Why better context decides winning advertisers in the AI age

Performance creative teams have too much context

Thanks to AI, producing more ads is now easier than ever. But are you actually creating more winners now than before LLMs came along? For most advertisers, more doesn’t equal better if they’re not creating the right things. And the reason you don't know what that right thing is is that there is simply so much context that could be used for better creative decision making that it keeps getting lost or forgotten.

Creative teams build hundreds of tests and new creatives every month. They generate real insight: which hooks hold cold audiences, which formats fizzle out fast, which angles were tried and didn't convert and which creative directions actually compound over time. It vanishes into unmaintained docs, reporting that captures outcomes but not reasoning or into the memory of someone who's already left the company.

Every sprint, some version of the relearning happens. The team starts closer to scratch than the data they've accumulated should allow. AI promises to change things by making storing and using this learning much more feasible. But simply prompting your LLM of choice won’t be enough; the winners of the new way of creative work will be decided by who holds the right context.

What creative work will look like in the future

Whether you believe the most outrageous claims of the most AI-bullish people or not, it's certain that the technology has and will keep transforming creative work. 

Over the past couple of years, most performance creative teams have gotten serious about adding AI to their workflows. The pitch is compelling: AI can write (better?) briefs, analyze what's performing, help with brainstorming and ideation, and even do video editing and build actual ad creative. 

Some of the promises are very real: things like video editing or motion graphics have become a lot more accessible. Focal has also benefited from these improvements: our content recognition, tagging and search solutions would not be possible without improvements in AI capabilities. But where are these developments leading us?

When it comes to creative work, humans will always have a place in the loop. Real creativity, original ideas and some of the other elements of the creative process are something AI will never replace entirely, but it has already proven itself as a serious productivity enhancer for people with taste and talent. But I'd argue most of that productivity enhancement so far has gone towards increasing creative volume. Higher ad volume helps advertisers, but not as much as many people think. 

As the effort required to produce ad creative goes down, it's tempting to simply create more and more. But drowning your audiences in AI slop is unlikely to be a winning strategy. What leads to much better results is creating more of the right things. 

Current AI tools and tool stacks are a step in this direction. Connecting MCP servers, APIs and skills to Claude has made it possible to do a lot of creative work through prompting a text interface, when it would have been done by hand inside a software UI in the past. LLMs also make it much easier to recall decisions made and actions taken, because they (in a way) document everything you do, without you needing to write manual documentation.

Prompting an LLM system (or talking to your computer with something like Whispr Flow) will increasingly become a part of most knowledge work, and creative work is no exception. But this approach to using AI still has limitations that start to show when you reach sufficient scale as an advertiser (typically at six-figure monthly budgets and beyond).

Why AI tools aren't closing the performance gap

All AI tools have one basic limitation. They can only work from the context they have access to. And when it comes to creative, the full context, the complete creative record, doesn't exist anywhere as a single object that could be fed to an LLM.

Where does a creative actually live in your stack? For most teams, there's no single place where a creative exists as a single entity: 

• The brief for the asset is in Notion or Docs. 

• The files are stored in Google Drive. 

• Feedback and reviews are done in Slack. 

• Production tasks and status are tracked in a spreadsheet or project management tool like Monday 

• The asset is run inside an ad on Meta, TikTok, and other platforms.

• Performance is tracked in a spreadsheet, dashboard or ad platform reporting. 

When we think of a “creative asset”, we are actually thinking of a set of fragmented decisions and data points that exist across half a dozen different systems, with no formal connections recorded between them. A creative as a complete, single object, with its history and meaning intact, doesn't exist anywhere.

When you add AI tools to this structure, including through MCPs connected to separate individual systems, you get faster access to fragments. The AI can pull Drive files, read a Notion doc, and see ad performance via Meta Ads MCP. But it can't fully understand the connections between them, because those connections were never recorded anywhere. The AI works from the same incomplete picture the team was already working from. Faster, but just as incomplete.

What a connected creative record changes

In Focal, every ad creative and its entire history become one connected record that includes the hypothesis, the brief, who made it and how long it took, the asset and its elements, the tags and the ad performance. 

With Focal, you can now easily store, access and make decisions with the full creative context, not just by connecting fragments of information in your head. 

When this connected record is accessible to AI through Focal's MCP, the kind of questions you can actually answer change. 

• How is my creative production progressing? What is stalling or needs my review?

• What was the hypothesis behind the top performers in Q1, and did it hold in Q2? 

• What kind of B-roll footage do we have that we could repurpose for ads?

• What does my ad performance look like when broken down by script or hook used?

• Based on the performance of ads created from my previous briefs, what should my next brief to the creative agency or designer contain?

• Is there a consistent difference in performance between creatives we get from different creative agencies?

• What kind of angles have we tested for a specific audience persona, and which ones have worked best?

• How many concepts did we create for last Black Friday, and how many of them actually spent over a certain amount and drove sales?

Those questions are answerable with real specificity, because the answers are connected to each other, not siloed across six systems where a human has to stitch things together in their head first. A stack of MCPs connected to individual tools can retrieve data from each silo. It can't hold the connections between them. Those connections are where the intelligence lives.

What this does and doesn't change

I want to be direct about what we enable, because the "AI and creative work" conversation (especially in places like LinkedIn) is filled with empty promises and hype.

The actual making of the creative happens with your editors, UGC creators, design team and AI generation tools. Focal doesn't replace the strategist's judgment about what will resonate in your audience, or the craft that goes into production. 

What changes is the quality and breadth of context the team uses to make every decision. 

• A creative strategist who can recall the full history of what their team has tested, across concepts, formats and timeframes, with full transparency into the images, videos, b-roll and other assets currently available, makes different briefs than one reconstructing from fragments. 

• If the team can get a real-time production status read with a prompt rather than a round of status pings, they’ll be able to move a lot faster. 

• When preparing for a sprint retrospective takes ten minutes instead of a half-day building a reporting deck, you’re left with much more time for thinking and making the actual valuable creative decisions.

These kinds of improvements add up over time. The human keeps making the critical judgment calls while Focal provides the connected record carrying the history and context so the team can make more impactful decisions. The 50th sprint is smarter than the fifth because the context gets richer every time. 

If the insights you’ve acquired from months and years of running campaigns sit in a spreadsheet that is rarely opened or in the head of a creative strategist who ends up leaving the company, your performance will eventually plateau. These teams keep relearning the same lessons, simply because they lack the context and memory to make incrementally better decisions. 

The context gap between these two teams widens with every sprint. For the teams operating with full creative context, it quickly starts to look like a moat. Every creative angle you've tested taught you something. It's time you start remembering and applying all you've learned.