Retrieval, then synthesis
Most AI answers are built in two steps. First the engine *retrieves* candidate sources (from its index, a live search, or both). Then it *synthesizes* an answer and decides which sources to name. You can influence both steps.
What gets retrieved
Retrieval favors content that is clearly relevant to the question and easy to parse. That means:
- Pages that directly answer specific questions, in plain language
- Clear headings that map to how people actually ask
- Content that loads fast and isn't buried behind scripts
What gets cited in the answer
Among retrieved candidates, engines tend to name sources that are:
- Specific and factual (concrete claims beat vague marketing copy)
- Trustworthy (recognizable entity, real expertise, corroborated elsewhere)
- Consistent with other sources (agreement increases confidence)
- Recent, when the question is time-sensitive
The practical takeaway
You can't control the model, but you can control your inputs: answer real questions directly, be specific and accurate, make your identity unambiguous, and earn corroboration across the web. The rest of this course turns each of those into concrete actions.