The real story behind all the headlines: traffic is getting intermediated
Read those headlines as a single narrative and a pattern snaps into focus:
- AI Overviews reduce clicks by 58%.
- Answer engine optimization vs. traditional SEO.
- Why GA4 alone can’t measure the real impact of AI SEO.
- PPC budget rebalancing as AI changes where money flows.
- Washington Post trying to win back traffic.
Google, Microsoft, Meta, TikTok, Amazon: they’re all doing the same thing-sitting between you and your customer, answering questions directly, and keeping more of the interaction on their own surfaces.
The short version: your “traffic” is becoming their “product surface.”
For CMOs, performance marketers, and media buyers, this isn’t another “trend.” It’s an operating system change. If you keep optimizing for clicks and last-click ROAS as if 2019 never ended, you will slowly but surely fund the platforms’ P&L more than your own.
From search engine optimization to answer engine arbitrage
Traditional SEO and paid search assumed a simple flow:
Query → Results page → Click → Your site → Conversion
AI overviews, answer engines, and social feeds now look more like:
Intent → Platform constructs an answer → User stays put → Maybe a click, maybe not
That’s not a cosmetic change. It rewires:
- What you measure (impressions and “assists” matter more than raw clicks).
- What you own (fewer first-party touchpoints, more rented context).
- Where you spend (PPC, SEO, social, and CRM budgets start to blur).
The operators who win won’t be the ones who “do more AI content.” They’ll be the ones who treat answer engines as arbitrage opportunities, not as traffic sources.
Three uncomfortable truths about AI-era performance
1. Your SEO KPIs are lying to you
Most SEO dashboards still obsess over:
- Sessions
- Average position
- Organic CTR
- “SEO revenue” based on last non-direct click
In a world where AI overviews cut clicks by half, those KPIs reward the wrong things:
- You can lose traffic but gain more qualified demand that converts via branded search, direct, or sales.
- You can gain traffic while answer boxes skim off the highest-intent users before they reach you.
If your board deck still has “organic sessions” as the headline SEO metric, you’re managing optics, not economics.
2. GA4 can’t see the full game you’re playing
GA4 is built to track what happens on your properties. AI answer engines push more of the decision-making off your properties:
- Decisions made inside AI overviews, chatbots, and shopping units.
- Discovery that starts and ends in TikTok, Instagram, and Maps.
- Multi-device journeys where the first and last touch are both off-site.
Your analytics stack is now blind to a growing share of influence. If you keep insisting on “GA or it didn’t happen,” you’ll cut the very channels that are quietly making your branded search and direct traffic work.
3. Budget lines are fictional; users don’t care if it was “SEO” or “PPC”
Headlines about PPC budget rebalancing and AI shifting spend are really about this: the user doesn’t care whether they clicked an ad, a local listing, an AI answer, or a video. They just care if you solved the problem fast.
But most orgs still budget like this:
- SEO team: “free traffic” and blog posts.
- PPC team: search + social ads, judged on last-click ROAS.
- Brand: TV, sponsorships, Super Bowl moments.
- CRM: email and lifecycle, judged on “email-attributed revenue.”
AI intermediaries don’t respect those silos. You’re either the best answer in the moment, or you’re invisible.
A new operating system: answer-centric, not channel-centric
Here’s the shift: stop thinking in channels, start thinking in answers to specific intents.
For each high-value intent, ask:
- Where does this intent show up first? (Search, social, Maps, marketplaces, AI chats.)
- What does a “good answer” look like in that context?
- How can we be the default answer-paid, organic, or embedded?
- How do we prove that being the answer drives revenue, even if the click happens later somewhere else?
That requires structural changes, not just new tactics.
Five practical moves for CMOs and performance leaders
1. Rewrite your performance scorecard around “answer share,” not just traffic
Keep traffic, but demote it. Add metrics that reflect how often you’re the chosen answer:
- Impression share by intent across paid + organic search, Maps, shopping, and key marketplaces.
- AI answer presence: inclusion rate and positioning in AI overviews for core queries.
- Brand search lift and direct traffic lift correlated with answer-engine presence.
- Downstream conversion rate from users exposed to your answer vs. control where possible.
