For years, the SEO industry has been locked in a high-stakes debate: Does "topical authority"—the practice of deep-diving into a specific subject to signal expertise—actually influence generative AI search, or is the AI a blank slate that evaluates every query in a vacuum?
That debate was effectively settled this July by Kevin Indig. In a comprehensive analysis published on his Growth Memo Substack, Indig leveraged six months of proprietary data from Semrush (covering January through June 2026) to determine how ChatGPT decides which brands to recommend. The results are a wake-up call for marketers: Topical authority is not just a legacy SEO myth; it is a critical, durable asset in the age of AI. However, the data also reveals that the "land grab" for digital dominance is far from over.
The Anatomy of the Study
To reach these conclusions, Indig analyzed a massive dataset comprising 1,094 U.S. categories. Across these categories, he processed five distinct prompts per niche, totaling more than 50,000 brands and over 600,000 citations.
The study categorized brand presence into three buckets:
- Clear Owners: Brands appearing in at least four out of five test prompts, maintaining a lead of five percentage points or more over their closest competitor.
- Emerging Leaders: Brands leading at least three prompts without reaching the "Clear Owner" threshold.
- Unsettled Categories: Categories where no brand captures even three of the five test prompts.
The findings offer a rare, empirical look at how LLMs (Large Language Models) prioritize information. While authority is indeed a factor, the distribution of that authority across the market is surprisingly fragmented.
A Wide-Open Frontier: The Data Breakdown
The most striking revelation of the study is the state of the current digital landscape. As of June 2026, the market is remarkably unsettled. Only 15.2% of the categories analyzed had a "clear owner." Conversely, over 53.7% were "open fields," characterized by multiple contenders with no single brand holding a dominant grip.
When Indig cross-referenced these findings with estimated AI search volume, the opportunity for brands became even more apparent. He split the 1,094 categories into two equal groups by volume. The higher-volume half—which accounts for a staggering 98% of total AI search demand—boasts an owner rate of just 11.3%.
The implication is clear: 89.3% of the total estimated AI search demand resides in categories that currently have no clear winner. For CMOs and SEO practitioners, this represents a massive, fleeting window of opportunity. The categories that hold the most financial value are, quite literally, the ones that have yet to be claimed.
The Durability of the Lead
Indig’s research also addressed the "stickiness" of AI-driven market share. Once a brand secures a dominant position, it becomes incredibly difficult to unseat. By tracking month-over-month leadership changes, Indig found that a "clear owner" retained its top spot in 90.4% of comparisons.
However, the "dethroning" of a leader is not impossible; it is simply rare, and it usually occurs when the incumbent’s lead is thin. In categories where a leadership change occurred, the median lead going into the shift was a razor-thin 1.3 percentage points. Where the leader held on, the median lead was 2.9 points.
This data suggests that momentum alone is not enough to secure a category. A brand with a one- or two-point lead is not "winning"; it is in a knife fight. Brands that fail to solidify their lead with consistent, cross-prompt coverage are effectively leaving the door open for faster-moving competitors to overtake them.
The Myth of the Citation
Perhaps the most counterintuitive finding in Indig’s analysis is the relationship between citations and brand mentions. Many SEOs have pivoted their strategy toward Generative Engine Optimization (GEO), focusing heavily on increasing citation counts to signal relevance to AI models.
Indig’s data reveals that this may be a strategic dead end. There is a weak, slightly negative correlation (-0.229) between being the most-cited domain and being the most-mentioned brand. In fact, the most frequently cited domain matched the most frequently mentioned brand only 20.8% of the time.
What this means is that while citations are a signal of presence, they are not a proxy for trust or preference. ChatGPT does not necessarily favor the brand with the most backlinks; it favors the brand that appears most consistently across the "full stack" of a user’s intent:
- The Definition: What is this?
- The Comparison: How does it stack up against alternatives?
- The Alternatives List: Who are the players?
- The Use Case: How is it applied?
- The Buying Decision: Why choose this brand?
Implications for Modern SEO
For years, agencies have sold the concept of "topical authority" as a way to structure content calendars. Indig’s study provides the first hard evidence that this is not merely a "convenient story" for clients, but an algorithmic reality.
1. Shift from Backlinks to "Answer Presence"
Practitioners must move away from the obsession with raw citation counts. If your GEO strategy begins and ends with "get more citations," you are likely optimizing for the wrong metric. Instead, the goal should be to be the brand that the model consistently references across all five types of intent prompts.
2. The Cost of Inaction
The 89% of demand currently sitting in "unsettled" categories will not stay that way forever. As AI models become more refined and the training data matures, the "barrier to entry" for claiming a category will rise. Brands that are currently treating category focus as optional are effectively ceding nine-figure market segments to competitors who are willing to show up consistently.
3. The New "Formula"
While Indig is careful to avoid suggesting a "magic formula," he identified common traits among winners:
- Higher branded search volume (55.7% of pairs).
- Higher overall organic traffic (48.4%).
- Higher Semrush Authority Scores (52.5%).
These are not the causes of AI dominance, but rather indicators of a brand that has successfully established a multi-channel footprint. The AI is not rewarding these metrics directly; it is rewarding the brand that is ubiquitous enough to appear in every corner of the research funnel.
Conclusion: The Closing Window
The lesson from the first half of 2026 is that AI search has not leveled the playing field—it has simply changed the rules of engagement. While it is true that a "generalist" brand could potentially walk into a category and challenge an incumbent, they cannot do so through sporadic content efforts.
Topical authority is, at its core, a measure of reliability. When an LLM acts as an assistant to a user, it seeks out the most "reliable" source for that specific intent. Reliability is built through depth, consistency, and a presence that spans the entire user journey.
For organizations, the mandate is clear: identify the categories where you have a foothold, map out the five core prompt types (definition, comparison, alternatives, use case, and purchase), and ensure that your brand is the default answer for every single one. The market is currently wide open, but as Indig’s data demonstrates, the doors are beginning to close. Those who treat this research as a blueprint for the next two quarters will define the competitive landscape of 2027; those who continue to chase vanity metrics like citation counts will likely find themselves locked out of the room entirely.
