Meta’s algorithms to dynamically distribute funds to the best-performing ad sets and creatives.
• Consolidated Ad Account Structure: A structure where a single campaign houses multiple ad sets, each containing diverse ad formats, with budget controlled at the campaign level.

Sequential Learning & GEM: The Future of Ad Delivery

The core of Meta’s advertising evolution in 2026 revolves around two key systems: sequential learning and GEM. Sequential learning is described as the mechanism by which Meta determines the optimal order in which to serve ads to individual users. The system recognizes that each person responds differently, making manual ad sequencing ineffective. The goal is to present ads – awareness, consideration, purchase, repurchase, upselling – in a sequence most likely to drive conversion for that specific user. This necessitates the creation of creatives designed for each stage of the customer journey.

GEM (Generative Engine for Meta) represents a further advancement. It’s positioned as Meta’s own Large Language Model (LLM) dedicated to ad delivery. GEM analyzes all available data points – user information and advertiser specifics – to predict user behavior and optimize ad serving, including timing and sequence. Essentially, GEM aims to anticipate the next action a user will take.

The Importance of Creative Diversity

Given the capabilities of sequential learning and GEM, creative diversity is identified as the most important aspect of ad creation in 2026. This isn’t simply about having different images; it’s about employing a variety of ad formats, messaging, and subjects. The rationale is that Meta’s system is designed to deliver the right ad to the right person at the right time and in the right order, and this requires a diverse creative pool to draw from. A lack of diversity limits the system’s ability to personalize the ad experience.

Structuring Ad Accounts for Budget Liquidity

To effectively leverage these systems, a specific ad account structure is recommended: a consolidated approach focused on budget liquidity. This involves a single campaign containing multiple ad sets. These ad sets can be segmented based on either personas (target audience profiles) or stages of awareness (e.g., initial awareness, consideration, purchase intent).

Within each ad set, a diverse range of ad formats should be utilized. Examples provided include: podcast videos, founder videos, testimonials, screenshots, and comparison videos. Crucially, the budget is allocated at the campaign level, not the ad set level. This allows Meta’s algorithms to dynamically distribute funds to the ad sets and individual ads that are performing best, and to determine the optimal ad sequence for each user. The speaker emphasizes that this approach allows Meta to identify which “angles” and which specific ads resonate most effectively.

Segmentation Strategies

The transcript highlights flexibility in how ad sets are segmented. While personas are one option, segmentation can also be based on stages of awareness or specific pain points. The key takeaway is that the segmentation strategy should support the delivery of diverse creatives and allow Meta’s algorithms to optimize performance.

Looking Ahead

The speaker concludes by noting that this is part two of a three-part series, encouraging viewers to follow for the final installment.

LLM (Large Language Model): A type of artificial intelligence that uses deep learning algorithms to understand, summarize, generate, and predict new content. GEM is Meta’s LLM specifically designed for ad delivery.

Persona: A semi-fictional representation of your ideal customer based on market research and data about your existing customers.