a builder's codex
codex · operators · Kevin Indig · ins_kevin-indig-reasoning-lift-two-universes

High-reasoning AI mode and standard mode share only 25.6% of cited domains; AEO built for one misses three in four sources the other cites

By Kevin Indig · Growth advisor and SEO researcher, Growth Memo · 2026-05-18 · research · Reasoning Lift

Tier A · TL;DR
High-reasoning AI mode and standard mode share only 25.6% of cited domains; AEO built for one misses three in four sources the other cites

Claim

High-reasoning AI mode cites a fundamentally different set of domains from standard mode. The overlap is 25.6%. AEO programs built for one population leave a 74.4% non-overlap unaddressed.

Mechanism

High-reasoning mode fires 4.6x more search queries per answer. Those additional queries surface domain-specific primary sources, expert interviews, and structured evidence that keyword-optimized content does not earn. Standard mode prefers high-volume, well-linked pages. Reasoning mode prefers depth, evidence density, and source credibility. The retrieval population is different, so the citation population is different.

Conditions

Holds when: AI systems offer a reasoning or extended-thinking mode. Applies to queries where users choose or are defaulted into high-reasoning. Fails when: reasoning-mode rollout is limited to a small user share, or when queries are simple enough that standard and reasoning modes produce identical retrieval behavior.

Evidence

Indig's May 18 Growth Memo research measured citation behavior across minimal and high-reasoning AI search modes. High-reasoning fires 4.6x more queries per answer, cites 99 domains minimal mode never touches, and lifts citation rates from 50% to 68%. The domains cited are largely different sets.

"The brand that wins under minimal reasoning is not the brand that wins under high reasoning."

Signals

Counter-evidence

Reasoning-mode rollout is uneven across AI products. Audiences that skew toward casual users may still be primarily served by the standard-mode citation universe. Some standard-mode content may naturally earn reasoning-mode citations without deliberate optimization.

Cross-references

Open the interactive view → View original source → Markdown source →