Why ChatGPT Recommends One Brand and Ignores the Rest

ChatGPT recommends one brand and ignores the rest because, in a 1,655-category study updated October 6, 2026, the number-one name carries 53.3% of that engine's category recommendations on a median list of 23.9 brands. Google AI Overviews picks the same number-one brand in 93.0% of those categories, then keeps naming further down a longer list.

I'm William Spurlock, founder of Spurlock Studios LLC and an AI Systems Architect and Fractional AI CTO. I've built 600+ automations with 500+ still live, put 20,000+ hours into agentic systems, and watched that work save clients 35,000+ hours. None of those receipts make a brand the number-one name. A page a buyer can check might. This post does not pretend the hours do the job.

This page is the shortlist math. It is not the retrieval check in why ChatGPT recommends your competitor instead of you. It is not the seven-day log in how to audit a ChatGPT shortlist in a week. The build that earns the name lives in how to get ChatGPT and Perplexity to recommend your business. I only want three labels here: the name, the also-ran, or absent.

I read the shortlist study and Forrester's April 30, 2026 post on October 7, 2026. Every hard number below is from one of those two pages, or it is arithmetic I label as arithmetic.

Why does ChatGPT recommend one brand and ignore the rest? #

ChatGPT's number-one brand carries 53.3% of the recommendations inside its category, so most of that engine's mention pile sits on one name and every later name splits what is left. The ignore comes from a short list with one heavy leader. It is not a grudge, and it is not the first line of a chat you ran this morning.

The shortlist study, updated on the page October 6, 2026, measured ChatGPT (OpenAI) and Google AI Overviews across 1,655 buyer categories from October 19, 2025 to April 25, 2026, spanning 48,302 distinct brands (shortlist study). A category had to draw at least 20 distinct recommendations on both engines. A brand named twice inside one answer counts once, because the count is once per answer, brand, and need. The page says this compares aggregated category leaders, not individual answers. It was published June 30, 2026.

The same page is where the concentration numbers live:

What the study counted	ChatGPT	Google AI Overviews
Categories in the matched sample	1,655	1,655
Share of categories with the same number-one brand	93.0%	93.0%
Median brands named per category	23.9	36.4
Share of that engine's category recommendations held by its own number-one brand	53.3%	41.5%
Mean overlap of the full recommended lists	0.25	0.25

Read 53.3% as the study wrote it. If you line up 100 counted recommendations in a ChatGPT category, the number-one brand accounts for 53.3 of them. That 100 is the definition of the percentage. It is not a category I went out and counted. Google's number-one brand accounts for 41.5 of 100 on the same kind of reading. The rest of Google's longer list holds more of the pile than the rest of ChatGPT's list does.

What I will not conclude from that table:

• That a single screenshot is the 1,655-category result. The page is explicit that these are aggregated leaders.
• That brand 14 on Google was "close." Close is not a column in the study. The column that transfers is number one.
• That the 7% of categories where the engines disagree are the plan. That split is 116 categories out of 1,655. I treat it as the exception, not the brief.
• That overlap of 0.25 means the lists are useless. It means they share a leader far more often than they share the field.

I think owners hear "ChatGPT ignored us" and picture a model with a temper. The pattern is duller. The two engines crown the same winner in 93.0% of these categories. They do not crown the same bench. If you are not that shared winner, ChatGPT's shorter median list is where you vanish first. Google can still print a long roster and look fair. You were not the name.

What does it mean to be the number-one name in a ChatGPT answer? #

In this study, the number-one name is the brand with the most recommendations in a category after repeats inside a single answer are collapsed. It is not whoever happened to be typed first in a chat you opened today. One prompt can lead with a different brand. The study will not move because of that prompt.

I keep two clocks apart, because mixing them is how a meeting goes bad:

Clock	What "number one" means	What it cannot tell you
One ChatGPT answer, search on	The first brand in that answer, on that account, for that wording	The 53.3% category share
The shortlist study	The brand with the most counted recommendations in a category across the window	What your prompt returned this morning
A Google result you already rank for	A page Google chose to show	That ChatGPT must name the same URL

How I say it when someone Slacks me a screenshot:

1. The first brand in that answer is the name for that answer only.
2. If our name shows up under that brand, we are the also-ran for that answer.
3. If our name never shows, we are absent for that answer.
4. None of those three labels is the 93.0% agreement figure. That figure needs the whole category sample.
5. I do not average the three labels into a score. A score hides which seat we actually hold.

A screenshot and the study can both be true. The screenshot is what that account saw. The study is how recommendations piled up across 1,655 categories through April 25, 2026, with 48,302 distinct brands in the set. People paste one answer into a deck and call it the market. I will not. If the deck needs a market number, the market number is the study, with the date on it, and the deck has to say "aggregated leaders."

The problem is narrower than the panic suggests. Being the name in one answer is a lead worth keeping only if the cited URL is a page you want quoted. Being the name in the study's sense means you already collect the heavy share of counted recommendations in that category. Most businesses I talk to are asking about the first clock while quoting a statistic from the second. I stop the meeting until we pick one.

