Result Ranking and Relevance Signals
What influences the order of results a search API returns, and why it matters for agents.
What it means
Result order reflects the underlying search engine's own ranking algorithm — a mix of relevance signals, content quality heuristics, and (for some result types) commercial placement — not a neutral relevance score you can fully trust at face value. For an agent that treats the first result as automatically the best answer, understanding what's actually driving position matters, because paid or SEO-optimized content can outrank a more substantively correct but less search-optimized source.
In practice
A query about a product comparison might return a heavily SEO-optimized affiliate roundup article at position one, ahead of the manufacturer's own more accurate spec page at position four — an agent naively trusting rank order alone would summarize the less authoritative source first.
Tradeoffs
Re-ranking results using a domain-authority heuristic or source-type classification (favoring `.gov`/`.edu` for factual queries, deprioritizing known content-farm domains) improves output quality but adds complexity and can occasionally deprioritize a genuinely good result that happens to come from a less typically authoritative domain — it's a real tradeoff, not a strictly better default.