How to Get Your Brand Recommended by ChatGPT, Gemini and Perplexity

To get recommended by ChatGPT, you need to be named on the pages it reads: the ones it learned from in training, and the ones it fetches when it searches. Which pages those are depends on the engine. By late September 2026 ChatGPT cited Reddit in about 1 in 43 answers, down from 1 in 5 in March¹, while in Ahrefs' November 2026 lists Perplexity⁶ and Gemini⁵ still cite Reddit more than any other site.

How ChatGPT picks the brands it names

ChatGPT has two levers. Memory is what the model learned in training, and it changes only when a new model ships. Search is the live part: ChatGPT "may search the web automatically when your question would benefit from current information", then names and cites what it found¹⁰.

Memory does most of the work. Ahrefs looked at 66.6 million free, logged-out ChatGPT responses from August and September 2026 and found search queries in only 14.4% of them¹. When ChatGPT does search, it sends shorter queries to search partners, and the ones OpenAI's help page lists are Microsoft and Shopify¹⁰, so check that Bing has indexed your key pages too.

Perplexity and Gemini pull the levers differently. Perplexity says it searches the internet "in real-time" when you ask, and every answer carries citations¹⁷, so its answers come from search. Gemini answered 64% of the questions in Ahrefs' 20-day citation test without citing a single source⁷.

Run the ask-twice test

Before you spend a dollar, find out which lever your category runs on. It takes about 10 minutes per assistant, and the idea comes from a practitioner on Reddit:

OpenAI's help pages on temporary chats¹¹ and web search¹⁰ have the current menu names. Repeat the test in Perplexity, which searches every time (skip the off runs), and in Gemini.

The sources each assistant leans on

Each assistant cites a different mix of sites. Ahrefs publishes a monthly top 50 per engine from its Brand Radar data, and the November 2026 lists, side by side, look like this:

For a brand, the table splits the work in two. In Gemini, Perplexity and Google's AI answers, Reddit and YouTube together take 35% to 43% of the top-50 citations, so community pages carry those answers. In ChatGPT, Wikipedia leads and the rest of the top 10 is testers, retailers and health and research bodies, so your own pages and your category's testers weigh more.

ChatGPT's row is the one that moved. Reddit went from about 1 in 5 of its answers in March to 1 in 43 by late September¹, and other trackers rank it lower than Ahrefs does (LLM Pulse had it outside ChatGPT's top 10 in the week to Nov 2⁸). The dataset-by-dataset story is in what changed in ChatGPT's Reddit citations.

Get mentioned on the pages searches land on

When ChatGPT cut back on Reddit, Wikipedia and YouTube, the citations went to research bodies, regulators, brand and support pages, and hands-on testers¹. Pages that only summarize someone else's work lost the most.

For a brand, that points at three kinds of page:

• Your own pricing, specs and support pages, with current prices written as text and a date. In September 2026 "price" appeared in 8.88% of ChatGPT's search queries, and 75.9% of its site: searches aimed at the most-searched sites went to brand sites¹.
• Comparison and "best X" pages other people write. Review platforms like Yelp and G2 came up often in recommendation answers in Peec AI's 30M-source study (March 2026)⁹. Get listed, and ask for stale listings to be corrected.
• Testers and reviewers in your category. Consumer Reports is ChatGPT's 3rd most-cited site in Ahrefs' November list, and RTINGS its 23rd². If your category has a tester, send them the product.

Threads and videos that keep getting read

Outside ChatGPT, the table's leaders are community pages: Reddit tops Gemini and Perplexity, and YouTube tops Google's AI Overviews. Google's AI answers are built from pages in Google's index¹⁶, so a thread that ranks for a buyer's question is in the pool they pick from, and it keeps getting read for as long as it ranks.

One comment, read for months

Three checks before you comment: the thread ranks for the buyer's question, it's still open, and the question in it isn't fully answered yet. How to find the threads Google and ChatGPT already use walks through the searches, and how to market on Reddit without getting banned covers writing something the subreddit keeps. Review and comparison videos work the same way (comments under the videos buyers watch).

Your ask-twice sheet also tells you whether threads matter in your category at all. If no assistant cites a thread or a video for your buyer questions, put the hours into pages instead. To see what a helpful comment looks like for one thread, paste it into the free Reddit comment generator.

Check it without fooling yourself

One screenshot proves almost nothing. Ahrefs tracked about 37,500 citations across five AI tools for 20 days: most of the sites cited one day were gone the next, and most of those came back later⁷.

So log two things separately, as the sheet does: whether you're named, and whether your pages are cited. Brand names held on longer than citations in the same study, and "best X" questions were among the least stable, so run each question on several days before you call a trend. How to track brand mentions in ChatGPT, Gemini and Perplexity has the free method and the trackers that do it at scale.

Measure · Move

What doesn't work

Buying a place in the answer

ChatGPT has shown ads since Feb 9, 2026, in the US first and in more countries since, to Free and Go users¹⁴. They sit below the answer, labelled as sponsored, and OpenAI's ads FAQ says ads "do not influence ChatGPT's answers"¹³. An ad can put you under the answer. It can't put you in it.

Submitting your site

There's no submission form for websites. OpenAI's instruction is to allow OAI-SearchBot and its published IP addresses, which makes a site eligible to be cited, not certain to be¹⁰. Products are the exception: merchants can apply to share a product feed for shopping in ChatGPT, which is live in the US, and Shopify and Etsy catalogs are already connected¹⁵.

Special files and markup

Google says you don't need new machine-readable files, AI text files or special schema to appear in its AI features, because Google Search doesn't use them¹⁶. Keep the structured data you have for rich results, and spend the time on the pages that get cited.

Where SlopMog fits

SlopMog works the thread side of the search lever, on Reddit and YouTube. It finds threads for your keywords every day and drafts a comment in each thread's tone. You edit it, approve it or skip it, or switch a site to hands-free, and comments go out on a staggered schedule, Reddit ones from aged Reddit accounts.

High-Value discovery is the part built for AI answers. It writes 40 buyer-style questions without your brand name, asks each one to Gemini, Claude, GPT and Grok with live web search, and ranks every Reddit and YouTube thread they cite. Those are the models' APIs, not the ChatGPT app your buyers open, so treat the list as the threads the models reach for, not a copy of a buyer's screen. It runs weekly, when you add a site, or on demand (how High-Value discovery works). What those runs cite, in aggregate, is in which threads the 4 models cite.

Plans start at $99 a month for 1,000 credits. A regular comment is 6 credits ($0.60) and a High-Value comment is 60 ($6), so that plan covers about 166 regular comments, and credits are spent only when a comment posts (what each plan covers). The simulation below runs the idea with your brand name in a made-up category: