Appearing in assistants' answers
Some of your customers no longer search, they ask. The answer arrives written out, with two or three sources cited, and the rest of the web does not exist. Being one of those sources can be worked on.
What assistants quote
Assistants cite pages that answer a precise question directly, in few words, with facts that are dated and attributed. They handle badly the pages that circle a subject without settling it, those whose content only appears once a script has run, and those with no identifiable author. A good page for an assistant looks a great deal like a good page for a reader in a hurry.
The work overlaps with ordinary search
The consequence is reassuring and disappointing at once: the work largely overlaps with organic search. A page that is fast, readable without JavaScript, structured and honest serves both. The difference lies in the shape of the answer. One question, a clear answer in the opening lines, then the detail. Not the other way round.
Where it starts
Before anything is written, two readings are taken. What the assistants already answer: we ask them the questions your customers actually ask, and note who is cited, on what, and from which sources. Then what your own tools already know, where they are in place: the queries bringing people in today, and the visits that already came from an assistant, recognisable by where they came from. The first reading says where you stand in the answers, the second says whether it is already bringing you anything.
When a competitor is cited, or the answer is wrong
The reading usually brings back two surprises. A competitor cited in your place on a question you are the specialist on: the answer does exist on your site, it is simply buried in a page about something else. And an inaccurate answer about you, a price that no longer exists, an address from three years ago, a service you no longer offer. The assistant is not inventing it, it is picking it up somewhere. The work is then to find where it comes from, directories, old press releases, forgotten pages on your own site, and to correct the source rather than argue with the answer.
What we act on
- Content readable without JavaScript: what a crawler cannot see does not exist for it.
- One question per page, and its answer in the opening lines.
- Facts that are dated, quantified and attributed, the only material an assistant can safely reuse.
- Structured data describing the company, its services and its pages.
- An author and an update date visible on substantial content.
- The files that explicitly allow or refuse the assistants' crawlers.
What can be counted, and what escapes
As for measurement, two things have to be separated. Visits arriving from an assistant can be counted: they carry its address as their referrer, and analytics tools now isolate them. What stays out of reach is the read without a click, where your name is cited and nobody comes. For that share, the work is done by convergence: regular tests on the questions that matter, brand mentions followed, direct visits rising without a campaign. The measuring results page details what actually gets counted.