The pattern you should care about (and it’s not TikTok vs. Pinterest)
Scan those headlines and you see the same surface story everyone talks about:
TikTok drama, Instagram tools, Pinterest stats, AI keyword prompts, new ad formats, new CEOs talking about AI.
Underneath that noise is the only pattern that matters for working operators:
Media is fragmenting, AI is intermediating, and your current marketing org is structurally not built to handle either.
The winners over the next 24-36 months won’t be the ones who “adopt AI” or “test new channels.” They’ll be the ones who quietly redesign their operating system for how media, creative, data, and AI work together.
What’s actually changing (beyond the headlines)
1. Channels are unstable, but your bets are still static
TikTok may be sold. Instagram keeps rolling out new tools. Pinterest is having a moment (again). Criteo’s still doing a billion a quarter. Agentic commerce is forecast to hit $1T by 2030. Every week, some platform becomes “too expensive” or “the next big thing.”
Meanwhile most orgs still:
- Lock budgets in quarterly or annually by channel.
- Run separate teams and agencies by platform (search, social, programmatic, CRM).
- Measure success in siloed dashboards that don’t agree with each other.
You’re trying to play speed chess with a Jenga tower.
2. AI is no longer a tool; it’s becoming an environment
Look again at the AI headlines:
- “Agentic AI vs. Generative AI”
- “Claude Code for Everyone”
- “Ads in ChatGPT”
- “Google adds more links & link context to AI search”
- “Your managed WordPress might be blocking AI bots and you can’t see it”
This isn’t just “use AI to write ads.” It’s:
- AI crawling and rewriting the discovery layer (AI search, AI overviews).
- AI agents making or influencing purchase decisions (agentic commerce, shopping copilots).
- AI tools embedded into every workflow: content engineering, keyword research, social scheduling, sales outreach.
Your media and content are now being consumed, re-ranked, summarized, and sometimes purchased by machines on behalf of humans.
3. Trust and control are becoming the real constraints
You see it in:
- “AI’s trust problem: The cost of outsourcing your message in a SaaS recession.”
- “Modern Retail: Marketing workflows benefit from AI, but trust is still a barrier.”
- 80% of CEOs worrying their job is at risk if AI fails this year.
Most teams don’t actually lack AI tools. They lack:
- Clear guardrails on what AI can and can’t do.
- Ownership of prompts, templates, and workflows.
- Confidence that AI-driven outputs won’t break brand, compliance, or performance.
So AI sits in the experimentation bucket while CPCs keep climbing and channel volatility gets worse.
The uncomfortable truth: your org chart is now your biggest media tax
You can’t solve this with “more testing” or “better creative briefs.” The problem is structural.
Most marketing orgs were built for a world where:
- Search was search, social was social, CRM was CRM.
- Humans made most of the decisions, platforms executed them.
- Attribution was messy but at least recognizable.
Now:
- Search results are half AI-generated, half links.
- Chat interfaces (ChatGPT, Claude, Perplexity, Gemini) are becoming ad surfaces and shopping advisors.
- Agentic systems will soon compare, negotiate, and transact for people.
- Your content is being ingested into models, not just indexed by crawlers.
Yet your operating model still assumes:
- Channel managers own strategy inside their silo.
- Data lives in BI or analytics, not inside the daily tools.
- AI is a side project, not part of the production line.
That gap is where you’re losing money. Not in your bid strategy, not in your latest A/B test, but in the friction between how your team is structured and how the ecosystem now behaves.
The new operating model: from channel-first to system-first
So what does a system-first media operation actually look like in practice?
1. Organize around journeys, not channels
Instead of:
- Paid search team
- Paid social team
- SEO team
- Lifecycle/CRM team
Move toward:
- Demand creation pods (problem-aware to solution-aware).
- Demand capture pods (high intent, bottom-of-funnel, retargeting, branded search, marketplaces).
- Customer growth pods (onboarding, expansion, retention, referrals).
Each pod owns:
- A clear stage of the journey.
- A set of metrics (not 40 KPIs, 3-5 that matter).
- A cross-channel budget they can move dynamically.
- Embedded AI workflows for research, production, and optimization.
Channels become tactics, not fiefdoms.
2. Treat AI as infrastructure, not a side project
The teams that pull ahead will treat AI less like a copy intern and more like a shared utility:
- Central “prompt library” and templates for briefs, ad variants, keyword expansion, audience research, and QA checks.
- Content engineering standards (as the Ahrefs team is already doing) so content is structured for both humans and AI systems.
- Agent-ready assets: product feeds, pricing, and policies structured so AI agents can easily compare and recommend you.
