---
title: "In 13 Categories Where All Six Buyer Questions Publish, No Domain Wins All Six"
description: "Today's Machine Relations Index publishes 81 buyer-question leaderboards. Forty-four different domains hold the #1 slot. Reddit holds 19 of them and YouTube 10 — and in every category where all six questions publish, the winner changes with the question."
canonical: https://authoritytech.io/curated/question-shape-decides-ai-citation-winner-2026
last-updated: 2026-09-19
---

# In 13 Categories Where All Six Buyer Questions Publish, No Domain Wins All Six

Today's Machine Relations Index publishes 81 buyer-question leaderboards. Forty-four different domains hold the #1 slot. Reddit holds 19 of them and YouTube 10 — and in every category where all six questions publish, the winner changes with the question.

Canonical URL: https://authoritytech.io/curated/question-shape-decides-ai-citation-winner-2026
Published: 2026-09-19
Author: Jaxon Parrott
Tags: Afternoon Brief, AI Search & Discovery, Measurement

"Who dominates AI answers in our category" is the question every brand asks first. Today's [Machine Relations Index](https://machinerelations.ai/index) release says it has no answer, because the premise is wrong. Across the 13 categories where all six buyer question shapes publish a leaderboard, the number of distinct domains holding the #1 slot is never one. It averages 4.5. In Fintech, Cybersecurity and Consumer Finance it is six of six — a different winner for every question a buyer asks.

The unit that decides who gets cited is not the category. It is the shape of the question.

## What today's release publishes

Release `mri_score_v2.0+2026-09-19+0cad03121f60`, generated 2026-09-19, observation window 2026-05-10 through 2026-09-19.

A segment is one subject category paired with one buyer question shape, and it publishes a ranked leaderboard only after clearing 10 observed answer runs across 7 distinct observation dates. Today 88 of the 157 segments in the universe clear that floor: 81 on the six buyer shapes, 7 on the news shape. The rest are still collecting, which is a different state from zero and [was the subject of yesterday's brief](https://authoritytech.io/curated/ai-citation-rate-unavailable-not-zero).

Those 81 published leaderboards are the entire evidence base below. Every rank, rate and source-role label in this brief is read from the segment's own public page.

| Measure | Value |
|---|---:|
| Segments in the universe | 157 |
| Publishing a ranked leaderboard | 88 |
| On the six buyer question shapes | 81 |
| On the news shape | 7 |
| Distinct domains holding #1 across the 81 | 44 |

Six engines reported runs inside the window: [ChatGPT](https://chatgpt.com), [Claude](https://www.anthropic.com/claude), [Gemini](https://gemini.google.com), [Google AI Mode](https://blog.google/products/search/ai-mode-search/), [Google AI Overviews](https://developers.google.com/search/docs/appearance/ai-features) and [Perplexity](https://docs.perplexity.ai).

## The winner changes with the question

Thirteen categories have all six buyer shapes published. Here is the #1 domain in each, read from that segment's leaderboard.

| Category | Best tools | How buyers choose | Is it worth it | Problem-first | Top lists | Comparisons |
|---|---|---|---|---|---|---|
| [AI Infrastructure](https://machinerelations.ai/index/categories/ai-infrastructure) | medium.com | medium.com | reddit.com | reddit.com | spheron.network | youtube.com |
| [AI Security & Privacy](https://machinerelations.ai/index/categories/ai-security-privacy) | arxiv.org | teramind.co | witness.ai | youtube.com | ovaledge.com | arxiv.org |
| [AI Visibility & GEO](https://machinerelations.ai/index/categories/ai-visibility-geo) | youtube.com | youtube.com | reddit.com | youtube.com | youtube.com | hubspot.com |
| [Consumer Finance](https://machinerelations.ai/index/categories/consumer-finance) | forbes.com | bankrate.com | reddit.com | youtube.com | nerdwallet.com | acorns.com |
| [Consumer Health](https://machinerelations.ai/index/categories/consumer-health) | healthline.com | nih.gov | uclahealth.org | clevelandclinic.org | forbes.com | healthline.com |
| [Consumer Products](https://machinerelations.ai/index/categories/consumer-products) | forbes.com | youtube.com | reddit.com | reddit.com | chewy.com | youtube.com |
| [Cybersecurity](https://machinerelations.ai/index/categories/cybersecurity) | huntress.com | microsoft.com | paloaltonetworks.com | github.com | hoxhunt.com | exabeam.com |
| [Deep Tech & Hardware](https://machinerelations.ai/index/categories/deep-tech-hardware) | reddit.com | anysilicon.com | vlsishuttle.com | analog.com | unmannedsystemstechnology.com | reddit.com |
| [Education & Training](https://machinerelations.ai/index/categories/education-training) | reddit.com | reddit.com | reddit.com | linkedin.com | reddit.com | uopeople.edu |
| [Emergent Prosumer](https://machinerelations.ai/index/categories/emergent-prosumer) | reddit.com | reddit.com | reddit.com | reddit.com | digitalcameraworld.com | youtube.com |
| [Enterprise Software](https://machinerelations.ai/index/categories/enterprise-software) | erpresearch.com | erpresearch.com | erpresearch.com | reddit.com | procuredesk.com | erpfocus.com |
| [Family Software](https://machinerelations.ai/index/categories/family-software) | techradar.com | apple.com | reddit.com | reddit.com | apple.com | useboomerang.com |
| [Fintech](https://machinerelations.ai/index/categories/fintech) | airwallex.com | pxp.io | paymentsandrisk.com | chargeflow.io | stripe.com | wise.com |

Read down a column and you see a market. Read across a row and you see the problem with category-level AI visibility advice: in Cybersecurity, the domain that wins "best endpoint security tools" is not the domain that wins "is Palo Alto worth it," and neither is the domain that wins a problem-first search. Six segments, six incumbents, one category.

