AI-Powered Growth

June 3, 2026

Captain's log

The Level team is bringing an interactive experience to our SMX Advanced booth this week called “The Last Keyword.”

The experience asks people to submit their idea of the last keyword typed into Google before AI made keyword-level thinking obsolete. It could be their last keyword, a funny idea of a last keyword (“what does 6 7 mean?!”), or something deeper about humanity’s last keyword. Then we display a rotating gallery of those submissions at the booth. Sure, it’s a tongue-in-cheek conversation piece, but it's also a real question about the shift in user search behavior.

At Google Marketing Live last month, one of the more honest things overheard by our President Bill Buchanan was that Google isn't redesigning Search to change user behavior. They're redesigning it because the behavior already changed. (Check out our recent podcast episode featured below that focuses on all the GML 2026 updates.)

The industry spent years trying to decode Google's intent, when the real signal was sitting in the search bar the whole time. People stopped typing "cheap flights Boston” and started typing "compare flights from Pittsburgh to Boston for two adults in late June, we're flexible on airport, and one of us has TSA PreCheck." Google's response was to build a box that expands when you type. This is design thinking in action: building products to meet users where they are rather than where you want them to be.

AI Mode crossed one billion monthly users in its first year, and queries have been more than doubling every quarter since launch. Overall search volume hit an all-time high last quarter. Ok, the keyword isn't dead. But it’s no longer the unit of measurement.

So what would your idea of the last keyword be? My guess is something embarrassingly mundane. "Weather." "Pizza near me." "Fix a leaky faucet." The kind of query someone fires off in three seconds without thinking, because that reflex is twenty years old and it dies last.

A single person who is not a seasoned developer built “The Last Keyword” experience in a single day. No creative brief, no three-week timeline. Ideation happened on a Wednesday afternoon, and the thing was built and deployed by Friday. The taste and judgment that shaped it are irreplaceable. The production time is not.

That's what this moment actually looks like in practice.

Patrick Patterson

CEO at Level Agency

Podcast spotlight

What Google Marketing Live 2026 Really Means for Marketers
With Bill Buchanan and Nicole Jennings

Google Marketing Live 2026 dropped a wave of announcements, and this episode is the best 45 minutes you can spend making sense of them. Brad Stephenson sits down with Level Agency President Bill Buchanan, who attended GML in person including a rare behind-the-scenes session with Google's keynote team, and Nicole Jennings, our EVP of Service Delivery, to break down what actually matters and what brands should do right now.

The throughline: Gemini is now the operating system of Google Ads. Ask Advisor, AI Max, agentic commerce, Meridian inside Analytics 360. Every announcement assumes AI is the default surface. The question for marketers is whether their data foundation is clean enough to take advantage of it.

Apple Podcasts | Spotify

What's interesting this week

Google shares guidance on visual AI generation
Google published prompting guidance for Gemini Omni's image generation, and the tips transfer to every other AI image tool your team is already using. Specificity about lighting, style, composition, and subject matter produces dramatically different results than vague descriptions. Most teams are leaving a lot on the table in how they prompt for visuals.

Key takeaway: Better prompts are a skill. This guidance is a fast training document for any creative team that keeps getting generic-looking AI output.

Netflix's ad growth is nipping at YouTube's heels
eMarketer is forecasting Netflix at roughly $3 billion in US ad revenue this year, doubling what its ad business produced in 2025. YouTube is forecast at around $9.2 billion in CTV ad revenue for the same period. Netflix's growth rate is the fastest among streaming ad platforms in percentage terms. The ad-supported tier now accounts for more than 60% of new sign-ups in markets where it's available.

Key takeaway: Netflix is no longer an emerging channel. The brands testing now are building the case history everyone else will wish they had in 2027.

The 50 most-cited websites in Google AI Overviews (June 2026)
Ahrefs analyzed millions of AI Overview responses to find the domains getting cited most often. The sources share a few things: content authority, topical depth, and consistent original reporting. This is the clearest map yet of what Google's AI trusts enough to cite by name.

Key takeaway: If your domain isn't in this conversation, the question isn't about schema or technical SEO. It's about whether you're producing the kind of content an AI would trust enough to surface.

Deep dive: AI

Inside The Trade Desk's Claude-powered campaign agent

This one came out a few weeks ago and is worth a read if you missed it.

The Trade Desk has a long history of AI ambition and a more complicated history of AI delivery. Its latest effort, a system called Koa Agents, is notably smaller in scope than previous announcements. That's exactly what makes it credible.

Koa Agents connects to whatever AI model a partner is using, including Claude, through an interoperability layer called Open Agentic Kit. A workflow built on Claude can ingest a media plan, reformat it for The Trade Desk's system, build the campaign, troubleshoot creative issues, and surface recommendations with the reasoning behind them.

The work it handles isn't glamorous. It's the stuff that takes hours and is easy to get wrong under deadline pressure. If it runs reliably, that's a real argument for agentic media buying.

Key takeaway: The programmatic workflow is being rebuilt around whoever shows up with clean inputs and the technical fluency to use them.

Deep dive: Growth

Why the best brands are relearning how to entertain first, advertise second

Worth reading slowly. The data is sharper than it's been on a pattern most senior marketers already sense. Brands putting up strong attention numbers right now built something worth watching before they asked for the click.

The shift Digiday is tracking is from posts to programming. Owned media experiences that compound over time rather than individual pieces optimized for a single conversion. Interruption is getting more expensive every quarter, and the numbers in this piece show where the budget is starting to move.

Key takeaway: If the creative wouldn't hold attention without an ad attached to it, the media spend is working harder than it should have to.

AI tool of the week

Claude Opus 4.8

Anthropic released Claude Opus 4.8 last week. The improvement that keeps showing up in early tester feedback across coding, legal, and financial workflows: it catches its own mistakes. Flags uncertain outputs before you have to. Pushes back when a plan isn't sound. Anthropic's evaluations put it at roughly four times less likely than Opus 4.7 to let flawed work pass without flagging it.

Two things launched alongside it worth paying attention to: effort control, which lets you dial how deeply the model thinks based on the task, and dynamic workflows in Claude Code for large-scale parallel operations. Curious to see how the effort control setting changes the experience across different kinds of work.

What does ChatGPT actually say about you?

When a potential buyer asks an AI agent for a Top 5 list in your category, do you make the cut? More importantly: Why or why not?

Don't leave your brand’s reputation to an algorithm you haven't audited. Our scorecard pulls back the curtain on how AI platforms frame your brand sentiment and which sources they use to vet your authority.

Inside your custom report:

Sentiment analysis: Is the AI's summary of your brand accurate or outdated?
Citation audit: Discover which third-party sites are feeding the AI your info.
AEO roadmap: Immediate steps to improve your "AI discoverability."

SEE WHAT AI SAYS ABOUT MY BRAND

If this email was forwarded to you, sign up here to get future issues.