Being the answer: search’s new front door

What it’s actually for

The buyer at stage 02 used to google “best marketing automation for mid-market” and skim ten blue links. Increasingly, they ask an assistant the same question and get three names in a paragraph. If you are not one of the three, you were never in the room — there is no page two of a chat answer.

AI answers search is the channel of being retrievable. When ChatGPT, Perplexity, Gemini or Google’s AI overviews assemble an answer about your category, they pull from sources they can find, parse and trust. The channel’s job is making sure that what they pull says your name, accurately, in the contexts where you deserve to be named.

This is not a passing fashion to wait out. It is the same buyer doing the same Looks-stage research through a new front door. The behaviour moved; the funnel stage did not.

Where it fires

AI answers search → Looks and Sizes up. The same intent as classic search, through a new front door.

A targeted, earned channel.

On the marcomms map, AI answers search fires at Looks and Sizes up — exactly where classic search fires, because it is the same intent being served differently. At Looks, the buyer asks the category question and the assistant drafts the consideration set. At Sizes up, they ask the comparison question, and the assistant summarises your proof, your pricing and your reputation whether you participated or not.

The numbers that matter

This channel’s measurement is young, so the discipline matters more, not less. The activity layer: run the retrieval checks yourself — ask the major assistants your buyers’ actual questions, monthly, and log whether you appear, what they say, and what they cite. Track referral sessions arriving from assistant surfaces, and watch branded search volume as the downstream tell: people who met you in an answer go and look you up.

The commercial layer is the same as organic search has always owed: movement from stage 02 into stage 03. Citations are this channel’s impressions. Pleasant, but not yet pounds.

When not to use it

Do not stuff content with question-shaped filler hoping the models hoover it up. Generic slop is the failure mode — the assistants are pattern-matching for specificity and authority, and they are better at smelling padding than most human readers. Do not abandon search fundamentals either: the models retrieve from the same crawled, structured, source-worthy web that classic SEO built, so entity clarity, specific claims and pages worth citing serve both doors. And do not let an agency sell you “AI SEO” as a separate magic discipline at a separate magic price. It is mostly the old discipline, done properly, with a new reader.

One row of twenty

This is one row of the marcomms map: every channel, every stage, every targeting mode, on one page.

The full system lives on the Methodology page.