AI Search Is Becoming B2B's First Touch
The B2B customer-to-brand journey is being upended by AI: 94 percent of B2B buyers use LLMs somewhere in the buying process (6sense). 84 percent of B2B tech queries show Google AI Overviews (BrightEdge). A potential customer's first experience of your brand is likely to be a paragraph that AI assembles based on what it can find on your company. Or, for general queries, like "best software for X," your company may not show up at all. The good news is that you can take action to shape your buyer's experience in the AI first touch. We detail more in this blog post.
Understanding AI citations for B2B
When AI overviews appear, clicks drop. Semrush data found the presence of Google's AI Mode pushes the zero-click rate to 93% (Semrush). But, the good news is that frequent AI users, which many buyers of B2B services are, say that they always or very often fact-check (TrustRadius). This extra level of research gives B2B companies another chance to reach buyers.
B2B companies often appear two ways in AI
B2B companies appear in AI search either by a direct-name search or in an answer to a "best-of" query. When your company is searched by name, AI has something to work with, pulling from the website, LinkedIn presence, and other indexed information to write a paragraph. The AI write-up might be outdated, incomplete, or just wrong, but your company is captured in the answer.
When someone does a category search, such as “the best software for X,” a different process occurs. The model looks for third-party citations that would render a company qualified for inclusion. If the information about your company is sparse or conflicting, you will likely not be included.
Test your brand's LLM presence
The best way to see how your company will appear in an AI answer is to test it in the LLM. The way to do this is run two test series: one in ChatGPT, Perplexity, Claude, and Gemini, and one in Google itself to trigger Google AI Overviews.
For each LLM and Google, run two searches:
Your company name.
The category query your buyer would likely use, phrased how they'd phrase it: "best [category] software for [use case/industry]."
For the first query, track:
Is my company's description and other information correct?
Does outdated information appear?
What citations are included?
For the second query, record three things:
Does my brand appear?
Where does it rank in the answer?
Which sources does the AI cite?
For Google, note whether a Google AI Overview appears, and capture the same information above.
Understanding Google AI Overviews
Not every query results in a Google AI Overview. The decision for AI inclusion is made algorithmically by Google, and phrasing can make a difference. We recommend testing different phrasings of each query to see if the AI overview will appear.
Google results can also vary by account log-in, location, and even A/B test rollouts. We recommend running AI Overview tests in both regular and incognito browser mode.
Location can change the results. AI Overview coverage differs by region, so search from the location your actual buyers are searching from, not just your own. You can log into a VPN to do an actual search, or use tools, such as Semrush, for an AI signal presence at scale.
Check your server logs for fetching and citing
AI companies run two different bot searches:
Training crawlers, such as GPTBot, ClaudeBot, and PerplexityBot in its indexing mode, visit your site to ingest content into the model's training data or search index. These searches happen continuously, and at scale, regardless of whether your content ever gets surfaced for anyone.
Retrieval/answer bots, such as OAI-SearchBot, ChatGPT-User, Claude-User, and Perplexity's live-query fetcher, hit your company's page in real time when a specific user's question makes your content relevant for an answer-pull. It is this step that actually pulls a citation.
Much of the bot activity on a site falls into the first category. Regularly check your server logs for activity to understand which ones are crawling.
Build a brand source map
Traditional SEO used backlink profiles to monitor site reach. AI search has something similar: a brand source map, which is a running list of the pages an AI tool names as source in query answers.
Building a source map
When you run your brand name and your core category queries across ChatGPT, Perplexity, Google AI Overviews, and Gemini, open the actual cited source for every answer. Then log the following four data points:
The URL and what type of page it is: your own site, a review site (G2, Capterra), a forum thread on Reddit, or an earned media article.
Whether the listing is current. Outdated pricing, positioning, and features can be repeated verbatim in the AI's answer if the information has not been updated.
Which other companies are named and in what order.
How often a source appears across different AI tools and queries.
Pay special attention to the fourth point. A single source that keeps recurring across multiple queries and AI tools is a heavy lifter and is a page source to prioritize.
How to engage sources
Once you know the pages and sources that are feeding the model's answer, look for interaction points:
For company listings, make sure the most accurate messaging, products, and industries served are represented. Edit or request changes for any outdated information.
For review sites like G2, make sure your company listing is complete, as incomplete listings often cause thin or generic descriptions in AI.
For third-party round-ups that either have incorrect information or don't include you at all, contact the author directly requesting inclusion and offering a briefing on the company and product.
For Reddit and other forums, proceed carefully, as the communities do not react well to self-promotion. A simple introduction of your company and product (if allowed by the moderator), either in response to a query or in general, can work.
Earned media requires time and the guidance of a seasoned public relations practitioner. If you are missing from regular coverage, a communication to the journalist about your company and what it brings to the industry can be helpful. Ideally, the communication would include market insights or data that points to larger trends.
Analyst reports are the result of ongoing relationships. Many analysts will take an introductory briefing on a company to learn about its solutions, and this is worthwhile.
Finally, the source map is not a static document. Re-run the map monthly to see changes that occur as AI tools re-crawl and re-rank what they cite.
Conclusion
The paragraph AI assembles about your company, or the answer it gives when your category comes up, should be viewed as your brand’s first touch with the buyer. This answer shapes decisions before your sales team ever gets a call. Your company can take an active role in how you appear by running prompt tests and building a citation source map, with an action plan. Consistent effort and monitoring will go a long way to accuracy in an AI answer.