Understanding it
A question like “which CRM for a 50-person manufacturing SME” cannot be handled in one search. The engine looks up CRMs suited to SMEs, those that handle industrial quotes, pricing and reviews separately, then cross-references the results.
Google presented it at the launch of AI Mode: the question is broken down into subtopics and many searches go out at once. For its in-depth research feature, Google speaks of hundreds of searches before a report. Assistants such as ChatGPT work in a comparable way when they browse the web.
The direct consequence: competition plays out on each sub-question, not only on the main query. A site that answers a neighboring question precisely can be cited alongside much bigger players.
A concrete example
Illustrative example: an HR software vendor
An HR software vendor ranks well for “HR software”, but is rarely cited when someone asks an AI assistant “which HR software for a multi-site company with shift workers”.
It publishes pages that answer the likely sub-questions: multi-site management, shift scheduling, collective agreement compliance, payroll integration. Each one gives the assistant a precise piece of the answer to reuse, and a reason to cite the vendor.
The common mistake
Writing one giant page that covers everything
A 10,000-word page that skims everything is rarely the best answer to a specific sub-question. Cover the neighboring questions, each with a clear answer: well-titled sections on one page, or dedicated pages linked together when the topic warrants it.