The question for every major intent becomes: “What percent of the time are we the answer, and what is that worth?”
2. Merge SEO, content, and paid search into a single “intent team”
The days of SEO rewriting 8,000 title tags in a silo while PPC fights for budget on the same queries are numbered.
Build a cross-functional “intent desk” that owns:
- Query and intent mapping (including AI answer opportunities).
- Unified creative: copy, landing experiences, structured data, feed quality.
- Joint P&L for search and answer surfaces: paid + organic.
- Experimentation across formats (ads, snippets, videos, tools, calculators).
Their mandate: maximize profitable coverage of high-value intents, regardless of channel label. If an AI answer is stealing your best queries, this team figures out how to get your brand inside the answer-through content, schema, partnerships, or paid units.
3. Stop treating AI content as a volume hack and start treating it as a precision tool
Yes, AI can churn out SEO content at scale. That’s also how you quietly destroy trust in your own brand and clog your site with cannibalized, low-signal pages.
Use AI where it gives you precision, not just volume:
- Generate structured variants for titles, meta, and ad copy for testing.
- Summarize complex internal knowledge into answer-friendly formats.
- Produce tailored versions of core assets for specific segments or geos.
- Draft, but never fully own, your brand’s point of view-human editors still guard the message.
The Copyhackers line about “AI’s trust problem” is the real risk: if you outsource your message to generic AI, you train both users and algorithms to see you as interchangeable.
4. Build “resistance infrastructure”: assets that can’t be easily disintermediated
If platforms are going to sit between you and the user, you need assets they can’t fully replicate or scrape away.
Think in three buckets:
-
Proprietary data and tools
Calculators, configurators, live pricing, availability, benchmarks, and interactive tools that require your backend. AI can summarize static blog posts; it can’t fake your real-time quote engine. -
Community and creators
The Washington Post investing in live blogs and video to resist AI scraping is the same play used by MrBeast or top TikTok creators: build formats and personalities that people seek out directly, not just stumble into via search. -
Service and experience hooks
Loyalty programs, workshops, and experiences (think Lowe’s kids workshops, or unique retail activations) that give people reasons to interact with you beyond a single query.
Ask your team: “If Google or OpenAI tried to fully replace us with an answer box, what would they struggle to copy?” If the honest answer is “not much,” that’s your roadmap.
5. Fix your attribution religion before it quietly kills growth
AI-era journeys are messy:
- First touch: TikTok or Instagram short.
- Middle: AI overview or chat summarizing options.
- Late: branded search, direct visit, or marketplace purchase.
If you still optimize on last-click ROAS in Google Ads or “email-attributed revenue” in your CRM tool, you will underfund the surfaces that actually create demand.
Practical steps:
-
Adopt a simple, opinionated model
You don’t need a PhD-grade MMM out of the gate. Start with a blended view: channel clusters, contribution to new vs. returning customers, and directional incrementality tests. -
Run cheap, dirty incrementality tests
Geo splits, holdout groups, and platform-level lift studies. Imperfect but better than pretending GA4 is ground truth. -
Set guardrails instead of single-source truth
For example: “We will not cut a channel solely because last-click ROAS dipped, if brand search and direct are up in the same regions.”
How to brief your team for the next 12 months
If you’re a CMO or growth lead, the most useful thing you can do this quarter is reframe the mission for your teams.
Instead of:
- “Grow organic traffic by 20%.”
- “Hit a 4x ROAS in paid search.”
- “Increase email revenue by 15%.”
Try:
-
“For our top 50 intents, we will be the best answer on every major surface.”
Measured by impression share, answer inclusion, and downstream revenue. -
“We will build three assets this year that AI answer boxes cannot fully replace.”
A proprietary tool, a flagship content format, and a community or program. -
“We will reduce our dependence on any single platform for more than 30% of new customer acquisition.”
With clear diversification targets and testing budgets.
That kind of clarity gives your operators permission to stop playing channel whack-a-mole and start designing for how people actually make decisions now-inside feeds, inside answer boxes, and inside ecosystems you don’t own.
The platforms have already updated their operating systems. The question is whether your marketing org will do the same before the next 58% of clicks disappear.