The real shift: AI is now a traffic source, not just a tool
Most of the headlines you’re seeing right now fall into two buckets:
- “How to use AI tools to do X faster” (ChatGPT SEO tools, custom GPTs, Copilot usage, etc.)
- “How AI is changing discovery” (AI Overviews, Gemini upgrades, misinformation experiments, brand visibility in AI answers)
The first bucket is about productivity. Useful, but not existential.
The second bucket is about distribution. That’s existential.
For performance marketers and media buyers, the real issue isn’t “How do I use AI to write more ads?” It’s:
“What happens to my funnel when AI systems sit between my brand and the user – rewriting queries, rewriting answers, and sometimes rewriting the truth?”
AI Overviews, ChatGPT, and Copilot are no longer just tools your team uses. They’re becoming:
- Gatekeepers of attention
- Interpreters of your brand and offers
- New “surfaces” where users make decisions without ever seeing your site
That has direct implications for:
- Attribution (your data is now even more wrong than you think)
- Brand safety (AI can hallucinate career-damaging claims)
- Media efficiency (you’re paying for clicks that never happen because the answer was given in-line)
- How you structure creative, landing pages, and even offers
This isn’t a “SEO people will figure it out” problem. It’s a performance and P&L problem.
AI as a middleman: how it’s quietly rewriting your funnel
Let’s simplify what’s happening in the wild:
-
AI Overviews & Gemini in search
Google is increasingly answering queries directly with AI-generated summaries. Those summaries:- Blend multiple sources (including your competitors)
- May or may not show your brand, even if you’re the source
- Can be wrong, biased, or incomplete
-
ChatGPT, Perplexity & other AI assistants
Users ask “What’s the best X?” and get a synthesized answer. The model:- Picks a few “preferred sources” (often big brands or strong SEO players)
- Summarizes offers, pricing, and positioning for the user
- May hallucinate details about your product or company
-
Copilot & device-based behavior
How people use AI assistants changes by device. On desktop, they might research; on mobile, they might ask for quick, transactional answers. That shifts:- Query intent distribution (more “do it for me” queries)
- Where in the funnel AI steps in (top, mid, and increasingly bottom)
The net effect: AI is compressing the funnel. Discovery, consideration, and comparison increasingly happen inside an AI interface, not across 5-10 of your carefully optimized pages.
If you’re still thinking in terms of “traffic to landing page to conversion,” you’re missing the new step:
Query → AI intermediary → (maybe) your property → conversion
That “AI intermediary” step is where you win or lose before you ever see a click.
What this breaks for performance marketers
When AI sits between the user and your funnel, three things break fast.
1. Your attribution model
AI assistants don’t show up as a neat channel in your analytics. They:
- Rewrite queries (“cheap CRM for freelancers” becomes “best affordable CRM tools for small businesses”)
- Send fewer clicks (because the answer is already on-screen)
- Change the mix of branded vs non-branded queries that actually reach you
That means:
- Your branded search might look “stronger” while generic discovery quietly decays
- Your retargeting pools shrink even though demand in the category hasn’t
- Channel ROAS comparisons become even more misleading
2. Your brand safety assumptions
We already have real cases of AI Overviews inventing damaging claims about people and brands. That’s not a PR problem; it’s a conversion-rate problem:
- A user asks, “Is [Brand] legit?”
- AI hallucinates a negative review or misinterprets an old forum post
- Your win rate on mid-funnel and bottom-funnel traffic quietly drops
You won’t see “AI hallucination” in your analytics. You’ll just see:
- Lower conversion rates on high-intent segments
- More “research-y” objections in sales calls and chats
- Longer time-to-close
3. Your creative and landing page assumptions
Most landing pages are built for humans skimming. Now you have a second audience:
- Humans
- AI models scraping, summarizing, and quoting you
If your page is:
- Vague (“solutions for modern teams”)
- Overly branded and under-specific
- Buried in jargon and feature soup
…then AI has nothing clear to grab. It will default to competitors with sharper, more structured, more explicit content.
The operator’s job now: design for AI as a distribution channel
You don’t control the models, but you can control the inputs they see and how well they can interpret you.
Think of AI systems as a new “meta-channel” that sits across search, social, and even email. Your job is to:
- Make your brand easy for AI to understand and quote
- Detect when AI is distorting or omitting you
- Adjust your media and creative to work with this new middle layer
Practical moves: how to adapt your funnel to AI intermediaries
1. Create “AI-readable” source content
You don’t need to “optimize for AI” with gimmicks. You need to make your positioning machine-readable and unambiguous.
On your key pages (home, category, comparison, pricing, high-intent blog posts), add:
-
Plain-language positioning blocks
Short, clear sections that answer:- “What is [Brand]?”
- “Who is it for?”
- “What does it help them do?”
