Fix the PMax Asset Gap That Costs You 12% Conversions

When you leave Performance Max video slots empty, Google automatically generates basic slideshows from your static assets. These templated videos hurt your brand and cost you performance, but producing custom video for every campaign eats your agency's margins.

To solve this bottleneck, agencies use generative AI models to build custom video, text, and image variations at scale. Supplying these AI-generated assets directly prevents Google from defaulting to automated slideshows, securing the 12% increase in conversions that advertiser-supplied videos deliver over system-generated ones.

Stop relying on default slideshows

When you launch a Performance Max campaign without video assets, Google does not skip video placements. Instead, the platform automatically generates video ads from your existing text, images, logos, and Merchant Center product feeds. While this prevents your campaign from being locked out of YouTube and Display inventory, the resulting outputs are basic, templated slideshows.

These automated assets present a serious brand risk. The system pairs generic taglines with standard product images, creating unpersuasive creative that blends in with competitors. More importantly, it costs you conversions. Internal Google data cited in Google Performance Max: Creating Video Assets with AI shows that campaigns running advertiser-supplied videos deliver an average of 12% more conversions than those relying on system-generated files.

You can identify these system-created videos by checking the Source column in your Asset Group details. If they say "Automatically created", you are losing performance to generic slideshows. Just as you regain control over your search terms by Automating Google Ads Negative Keywords With LLMs, you must take active control of your visual inputs to protect your client's brand.

Building video for every aspect ratio

Escaping the slideshow trap is difficult because Performance Max demands a high volume of video. Google recommends supplying at least five videos per asset group: two to three in landscape (16:9), two to three in portrait (9:16), and one or two square (1:1).

Producing this variety manually crushes agency margins. Instead, agencies use generative AI to construct compliant files in a fraction of the time. Text-to-video engines like Sora 2 and Veo 3.1 allow you to generate cinematic clips from direct text prompts across all three required aspect ratios. For e-commerce clients, image-to-video tools can animate existing static product photography, turning still shots into professional motion assets.

Audio is just as critical. Roughly 40% of Performance Max videos lack audio because they were exported as visual loops. To fix this, you can use Google's native text-to-speech models to synthesize narration from your existing headlines. According to Google Ads AI Voice-Over: Performance Max Video Guide, adding this AI-generated voice-over to silent videos drives an estimated 15% average view rate lift.

Scaling text and imagery in bulk

Performance Max is heavily dependent on the volume and diversity of its visual assets. According to How AI Visuals Are Transforming Google Display and ..., advertisers who supply enough varied text and image assets to achieve an "Excellent" Ad Strength rating see an average conversion increase of 6%.

To reach that rating without commissioning new photoshoots, agencies use tools like Google's Asset Studio. Powered by the Imagen 4 model, Asset Studio lets you automatically generate lifestyle photography from static product shots. You can upload a standard product photo on a white background, prompt the AI with a setting—like a winter landscape—and generate realistic lifestyle variations while preserving the physical details of the original product.

This workflow scales horizontally. The tool allows agencies to transform images in bulk, refreshing backgrounds and adding seasonal context for up to 100 photos simultaneously. For enterprise clients managing hundreds of locations, generative AI acts as a creative co-pilot. As noted in Advent Calendar Day 11: The Enterprise Creative Bottleneck, a national retailer can use generative AI in Google's Studio to automatically produce 200 localized ad variations from a single master template, inserting unique maps and store-specific offers into each image.

Text assets require the same scale. Using Google's Ask Advisor conversational experience, you can prompt large language models to scan your landing page and extract information to generate custom dynamic headlines. When you use SproutMe, our guardrails cap what an agent can spend testing these new visual and text combinations before a human signs off, keeping rapid creative expansion safely within budget.

Using hybrid workflows for quick UGC

Generative AI does not mean every asset has to be completely synthetic. Hybrid workflows blend authentic customer footage with AI processing to solve the bottleneck without losing credibility.

In AI-generated ads are becoming slop. Here's how I create 30+ ..., one media buyer outlines a process for generating dozens of ad variations in under 30 minutes. By recording a single customer interview through a platform like Riverside and using its AI-driven features, you can automatically extract and edit short-form segments ready for paid placement. The AI identifies testimonial moments and product demos, cutting them into separate assets without requiring hours of manual video editing.

For campaigns where you lack authentic customer footage, you can use generative talking avatars. These tools allow agencies to generate testimonial-style presenter ads by scripting an ideal customer profile and applying it to a digital presenter. This creates targeted variations in hours instead of coordinating expensive shoot days.

The risks of unguided AI generation

Relying on generative tools without strict boundaries creates new problems. When an algorithm is left to mix and match assets blindly, it often finds the cheapest traffic rather than the most valuable customers.

In a case study published in Pmax Auto Generated Videos and Experiences, a digital marketer found that while AI effectively generated promotional videos, the initial launch resulted in highly inefficient traffic. The automated setup brought in junk clicks from South Africa and automated chats that led to zero engagement. The AI chased cheap interactions because it lacked the strategic context of who the actual buyer was.

As detailed in The Agency Guide to AI-Driven Google Ads Performance, automated execution requires strict business constraints. An agent that generates creative but ignores your target audience will efficiently acquire the wrong customer. SproutMe agents operate from a data lake that holds your ICP definitions alongside your performance data, so they test new assets against your actual buyers rather than chasing empty engagement metrics.

Conclusion

Generative AI has moved from a novelty to a structural necessity in Performance Max campaigns. When you rely on automated platform defaults, you sacrifice control over your brand and miss out on double-digit conversion growth. By building custom video, imagery, and text assets using generative models, you solve the creative bottleneck and give the algorithm the raw material it needs to scale. The agencies that win are the ones that treat AI as a high-volume production engine, tightly constrained by human strategy and strict audience definitions.