Why does Google name more brands in an AI Overview than ChatGPT does? #

Google AI Overviews names a median of 36.4 brands per category in this sample, and ChatGPT names a median of 23.9, so Google's answer has room for names ChatGPT never reaches. The extra names are not a second copy of ChatGPT's list with a tail stapled on. Mean overlap of the full lists is 0.25.

The length gap and the share gap move together:

Fact on the study page	Number	How I use it with an owner
Google median list	36.4 brands	More seats exist. The winner is still the seat that matches ChatGPT in 93.0% of categories.
ChatGPT median list	23.9 brands	Fewer seats. Missing here is the common way to be "ignored."
Google number-one share	41.5%	The leader takes less of Google's pile, so the rest of that longer list still holds a real share.
ChatGPT number-one share	53.3%	The leader takes more of a shorter pile. Second place is a thinner remainder.
Categories with a different number-one brand	116 (the page calls this 7%)	Real, and not the base case. I do not plan Q4 around the exception.

A longer list is not the cash register. Forrester's April 30, 2026 post, which I read on October 7, 2026, cites the firm's February 2026 Consumer Pulse Survey: 71% of consumers reported using Google in the past month to search for products they are interested in buying, and 26% reported using ChatGPT in the past month to search for products they are buying (Forrester, April 30, 2026). The same post says Alphabet's Q1 2026 revenue was $109.9 billion, up 22% year over year, including Google Search revenue of $60.4 billion, up 19% year over year. That is Forrester describing the quarter. I did not open Alphabet's filing myself, so I will not dress those figures up as my own read of the 10-Q.

What the longer Google list changes:

• A buyer who only looks at Google can see names ChatGPT never prints. Median 36.4 against median 23.9 is the room.
• The shared winner still matters more than the extra names, because the engines agree on that winner in 93.0% of the categories.
• Full-list overlap of 0.25 means you can be on one roster and missing from the other without anyone "breaking" a rank.

What it does not change:

• It does not turn 71% of consumers in a survey into a share of all searches. The pulse is about who used which product in the past month. I leave it as that sentence.
• It does not mean a long overview is a friendly market. Brand 20 on a 36-brand list is still not the name.
• It does not mean I should drop Google work because ChatGPT's list is the dramatic one. The revenue line Forrester cites is still Google Search.

I would not pull a working Google program because ChatGPT concentrates mentions. I would not celebrate a Google overview that names us at the bottom and call the category won. The study says the number-one slot is the part that usually matches. Second place is where the engines tell different stories. Second place on a 36-brand list is still second place.

How do I tell if my business is the name, the also-ran, or absent in ChatGPT? #

You are the name only if you are the first brand in that answer, the also-ran if your name shows up later, and absent if the answer finishes without you. Write one of those three words. Do not blend them into a grade. One answer is not the 1,655-category study, and a week of the same prompts is still not that study. The day-by-day version of the log is how to audit a ChatGPT shortlist in a week. This section is only the label.

Seat	What you saw	What I say in the notes
The name	Your brand is the first brand in the answer	Keep the cited URL. If the URL is wrong, the seat is not a win yet.
The also-ran	Your brand appears after at least one other brand	You are on the list and you are not the 53.3% seat. Do not call this a win.
Absent	The answer ends and your brand was never printed	You are in the ignore. A long Google list does not erase that line.

What I write in the margin before anyone argues:

• The exact buyer wording, saved, with search on or off noted.
• The first brand, spelled the way the answer spelled it.
• Our seat: name, also-ran, or absent.
• The URL attached to the first brand, if one was attached.
• Whether the Google overview for the same wording named us at all.

Use a buyer question, not a vanity question. A prompt that already contains your brand asks for a reputation check. The study is about buyer categories, the kind of wording a person uses before they have picked a vendor. If you only test prompts that contain your name, you are measuring recall. You are not measuring the shortlist.

If the first brand is a directory, a magazine, or a marketplace, your seat is still also-ran or absent. That publisher is a brand in the answer. It is not a neutral frame around the "real" companies. I have watched owners skip the directory, count themselves as first, and walk out with the wrong label. The label follows the order on the screen.

I do not invent a city, a trade, or a competitor in this post. Put your own wording in the blank. If you need the ten prompt shapes and the seven-day rhythm, use the audit post. Copying that calendar here would be a second version of a page that already exists.

What should I fix before the holiday question spike if ChatGPT does not name me? #

I do not have a dated public count of how many more buyer questions show up in November, so I will not invent a spike. Before that window I pick the seat and do the one job that seat requires. A new theme, a pile of new posts, and an ads buy do not change the 53.3% concentration figure.

Seat today	The job I would actually do	What I will not spend Q4 on
Absent	Make the offer quotable on a URL both engines can reach. The retrieval steps are the competitor post, not a second essay here.	A homepage rewrite that still opens on a slogan.
Also-ran	Get the proof that already makes someone else the shared number-one onto pages both engines can see. They agree on that winner in 93.0% of the study's categories.	Ten more posts that restate the slogan and never state the offer.
The name	Leave the quoted sentence alone unless the cited URL is the wrong page.	A refresh that turns a checkable sentence into a tagline.

