The real shift: from feeds and SERPs to answer engines
Look past the noise about “trending TikTok sounds” and “Domain Rating benchmarks.” There’s a single structural shift running through a lot of recent headlines:
- Referral traffic is declining for smaller publishers.
- Google is testing AI search data and folding more owned surfaces into Analytics.
- On-page formats that “answer engines” favor are changing.
- Automated SEO and AI-powered lead gen are moving from experiments to systems.
Translation: discovery is consolidating into answer engines – AI-infused systems that decide what users see without sending as much traffic back to you. That includes:
- AI search experiences (Google, Perplexity, ChatGPT, etc.)
- Social feeds tuned to “time well spent,” not raw reach
- Marketplaces and walled gardens that keep users inside their own UX
For CMOs, performance leads, and media buyers, this isn’t an SEO curiosity. It’s a growth architecture problem. If you keep planning like it’s 2016 – chasing referral spikes and “winning the feed” – you will spend more to grow less.
The play now is to build an Answer Engine Ready growth system: content, data, and media working together so that wherever the answer is given, you still get the demand, the lead, or the sale.
Why referral is collapsing (and why that’s not the main problem)
“Referral traffic is declining for smaller publishers” is a symptom, not the disease. The disease is this:
Platforms are no longer in the business of sending people to you. They’re in the business of answering people for themselves.
Three forces are driving it:
1. AI summaries are eating the click
Search engines and social platforms are testing AI summaries, “quick answers,” and remix tools that:
- Use your content as training data
- Serve a synthetic summary or remix
- Reduce the need to click through
The user gets their answer in one screen. You get… brand exposure if you’re lucky, a logo if you’re very lucky, and often nothing trackable.
2. Feeds reward retention, not referrals
LinkedIn, TikTok, Facebook, YouTube: their incentives are simple – keep users inside, keep them scrolling, keep them watching. External clicks are a tax on their own engagement metrics.
So feeds favor:
- Native formats (carousels, polls, shorts, remixes)
- Content that can be consumed fully in-feed
- Creators who post often enough to train the algo
Your “pillar blog post” with a link preview is a second-class citizen in that world.
3. Walled gardens are growing up
Marketplaces, retail media, app stores, super-apps – all are quietly becoming answer engines for specific intents:
- “What should I buy?” → Amazon, Walmart, TikTok Shop
- “What’s happening in my industry?” → LinkedIn, niche communities
- “Which SaaS should I use?” → G2, Capterra, category newsletters
They capture demand, shape consideration, and increasingly close the transaction. Your .com is a verification step, not the main stage.
Judgment literacy beats prompt literacy
You’re being sold a comforting story: “Just learn better prompts” or “Just plug AI into your existing content process.” That’s not strategy. That’s a UI tutorial.
Ann Handley’s point that AI literacy is really judgment literacy is the one operators should care about. The advantage is no longer:
- Who can ship the most content
- Who can stuff the most keywords
- Who can test the most ad variants
It’s:
- Who can decide what is worth automating
- Who can decide where to own the experience vs. rent it
- Who can decide how to measure value when clicks disappear
That’s judgment. And it needs a new operating model.
From “SEO strategy” to Answer Engine Ready growth system
Let’s move from diagnosis to design. An Answer Engine Ready system has four pieces:
- Content built to be quoted, summarized, and remixed
- Owned properties tuned for conversion, not vanity traffic
- Media buying that assumes partial attribution by default
- Data and reporting that tell a story without perfect click trails
1. Design content to be the answer, not just rank for it
Old playbook: keyword → blog post → rank → get clicks → retarget. New playbook:
“If an AI or feed is going to answer this question in 2-3 sentences or a 30-second clip, what do we want it to quote?”
That means:
- Clear, quotable statements – Distinct POVs, data points, and definitions that can be lifted cleanly.
- Structured content – FAQs, comparison tables, step lists, and summaries that answer engines can parse and repurpose.
- Format variety – Short video, carousels, text posts, and longform living off the same core insight, so every surface has a “native” answer.
- Brand-linked language – Phrases and frameworks that, when repeated, point back to you (think “Jobs to be Done,” “Category Design,” etc.).
The goal is not just “be present.” It’s “be the canonical reference” on the questions that actually move pipeline or revenue.
2. Turn your .com into a conversion engine, not a museum
If fewer people arrive, the ones who do matter more. Operators who are still obsessing over Domain Rating while their on-site conversion is a mess are solving the wrong problem.
Steal from that Moz case study that grew inquiries 37%: treat your site like a living sales script, not a brochure. Concretely:
- Cut the fluff – Every page should answer: “Who is this for? What problem do they have? What happens next?”
- Align offers to intent – High-intent pages get direct calls-to-action (book a demo, get pricing). Low-intent pages get soft offers (tools, benchmarks, buyer’s guides).