- Automated hygiene: scripts/agents that watch for cannibalization, broken tracking, blocked bots, and SEO basics at scale.
This doesn’t mean “replace your team with AI.” It means your team designs the system; AI runs the repetitive loops.
3. Build for AI-mediated discovery, not just human searchers
Google adding more links and context to AI search, PR suddenly mattering more for SEO, AI overviews rewriting SERPs – all of that points to one shift:
you’re now marketing to algorithms that market to people.
Practically, that means:
- Investing in entity-based SEO: brand, people, products, and categories clearly defined and reinforced across the web.
- Coordinating PR + SEO so earned coverage feeds AI systems high-authority context about what you are and where you fit.
- Publishing structured, canonical content that AI can safely summarize without distorting your message.
- Monitoring where and how your brand shows up in AI answers and chat interfaces – not just in blue links.
If your only “visibility” plan is rank on page one, you’re already late.
4. Redesign your measurement for messy, AI-heavy journeys
The “whole point was the mess” headline about search applies to your funnel now. Journeys are non-linear, multi-device, and increasingly AI-assisted.
You won’t fix this with a new attribution vendor. You need a measurement philosophy that accepts mess and still guides decisions:
- Fewer, better north stars per pod (e.g., qualified pipeline, CAC payback, incremental revenue), not channel ROAS in isolation.
- Experimentation as a muscle: holdouts, geo splits, incrementality tests baked into planning, not ad hoc “special projects.”
- Directional models combined with operator judgment, instead of pretending any model is “the truth.”
- AI-assisted analysis to scan logs, queries, and creative variations for patterns humans miss – with humans deciding what to do about it.
The goal isn’t perfect attribution. It’s actionable confidence about where to move the next dollar.
What CMOs and media leaders should actually do in the next 90 days
This all sounds big and abstract until you tie it to a 90-day plan. Here’s a concrete, operator-level sequence.
Step 1: Map your real system (not your org chart)
In one working session, with your core team:
- Draw the full journey from first touch to repeat purchase.
- List every major channel, tool, and team that touches each stage.
- Mark where AI is already in use (however small) and where it obviously could be.
- Circle the three biggest friction points: handoffs, delays, or blind spots.
This is your actual marketing system. Everything else is job titles.
Step 2: Stand up one cross-functional “pod” as a live test
Don’t reorg the whole company. Pick one high-impact slice, for example:
- B2B: “Demo-request to closed-won” demand capture pod.
- Ecommerce: “First-time buyer acquisition” pod.
- Retail: “Local store visit + online purchase” pod.
Give that pod:
- One shared target (e.g., incremental revenue or qualified leads).
- Authority to move budget across 3-5 channels within that slice.
- Dedicated support from data and creative.
- Explicit permission to embed AI in their workflows.
Step 3: Productize 3-5 AI workflows that save real time
Focus on repeatable, boring work that’s already happening:
- Search: AI-assisted negative keyword mining and query clustering.
- SEO: AI-assisted cannibalization checks and title/meta rewrites, reviewed by humans.
- Paid social: AI-generated first drafts of creative variants based on a master concept.
- CRM: AI summaries of customer feedback and support tickets, feeding copy and positioning.
Document prompts, guardrails, and approval flows. Treat them like mini-products, not one-off hacks.
Step 4: Align PR, content, and SEO around AI discovery
In one joint planning session:
- Define 5-10 core entities you want to be “about” (categories, problems, use cases).
- Plan content and PR around reinforcing those entities with consistent language and proof.
- Audit your top-performing pages for how “summarizable” and “quotable” they are for AI systems.
- Set a simple monitoring routine: monthly checks of how AI search and chat tools describe your brand and category.
Step 5: Change one incentive that keeps you channel-locked
You don’t need a full comp redesign. Pick one move that sends a clear signal:
- Shift variable comp for channel owners from channel-specific ROAS to journey-stage outcomes.
- Set a quarterly target for “budget reallocation speed” – how fast you can move 10-20% of spend between channels when performance shifts.
- Make at least one senior leader’s goal tied to AI-enabled efficiency (e.g., cost per qualified lead down 15% with no volume loss) instead of “AI adoption.”
The quiet advantage: building a media operation that can survive chaos
TikTok might get sold. Ads in ChatGPT might work or flop. CPCs might keep rising. Another platform will launch next quarter promising cheaper attention.
You can’t predict any of that. But you can decide what kind of system you run:
- Channel-first, tool-chasing, AI as garnish.
- Or system-first, journey-owned, AI as infrastructure.
The second one is less glamorous. It doesn’t make headlines. But it’s the one operators bookmark, because it quietly compounds every quarter – regardless of which logo is on the latest “must-buy” media plan.