How AI search changes SEO
AI search is not separate from SEO, but it does change how an agency should approach visibility. As Anirudh Agarwal explains, large language models answer questions using both pre-training data and information retrieved from the web.
Agencies cannot directly change a model’s pre-training data. They can influence what the model retrieves, which means they need to influence the pages, documents, and third-party sources that shape its answers.
This work is platform-specific and can also be model-specific. A strategy that works in ChatGPT may not work in Gemini or another AI search product. Google AI Overviews and Gemini also follow their own citation patterns.
Prompts matter more than isolated keywords
For AI search, agencies need to think in prompts rather than only in keywords. Customer conversations can reveal the questions, fears, needs, and contexts that should guide a content strategy. Useful sources include:
- Sales-call transcripts
- Support chats
- Reddit discussions
- Customer conversations
- “People also ask” queries
- Keyword data
ChatGPT also breaks a question into multiple related searches. If someone asks for the best CRM, it might search for CRM options for a senior engineer, pricing comparisons, and other related questions. It then synthesizes and re-ranks the information before answering.
Agencies should identify these secondary queries and optimize for them as well.
Find the sources AI already cites
One approach is to run target prompts, collect the citations, and categorize the cited pages by type. AI systems might repeatedly cite comparison pages, guides, listicles, pricing pages, documentation, or institutional sources.
This analysis shows an agency which formats are useful for a topic and which domains are already shaping the answer. The agency can then create relevant content and pursue mentions in the same places.
Reverse-engineering the queries can be even more useful because some query fan-outs are not disclosed. The citations can be sorted into buckets such as pricing, implementation, API availability, use cases, and alternatives.
Consensus is the new tiebreaker
Basic content has become easier to produce. Proprietary research and information gain still matter, but when two sites offer similarly good content, third-party consensus can become the deciding factor.
Consensus means that several independent sources repeatedly say the same thing about a brand. If AI systems scan ten pages and six or seven identify a brand as the most affordable option for a particular need, that repeated agreement strengthens the signal.
This makes context more important than optimized anchor text alone. AI systems can understand the surrounding discussion and whether a mention is favorable, not merely whether a link exists.
Mentions and links still matter
AI search has not reduced SEO to selling citations. ChatGPT uses Bing’s index, while Gemini uses Google’s index, so a brand must remain present in the relevant search indexes. There is substantial overlap between pages that rank traditionally and pages cited by AI systems.
A citation is not necessarily better than a link, and links are not inherently bad. However, brand mentions can carry considerable weight because models may learn from text without relying on the anchor or destination URL.
The practical implication is that agencies should treat mentions, links, citations, and index visibility as related parts of one strategy rather than separate deliverables.
Plan mentions around the customer journey
Agencies should create awareness-stage prompts around the brand across top-, middle-, and bottom-of-funnel topics. Run those prompts regularly, collect the citations, and identify the sites AI already favors for the brand.
Then pursue placement across those sources. The most common method is direct outreach. An agency may ask to be added to an existing listicle, submit an article, or request another type of brand insertion. These placements generally cost money.
Anirudh says the average price for a regular link or mention is often around $300 to $400, although placements on authoritative sites can reach $2,500 or $3,000. Publishers usually base prices on traffic, although sites that know they are cited by AI systems may charge more than their traffic alone would justify. Typical terms are one or two years, while lifetime placements are possible.
YouTube is a priority
YouTube should be a major focus for AI visibility. According to Anirudh, YouTube is currently the most cited website for Google AI Overviews, and Gemini also frequently cites Google-hosted videos because it can read their transcripts.
A brand mentioned accurately in a transcript can build context across multiple videos. Topical relevance matters more than the creator’s general audience size. A creator who consistently publishes useful content about a relevant topic may be a better target than a larger creator covering general business topics.
Agencies should therefore publish relevant videos and pursue mentions on established channels in the same vertical. They can track which channels are cited for target prompts, then ask to appear on a podcast or publish a video that discusses the brand in context.
A practical 80/20 approach
Agency owners can start with a straightforward process:
- Define the brand, target market, topics, competitors, and likely buyer personas.
- Build prompts for top-, middle-, and bottom-of-funnel questions.
- Run those prompts across the relevant AI platforms.
- Track responses, citations, visibility, sentiment, and source types.
- Identify gaps where competitors appear but the brand does not.
- Create or improve pages that answer the missing queries.
- Reach out to the domains and creators already cited in the relevant topics.
- Repeat the process and monitor changes over time.
Agencies can use an AI visibility platform or perform the work themselves. The goal is not to use every available tool, but to use the information to choose the right content, sources, and outreach opportunities.
Use customer conversations to guide content
Sales calls and customer questions remain important. AI conversations tend to be conversational, so understanding customers’ fears, doubts, and recurring questions can reveal the language people use when seeking help.
That information can guide both on-site content and third-party content. If an agency identifies a question customers ask frequently, it can publish a useful guide, appear on a relevant third-party site, or address the issue in a video.
What this means for agencies
AI search is pushing SEO toward a broader discipline. Agencies still need sound technical work, index visibility, useful content, links, and citations. They also need a coordinated effort to establish context and consensus around their clients’ brands.
For an agency owner, the most actionable advice is to start by publishing relevant YouTube content, tracking which sources AI cites for the topics that matter, and building outreach around those sources. AI is making it harder to rely on generic content alone and easier to benefit from doing SEO, branding, and business development together.