- Instrument everything – Scroll depth, click maps, form drop-off, chat interactions. You can’t fix conversion blind.
- Build direct response assets – Calculators, audits, ROI tools, and quizzes that capture leads when people do decide to click through from an answer engine.
Traffic is now a privilege. Treat it that way.
3. Buy media like attribution will always be incomplete
AI search, privacy rules, and walled gardens guarantee one thing: your reporting will under-credit the channels that actually create demand.
If your media buying model still assumes clean last-click paths, you’ll keep over-funding bottom-of-funnel branded search and under-funding the messy middle where people discover you.
Operators who are adapting are doing three things:
-
Separating demand creation and demand capture
Different budgets, different KPIs, different expectations. Top and mid-funnel channels are judged on:- Lift tests (geo splits, holdout groups)
- Branded search and direct traffic trends
- New account and opportunity creation
-
Using AI for scale, not for strategy
AI helps with:- Creative iteration (multiple hooks and formats)
- Audience expansion within guardrails
- Bid and budget optimization
But humans still decide:
- Which segments to prioritize
- What the actual promise is in the ad
- Where brand risk is acceptable (especially with new targeting rules around sensitive categories)
-
Planning around “dark” influence
Podcasts, newsletters, communities, and influencers are shaping purchase decisions without neat UTMs. Treat them like brand channels, not performance channels, and measure:- Self-reported attribution (“How did you hear about us?”)
- Search volume for branded and category terms
- Downstream CAC and LTV shifts in markets where you invest
4. Report like an investor, not a channel manager
Most marketing dashboards are still built for the referral era: sessions, CTR, CPC, ROAS by channel. In an answer engine world, that’s table stakes at best and misleading at worst.
You need a second layer of reporting that answers three questions:
- Are we being chosen as the answer?
- Are we converting the people who do reach us?
- Is the whole system creating profitable growth?
Practically, that looks like:
-
Visibility metrics
Not just rankings. Think:- Share of answers in key AI/answer surfaces (where you can measure it)
- Share of voice in key social feeds and communities
- Mentions and quotes of your frameworks, data, and POV
-
System metrics
Tie marketing to business outcomes:- Pipeline created by first-touch cohort, not just last-touch
- Payback period by program (not just channel)
- Incremental lift from campaigns vs. business-as-usual baseline
-
Judgment checks
Quarterly reviews where you explicitly ask:- What are we over-optimizing because it’s easy to measure?
- Which bets are working despite weak click-path proof?
- Where is AI helping us, and where is it quietly dumbing down our strategy?
Practical moves for the next 90 days
This doesn’t have to become a two-year “transformation program.” You can start making your growth system Answer Engine Ready this quarter.
1. Audit your “answer surface”
Pick the 20-30 questions that matter most to your revenue:
- Problems your best customers Google
- Objections your sales team hears
- Comparisons buyers make between you and alternatives
Then:
- Search them in Google, in AI search tools, and in social search.
- Look at who is being quoted, summarized, or shown.
- Score yourself: Are you invisible, present, or the clear reference?
2. Ship one “canonical answer” asset per week
For each key question, create:
- One strong, structured longform answer (on your site)
- One short video or carousel that answers the same question natively
- One memorable line or framework that you want repeated
This is not “more content.” It’s better bets on the questions that actually drive revenue.
3. Fix the top three conversion leaks on your site
Pull your analytics and:
- Identify the three highest-traffic pages with the worst conversion or engagement.
- Rewrite them like a sales conversation, not a brochure.
- Add one clear, relevant next step (tool, demo, guide, quiz).
- Test one friction-reducing change (shorter form, social proof, live chat).
If referral is down, the fastest way to defend growth is to squeeze more value out of the attention you still earn.
4. Reframe your media plan in terms of questions, not channels
Instead of “We’re putting 20% into LinkedIn and 30% into search,” ask:
- Which questions are we trying to own?
- Where do those questions naturally show up – search, feeds, communities, marketplaces?
- What’s the right mix of paid, partner, and owned to show up as the answer there?
Then adjust budgets accordingly. You’ll find some channels are overfunded simply because they’re easy to report on.
The quiet power shift you need to plan for
Ben Thompson’s “Power Shifts” framing applies here: power is moving from sites and feeds to answer engines and the companies that control them. You won’t reverse that trend with a better robots.txt file or one more TikTok sound.
But you can decide how your brand shows up in that world:
- As raw material for someone else’s answers
- Or as the source that answer engines, creators, and buyers keep coming back to
The operators who win the next few years won’t be the ones who memorize every new algorithm tweak. They’ll be the ones who treat AI, SEO, and media not as separate crafts, but as parts of a single, answer-first growth system – and who have the judgment to know what’s worth changing, and what isn’t.