If you still think “traffic” mostly means Google, you’re reading last year’s playbook.

This week pushed the trend into plain sight: answer engines are no longer just research tools. They’re becoming discovery layers, comparison layers, and increasingly, transaction layers.

OpenAI made that explicit in its new product discovery announcement for ChatGPT, saying more shoppers are starting inside ChatGPT, that major retailers are integrating into its commerce protocol, and that Shopify merchants are already flowing into discovery through the Shopify Catalog.

At almost the same time, OpenAI’s latest ChatGPT release notes on shopping updates said product results are becoming richer, fresher, and easier to compare side by side.

Google is moving in the same direction. In its 2026 letter on ads and commerce, Google said Search is shifting from keywords toward conversational, multimodal discovery and that AI Mode creates new opportunities for businesses to fit naturally into the conversation. Then, just two days ago, Google announced in its Search Live global expansion update that people in more than 200 countries and territories can now have voice-and-camera conversations with Search in AI Mode.

That is not a feature trend.

That is a distribution shift.

I’ve always loved clean funnels on whiteboards.

I’ve almost never loved them in real life.

In real buying behavior, people do not move politely from awareness to consideration to decision while waiting for your nurture sequence to catch up. They bounce. They compare. They ask weird questions. They pressure-test the category in private.

Answer engines make that behavior even faster.

A prospect can now ask:

And the engine can return a synthesized shortlist before your SDR has even opened HubSpot.

That matters because shortlist formation is where a lot of the game is already won.

This is the part most teams haven’t operationalized yet.

Your website can still rank.

Your ads can still run.

Your outbound can still work.

But if answer engines are summarizing the category, then being merely indexed is not enough. You need to be recommended, or at least consistently cited.

That’s why HubSpot’s latest piece on answer engine optimization caught my attention. It argues that AI-powered search is compressing the customer journey, and cites data showing 72% of consumers plan to use AI-powered search for shopping more frequently.

That number feels directionally right to me because it matches what founders are already seeing in the wild: buyers are arriving with stronger opinions, narrower shortlists, and much less patience for generic marketing.

This is where people drift into abstract “AI will change everything” talk.

I’m allergic to that.

The useful question is: what should a GTM team actually do on Monday morning?

Here’s my answer:

Your site should explain:

If an answer engine can’t parse it, it can’t recommend it well.

The old content move was “What is CRM?”

The new move is “Best CRM for consultants,” “HubSpot vs. Pipedrive for startups,” or “Email outreach tools with good deliverability for agencies.”

That’s not just SEO. That’s shortlist engineering.

Reviews, customer evidence, benchmarks, and category mentions matter more when a model is stitching together an answer from multiple signals.

If your best proof only lives in your founder’s head, you are invisible in the places that now matter.

Track it. Prompt-test it. Monitor which categories you appear in. Fix weak pages. Improve structured data. Refresh stale claims. Build assets that answer specific buyer-context questions.

If you wouldn’t ignore paid search, don’t ignore AI discovery.

The old homepage was a storefront.

Now the storefront is increasingly the answer itself.

That means your GTM job is changing from “get found” to “get chosen inside the recommendation layer.”

That’s a harder game.

It is also a much better one.

Because once you accept that shift, you stop obsessing over vanity traffic and start building the assets that actually win the shortlist.