Example:
“Acme CRM is a customer relationship management tool for solo freelancers and small agencies. It helps them track leads, send proposals, and follow up automatically without a sales team.” -
Structured comparisons
If you’re in a competitive category, create honest, structured comparison pages:- “Acme vs HubSpot: Key Differences for Freelancers”
- Tables with features, pricing, and who each is best for
AI loves lists and tables. Feed it good ones.
-
Explicit use cases and outcomes
Instead of “transform your workflow,” write:- “Save ~4 hours per week on manual follow-up”
- “Increase reply rates by 20-30% based on customer data from 2,000+ accounts”
2. Monitor how AI actually describes you
Don’t guess. Treat AI assistants like a new review site.
Build a simple monthly or quarterly “AI visibility audit”:
- Ask ChatGPT, Perplexity, and Gemini:
- “What is [Brand]?”
- “Best tools for [your category/use case]”
- “[Brand] vs [Top competitor]”
- “Is [Brand] legit?”
- Screenshot and log:
- How often you appear
- How you’re described
- Which sources are cited
- Track changes over time like you would rank tracking
This gives you:
- Early warning on misinformation or hallucinations
- A list of “source sites” you may want to strengthen (reviews, partners, PR, docs)
- Signals on which pages AI is pulling from when it does get you right
3. Adjust your media mix and expectations
AI intermediaries will blunt some of your organic reach and some of your “free” discovery. That has paid media implications.
-
Expect weaker performance from generic search over time
As AI answers more generic queries directly, treat non-branded search like a decaying asset. Don’t panic, but:- Watch impression vs click trends post-AI rollout
- Shift budget toward:
- Branded and near-branded terms
- Mid-funnel queries that AI still struggles to answer cleanly
-
Use paid to “force” exposure to your own surfaces
If AI is compressing the funnel, your job is to create more direct entry points:- Run more direct-response social to high-intent, AI-readable pages
- Drive traffic into email/SMS where you control the narrative
- Use lead magnets that capture contact info earlier (before AI mediates everything)
-
Re-think retargeting pools
With fewer organic clicks, your retargeting audiences will shrink. Compensate by:- Building more first-party audiences (email, communities, Slack groups, etc.)
- Using creative that speaks to “what you’ve probably read about us” and corrects it
4. Treat misinformation as a conversion problem, not just PR
When AI gets you wrong, don’t just send an angry email to support. Build a playbook:
-
Detection
Your AI audit (above) plus:- Ask sales/support to tag tickets where prospects mention “I read that…”
- Collect those phrases and test them in AI tools
-
Response
Depending on severity:- Update your own content to address the specific claim clearly
- Publish a short “Myth vs Fact” section on a relevant page
- Reach out to any misquoted sources (forums, reviews, blogs)
-
Conversion-layer fixes
If you know a bad narrative exists in the wild, don’t pretend it doesn’t. Add:- FAQ items that directly address it (“Does [Brand] share my data with X?”)
- Short explainer videos or snippets in your sales deck
- Ad copy that pre-empts the objection (“Unlike X, we don’t…”)
5. Use AI tools for ops, not for outsourcing thinking
The irony: while AI is distorting your funnel on the outside, most teams are using it internally for the least important work.
Instead of “write me 10 ad variations,” use AI to:
- Summarize large query logs to spot new intents AI might be intercepting
- Cluster search terms and map them to funnel stages (pre-AI vs post-AI rollout)
- Generate testing plans: “Given this drop in non-branded search, propose 5 experiments across paid and landing pages”
- Build internal playbooks faster (QA checklists, campaign setting audits, etc.)
The goal is not “more content.” It’s faster iteration on strategy and experiments in a world where the ground is moving under you.
What to do this quarter
If you run growth, media, or performance, here’s a simple 90-day plan that respects reality without overreacting to hype.
-
Run an AI visibility and narrative audit
Document how AI tools describe you, your competitors, and your category. Treat it like a new analytics report. -
Fix your “source of truth” pages
Update 5-10 key pages to be AI-readable: clear positioning, structured comparisons, explicit use cases, and outcome statements. -
Instrument your funnel for the new reality
Watch:- Non-branded vs branded search trends
- Retargeting audience growth vs last year
- Conversion rates on high-intent segments where AI is likely involved
-
Rebalance your media
Shift a small but meaningful slice of budget (5-15%) into:- Direct-response social to high-intent pages
- Lead-gen to build owned audiences
- Testing creative that assumes the user has already seen a summary of you elsewhere
-
Write a one-page internal “AI middleman” memo
Align your team on this simple idea: AI is now a distribution layer. Share:- What you’re seeing in audits
- What you’re changing in content and media
- How sales/support should flag AI-driven objections
The teams that win the next few years won’t be the ones who write the most prompts. They’ll be the ones who treat AI not just as a productivity hack, but as a powerful, messy, unavoidable middleman in their funnel – and design for it on purpose.