Indexing no longer guarantees anything

Duane Forrester wrote something on Search Engine Journal that a lot of people are afraid to say out loud: indexing no longer guarantees traffic. Not rankings — traffic, period. A page can be perfectly indexed, have the right keywords, solid backlinks — and still get zero visits, because the AI already summarized its content right there in the results, without sending a single human your way.

I don't see this as a disaster. It's the new reality, and you need to adapt to it the same way you once adapted to mobile-first indexing or Core Web Vitals. The rules shifted again. That's it.

Content now competes to get included, not to get clicked

The old task was simple: write something Google would consider relevant to the query. The task now is harder — write something the AI actually wants to pull as a source for its answer. These are two different games. In the first one, whoever matches the keywords best wins. In the second, whoever has something that can't just be paraphrased wins: a specific expert take, a comparison backed by real numbers, a conclusion nobody else has published.

If your page is just another generic "what is X and why it matters" piece, it has no shot. The AI will build the same answer out of ten similar sources, and your site will be a random footnote in that pile — if it makes the list at all.

Your site has to work on two fronts at once

Here's where I fully agree with Forrestier: a site today needs to do two things simultaneously. Be a source of structured data the AI can easily pull and use. And be a conversion point for people who already read the AI's answer and came to you for details or to take action.

That means Schema.org isn't a checkbox anymore — it's a working tool: markup for your brand, products, services, reviews. The easier it is for a machine to parse your data's structure, the better your odds of making it into the summary. And at the same time, your page needs to be ready for a visitor who already knows the basics and came looking for a solution, not an explanation of "what this is."

A simple test for every page

Forrester proposes an honest filter: can the AI give a good answer without this specific page? If yes — the page isn't earning its place as is. Either rebuild it so it contains something that can't be synthesized, or cut it and stop wasting resources on it.

I'd run this test across the entire site, not just a sample. Most content sections on a typical project are exactly the kind of generic material an AI will paraphrase without a second thought. Unique comparisons, honest expert conclusions, a real opinion on a contested question — that's what people will still come to the original source for, because a neural network still can't convincingly fake a personal stance.

Your analytics need to see the new channel

Here's a practical point that's easy to miss: you need to track AI-search traffic separately from organic and direct. Otherwise you have no idea whether this whole content overhaul is actually working. Without that split in your analytics, you're just guessing — and guessing is a bad strategy for a business that wants to outlive a single season.

What to actually do about it

Build content around real needs — comparisons, choosing between options, expert positions — instead of around keywords that used to pull decent search volume. Add markup so machines can lift your data more easily. And above all, get your site ready to receive people after the AI's answer, not before it. That's a different funnel, and you need to rebuild it now, not after your classic search traffic has already cratered.

If you want to figure out exactly where your site is losing visibility in AI search and how to rebuild your content around the new logic — come talk to us during an audit.