If you are absent, start with the page. Why ChatGPT recommends your competitor instead of you is the retrieval check: what the answer can fetch, and what a blocked or empty page does to the citation. I am not repeating that procedure. If you are the also-ran, more pages on your own site do not automatically take the number-one share. The study's agreement says the winner usually shows up on both engines. A trick that only one product can see is a bet that you live in the 116-category minority. I would not bet the quarter on 116 out of 1,655.

If you are already the name, cosmetic editing is the risk. Someone "improves" the sentence the answer was quoting. The seat was the sentence. Protect it. Do not add a second offer into the same first paragraph and hope the model picks the right one.

What I cut from a Q4 list when the seat is wrong:

1. A visual redesign that does not change the first two sentences a buyer, or an answer engine, can quote.
2. A content quota. The study counts brand recommendations, not how many URLs you shipped.
3. An ads purchase aimed at the 53.3% share. That figure is not a rate card. The study is not an ad test.
4. A dashboard that blends one ChatGPT screenshot with the 1,655-category table and calls the blend a score.

The build, once the seat is honest, is still the one in how to get ChatGPT and Perplexity to recommend your business. Math tells you which seat you hold. It does not lay out the page. I want owners to stop treating those as the same assignment.

FAQ #

How does an owner audit a ChatGPT shortlist in a week? #

A week of the same buyer prompts tells you your seat. It does not reproduce the 1,655-category medians. Lock the wording, run it with search available, and mark name, also-ran, or absent. The day-by-day log, including the prompt shapes, is in how to audit a ChatGPT shortlist in a week. I will not paste a second calendar here and pretend it is a new method.

Does ranking on Google page one get a brand named in ChatGPT? #

A Google page-one ranking is a different measurement from ChatGPT's number-one share. The shortlist study says the engines agree on the aggregated number-one brand in 93.0% of categories, and that full-list overlap is only 0.25, so plenty of Google-visible names never become ChatGPT's leader. Rank and "the name" are not the same sentence. The retrieval check is in why ChatGPT recommends your competitor instead of you.

What does it cost to be number two on a ChatGPT shortlist through Q4? #

I will not invent a dollar loss for second place. On ChatGPT, the number-one brand carries 53.3% of that engine's category recommendations, so every later name is splitting what that share does not cover, inside a median list of 23.9 brands. Forrester's February 2026 pulse still has 71% of consumers using Google in the past month for products they are interested in buying, against 26% using ChatGPT for products they are buying. Through Q4 I treat second place as a citation gap beside that survey, not as a revenue number I made up.

How do Google AI Overviews and ChatGPT split the same buyer question? #

They usually keep the same leader and split the field. In the study, 93.0% of categories share the number-one brand, mean list overlap is 0.25, Google's median list is 36.4 brands, and ChatGPT's is 23.9. ChatGPT's leader holds 53.3% of that engine's recommendations. Google's leader holds 41.5%. Your screenshot can still disagree, because the page compares aggregated leaders, not individual answers.

What proof does a brand need before ChatGPT concentrates recommendations on it? #

The study reports the concentration it observed. It does not publish the documents that cause the 53.3% share. I will not invent a checklist and pin it on that page. Both engines crown the same number-one brand in 93.0% of categories, so the proof has to be visible to both, not stored where only one product can see it. The layers I use on a live site are in the week-audit post. The build is in how to get ChatGPT and Perplexity to recommend your business.

Should a business buy ChatGPT ads or earn the citation first? #

Earn the name seat before you pay for a placement the shortlist study did not measure. I did not open an ads product page on October 7, 2026, so I will not describe a current ad unit or a price. The 53.3% figure is a share of recommendations in the study's categories. It is not a line you can buy off a rate card. If a vendor says spend equals that share, ask them to show the ad test inside this study. It is not an ad test.

How often should I recheck whether ChatGPT still names a competitor? #

Recheck your own seat when you change the page, and recheck the category math when the study page updates. The sample window closes April 25, 2026. The page I read was updated October 6, 2026. That is not a Monday rank tracker. Running one prompt again the same afternoon is a mood, not a trend. The weekly cadence for a live log is the audit post. Keep the prompt text identical or the comparison is junk.

Does getting cited in AI chatbots improve SEO rankings too? #

I do not have a study that ties a ChatGPT citation to a move in Google rank, so I will not claim a ranking lift. Forrester's April 30, 2026 post still describes Google Search revenue of $60.4 billion in Q1 2026, up 19% year over year, as they cite it. A citation and a rank are two logs. I keep both. I do not add them into one score and call the sum a strategy.

Get an AI-visibility-ready site #

I build AI-visibility-ready sites for owners who can see the first brand in the answer and know it is not theirs. The work is a page both engines can repeat, a seat log with three words, and a refusal to buy a share the study did not sell.

If you want that site, get an AI-visibility-ready site built. Bring the prompt, the first brand, and whether the Google overview still names you. That is the brief.