How to reverse-engineer competitor Performance Max asset groups

How can e-commerce operators extract actionable intelligence from an advertising channel that hides its own mechanics? While Google Performance Max obscures individual asset performance behind dynamic placements, growth teams can reverse-engineer competitor campaigns by systematically cataloging long-running visual elements. By deploying Notch, an AI-powered creative ad engine, performance marketers can translate those isolated visual formats into structured, on-brand variations ready for Meta and TikTok. This methodology strips the mystery from Google's automated systems, letting you deploy optimized creative variants based on your competitor's historical testing budget.

Our team of ex-Meta performance marketers built Notch to address the exact structural inefficiencies that keep high-growth brands trapped in slow testing cycles. Across the 5,000+ brands and agencies using our creative ad engine, we consistently see that scaling paid acquisition requires treating ad creative as structured data rather than unpredictable art. This guide outlines the precise workflow used by elite growth teams to systematically strip away the algorithmic opacity of Google's flagship ad product.

The PMax visibility problem

Most digital advertisers continue to analyze competitor activity using frameworks built for a legacy search environment. Traditional keyword research platforms rely on search engine scraping models that map specific search queries directly to static ad copy. This approach breaks down completely when confronting Performance Max (PMax), which now runs on roughly 73% of Google advertiser accounts according to a 2026 Brandsearch study. Because PMax does not rely on simple keyword auctions, standard competitor tools are functionally blind to its volume.

Traditional options like Auction Insights only provide macro metrics like overlap rate and impression share. They tell you that a competitor is present in the market, but they fail to show the actual creative combinations driving their customer acquisition. PMax operates as a programmatic matching system, taking headlines, descriptions, images, and video assets to assemble millions of distinct ad permutations in real time. The final ad only exists at the exact moment of impression, leaving no static ad unit for external platforms to scrape.

This architectural shift has turned modern ad creative into a software engineering problem. Google’s machine-learning algorithm distributes these dynamic assets across six channels simultaneously: Search, YouTube, Display, Discover, Gmail, and Maps. As detailed in a 2026 Segwise analysis, the system relies on an opaque feedback loop that rates asset performance merely as "Best," "Good," or "Low" relative to other assets in the same campaign. This leaves performance teams with a massive visibility deficit. To out-optimize the market, you must bypass the dashboard and analyze the raw creative inputs your competitors are feeding the machine.

Isolate the raw materials

To break through this black box, you must focus on the individual ingredients before Google's system mixes them. Our team at Notch recommends a systematic gathering process that targets the assets that have survived the algorithm's optimization process. This requires looking at the raw visuals rather than the dynamic text pairings. Use the following framework to organize your initial teardown:

• Filter competitor libraries to identify images or videos active for more than 90 days, indicating sustained profitability.
• Note if the top assets rely on user-generated demonstration, side-by-side comparison, or product-in-use closeups.
• Track the exact percentage of screen space occupied by static text overlays vs. raw visual action.
• Document whether the visual assets focus on bundle pricing, single-item discounts, or shipping guarantees.

Ad longevity is the most reliable proxy for performance in programmatic environments. As experienced media buyers know, an ad asset that stays active for three months or more is almost certainly profitable. Competitors will not waste ad spend on underperforming creative assets that Google's algorithm has deprioritized. You can map these long-running variables into a structured database to identify the exact visual boundaries that scale in your vertical.

To make this discovery process actionable, growth teams should build a competitor ad analysis template that actually predicts winners. Documenting these patterns allows you to ignore superficial production details and focus on the structural frameworks that convert. Once you have isolated these winning creative inputs, you can use our creative ad platform to systematically clone and iterate on them without manual video editing.

Extract the creative physics

Replicating a competitor's success is not about copying their exact footage or branding. Instead, you must isolate the underlying structural mechanics—what we call creative physics. This refers to the exact timing of visual cuts, the cadence of text overlays, and the psychological triggers used to capture attention in the first three seconds.

Deconstruct the layers

Every winning visual asset is composed of three distinct layers: the physical hook, the text overlay structure, and the pacing of the visual edits. In our analysis of top-performing direct-to-consumer accounts, the first three seconds dictate up to 80% of the asset's overall efficiency. If a competitor's ad uses a split-screen layout displaying a product problem on the left and a solution on the right, that structural choice is the physical hook. You must record these specific variables: the precise timestamp of the first cut, the presence of voiceover tracks, and the transition speed between product benefits.

Rebuild with intelligence

Once you have mapped these structural rules, you can bypass manual editing entirely. Instead of spending hours coordinating with creators or chopping clips in complex editing software, you can use Notch. Our Claude-powered AI engine allows you to drop a public competitor ad link directly into the workspace. The engine analyzes the visual pacing and hook structures, extracting the creative physics to build an optimized replica using your brand's own assets, logo, and brand guidelines. This workflow cuts production times from hours to minutes, allowing your growth team to stay focused on deployment.

To visualize the operational impact of this workflow shift, consider how Notch compares to traditional creative pipelines:

Feed the algorithm what it wants

Google's machine learning thrives on variety, but asset groups often starve from a lack of creative depth. When an asset group contains only a handful of static images and a single generic video, the matching algorithm quickly exhausts its testing variations. This asset limitation is where most PMax campaigns fail to scale, leading to rapid creative fatigue and climbing acquisition costs.

According to the official Google Ads API documentation, an asset group is a themed creative kit that requires a comprehensive suite of visual resources to perform. When you feed the machine a rich library of unique creatives, you provide the necessary raw material for Google's system to find profitable micro-segments. Using Notch, you can take a single reverse-engineered competitor concept and generate 20 to 40 unique variations in a single session.

Our platform ensures you never suffer from the "same faces" problem common among traditional AI generation tools. While other platforms recycle a limited library of 300 stock avatars, Notch generates unique variations for every brand, protecting your campaigns from visual fatigue. These assets—complete with varied hooks, unique visual layouts, and distinct text overlay structures—can be pushed directly to your Meta Ads Manager and Google accounts. This high-volume output allows you to continuously feed the algorithm with high-quality, structured inputs.

One Thing to Watch Out For

A common pitfall in competitor reverse-engineering is creating "frankenstein" assets that do not correspond with the broader conversion funnel. Many media buyers will clone a competitor's high-performing visual hook but connect it to an entirely unrelated landing page or offer. PMax evaluates the user journey as a single, cohesive system. If your visual asset promises a specific bundle or risk-reversal guarantee, your lander must immediately reflect that exact copy.

When the visual hook and the landing page copy do not match, conversion rates plummet and bounce rates spike. The algorithm quickly detects this drop in post-click engagement and will systematically deprioritize the entire asset group, regardless of how strong the individual ad components are. Before launching your newly generated creative variations from Notch, audit your destination URLs to ensure that the messaging, pricing, and visual style remain completely consistent.

Closing

Scaling modern performance campaigns requires moving past manual, creative guesswork. By systematically reverse-engineering the asset groups your competitors have spent thousands of dollars validating, you can build a highly predictable creative testing pipeline. Avoid the bottleneck of manual content creation and old, fragmented workflows. Drop a competitor's longest-running visual link into our platform, let the intelligence engine extract the physics, and generate your first batch of publish-ready variations at Notch today.