I've heard some version of that question at every dinner, every conference, and every new business meeting for the past year. Everyone's trying to figure out if they're keeping up.
It's the wrong question. "What is your AI strategy?" will send you in the wrong direction every time because it skips over the part that actually matters: what is your strategy, and where does AI fit inside it?
When you start with the AI question, you end up bolting tools onto existing workflows and calling it transformation. Then the pilots that looked clean in a demo hit real operations and fall apart. The organizations that are actually getting value from AI started somewhere different: they defined the outcome first, then built the system around it. Judgment and taste are scarce, and they're the one thing that will differentiate your brand when everyone has access to the same tools. AI handles the execution. Humans own the creative and strategic calls that can't be commoditized.
That distinction was the focus of my talk at the Adaptive Advantage Executive Summit we hosted in Pittsburgh. The summit brought together senior marketers, platform partners, and researchers for one full day of working sessions. Ten speakers. Six platforms.
Google made the case that measurement has to prove causation, not just report correlation. Meta showed that creative diversity is now a media efficiency lever. Reddit showed that AI search is reshaping purchase decisions whether you're managing the narrative or not. And a CMU professor used the story of how a birdwatcher redesigned the nose of a bullet train to explain why AI trained on existing solutions only produces more existing solutions. That one stopped the room.
Attendees also caught the Pittsburgh Pirates at PNC Park on arrival day, then spent Thursday evening with Second City, the legendary Chicago improv group, for a show at City Theatre. Friday morning started with a hands-on improv workshop back at the hotel. Turns out, spending time learning to think on your feet was a fitting way to kick off a full day of sessions about exactly that.
The full recap is on the blog. Every session, every insight, and an action item at the end of every section.
Now, here's what else caught my attention this week.
Patrick Patterson
CEO at Level Agency
What's interesting this week
The art of making fintech cool with Catherine Ferdon of Coinbase Coinbase's CMO has a framework worth borrowing: what is the product trying to say, who is it speaking to, and how do we reach the people who will actually care? The whole conversation is a masterclass in marketing products that don't physically exist. Her take on the Super Bowl karaoke ad
is especially interesting. Not everyone loved it. That was the point. Read more
How to use social data for target audience analysis Sprout Social put together a solid breakdown of how to use social listening data to actually build audience profiles instead of just guessing. Nothing revolutionary here, but a good reminder that the data most teams already have access to is wildly underused. If your team is running social and not feeding those insights back into targeting, you're leaving money on the table. Read more
Google's push for data strength is really a push for better bidding Google keeps talking about "data strength" inside Performance Max, and this piece unpacks what that actually means. The short version: Google wants more of your first-party data so its bidding algorithms can work better. That's not a bad thing if you're feeding it quality signals. It is a bad thing if you're treating it like a black box and hoping for the best. Read more
Marketing that pleases everyone converts no one This one hit close to home. The author compares safe marketing to a chef who plays it safe for a food critic and gets destroyed for it. His argument: when you strip tension out of your message, you strip out the reason anyone would act. The best line in the piece is about how no one person in the room is a coward on their own, but collectively, marketing by agreement makes them one. Read more
Meta is about to pass Google in digital ad revenue for the first time Emarketer is forecasting that Meta will surpass Google in worldwide ad revenue this year. Meta at $243 billion, Google at $239 billion. The gap is being driven by Meta's growth rate accelerating to 24% while Google holds at 12%. Together with Amazon, the three platforms now account for 62% of all global digital ad spending. If you're allocating budget anywhere in digital, this is the competitive landscape shaping what your impressions cost and where your reach actually lives. Read more
How much do TikTok ads cost in 2026? WordStream pulled real data from accounts they manage and cross-referenced it with third-party studies. The numbers: $0.30 to $1.50 per click, $4 to $10 CPM, and $0.01 to $0.07 per view. If you're running TikTok and haven't benchmarked your costs against these ranges lately, this is a good gut check. Read more
Deep dive: AI
Stanford's 2026 AI Index: the scorecard nobody can ignore
Stanford's annual AI Index dropped this week, and it's the most comprehensive snapshot of where AI actually stands right now. A few things jumped out.
First, the race between the US and China is closer than most people think. On the Arena ranking platform, Anthropic currently leads, but Chinese models from DeepSeek and Alibaba are only modestly behind. The margins separating the top models are razor thin, which means competition is shifting from raw capability to cost, reliability, and real-world usefulness.
Second, adoption is outpacing every previous technology wave. AI is being used by more than half of people worldwide. An estimated 88% of organizations use it in some capacity. That's faster than the personal computer and the internet.
Third, the benchmarks we use to measure AI progress are breaking. Some popular tests have error rates above 40%. Others can be gamed by training on the test data. And for emerging categories like AI agents and robotics, benchmarks barely exist. The takeaway: be skeptical of any headline that says a model "passed" a benchmark. The yardstick is probably wrong.
Finally, the job impact is real but concentrated. Employment for software developers aged 22 to 25 has dropped nearly 20% since 2022. Customer support roles are seeing similar trends. The full economic picture is still developing, but early career professionals are feeling it first.
Does free will exist? A neuroscientist is pulling the thread.
This isn't a marketing article. But if you work in marketing, you should read it anyway.
Neuroscientist Uri Maoz has spent years studying how the brain turns desires and beliefs into actions. His research started with a famous 1980s experiment by Benjamin Libet that many people interpreted as proof that free will doesn't exist. The experiment showed that an electrical signal in the brain appeared before participants reported deciding to move their hand. If the brain decides before we're conscious of it, who's really in charge?
Maoz reran those experiments with a twist. Instead of asking people to make meaningless choices (wiggle your wrist whenever you feel like it), he gave them real ones: choose which nonprofit gets $1,000 of actual money. The result? That brain signal disappeared. Libet's finding only showed up in arbitrary decisions, not meaningful ones.
That distinction matters for anyone trying to understand how people make choices. The trivial ones might be on autopilot. But the ones that carry weight, cost money, or change direction? Those appear to work differently. If you're building campaigns, you're in the business of moving people from autopilot to intentional action. The neuroscience is starting to explain why that's so hard and what it actually takes.
Opus Clip takes a long-form video and turns it into short-form clips ready for TikTok, Reels, and YouTube Shorts. You upload a video or paste a YouTube link, and the AI identifies the strongest moments, reframes them vertically, adds captions, and scores each clip with a predicted virality rating so you know which ones to post first.
If your team is producing webinars, podcast episodes, or event recordings and not repurposing them into short-form content, the issue is almost always time. Opus Clip removes that bottleneck. A 45-minute video produces 10 to 15 clips in a few minutes. You'll still want to review and discard the weaker ones, but the starting point is significantly better than scrubbing through footage manually.
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."