The nearest thing to a category-wide winner in the release is Healthcare Services, where [omnimd.com](https://machinerelations.ai/index/domains/omnimd.com) leads all three of the buyer shapes that published today — and three of six is the smallest sample in the table, because the other three shapes are still under the evidence floor.

## The source role that wins depends on the shape

Each leaderboard labels every domain with a source role from the Index's own taxonomy. Grouping the 81 #1 slots by that label gives the rubric.

| Question shape | Segments | #1 source role, most common | Second |
|---|---:|---|---|
| Best tools | 14 | Editorial publication (5) | Community and social platform (3) |
| How buyers choose | 14 | Other observed source (4) | Editorial publication (3) |
| Is it worth it | 13 | Community and social platform (7) | Other observed source (4) |
| Problem-first research | 13 | Community and social platform (7) | Search or media platform (3) |
| Top lists | 14 | Other observed source (7) | Editorial publication (4) |
| Comparisons | 13 | Other observed source (5) | Search or media platform (3) |

Three readings follow directly.

**The verdict questions belong to communities.** On "Is it worth it" and "Problem-first research," a community or social platform holds #1 in 7 of 13 segments each — 14 of 26 between them, against 5 of 28 on Best tools and How buyers choose. When a buyer asks whether a product is worth the money or how to solve a problem, the engine reaches for a forum, and [reddit.com](https://machinerelations.ai/index/domains/reddit.com) is that forum in 19 of the 81 segments.

**Best tools is where a publication can still win.** It is the only shape where editorial publications lead most often, at 5 of 14. It is also the shape with the widest leader rates: from 20.66% in Deep Tech & Hardware to 58.67% in Cybersecurity, median 34.21%. A shortlist question has a settled answer in a way a verdict question does not.

**Two platforms hold a third of everything.** Reddit's 19 and [YouTube](https://machinerelations.ai/index/domains/youtube.com)'s 10 are 29 of the 81 buyer segments, 35.8%, and the remaining 52 are spread across 42 other domains. Concentration at the top of the index is real and [has its own study](https://machinerelations.ai/research/how-concentrated-are-ai-citations-source-distribution-2026); what this cut adds is that even the two most concentrated platforms in the market lead all six shapes in exactly zero categories.

## The one shape a vendor's own site wins

Across all 81 buyer segments, a vendor-owned source holds #1 six times: [huntress.com](https://machinerelations.ai/index/domains/huntress.com) and [microsoft.com](https://machinerelations.ai/index/domains/microsoft.com) and [paloaltonetworks.com](https://machinerelations.ai/index/domains/paloaltonetworks.com) in Cybersecurity, [airwallex.com](https://machinerelations.ai/index/domains/airwallex.com) and [stripe.com](https://machinerelations.ai/index/domains/stripe.com) in Fintech, [hubspot.com](https://machinerelations.ai/index/domains/hubspot.com) on AI Visibility comparisons. That is 7.4% of the buyer-question leaderboards, and four of the six sit in two categories.

Then look at the news shape, where the same taxonomy is applied to the same engines over the same window.

| Shape | Vendor-owned share of top-ten slots |
|---|---:|
| How buyers choose | 6.4% |
| Problem-first research | 6.9% |
| Best tools | 7.1% |
| Is it worth it | 9.2% |
| Top lists | 9.3% |
| Comparisons | 13.8% |
| **News-driven citations** | **34.3%** |

News is the shape where a company's own newsroom is the most-cited class in the market, at more than double its best buyer-shape share. In Cybersecurity's news segment the top three domains are all vendor-owned. That is the measured version of a thing PR teams have suspected for two years: the press release moves the news answer and leaves the buying answer untouched.

It also sets the price of the news shape honestly. News leaders hold the lowest rates in the release — median 11.35% against 34.21% for Best tools — because a news segment runs hundreds of answers across dozens of dates and no single source holds many of them.

## What this changes for a brand

The practical consequence is that "get cited by AI" is six different projects, and only one of them is usually funded.

- **If the target is a shortlist question,** the incumbent is most often a publication, and the move is earned coverage in the specific outlet the segment page names — not any outlet with a big audience.
- **If the target is a verdict or problem question,** the incumbent is a community thread. No amount of owned content displaces it; presence in the thread does.
- **If the target is a comparison or top-list question,** the leader is usually a small specialist site the category has never heard of, and those are displaceable in a way Reddit is not.
- **If the target is news,** an owned newsroom is a legitimate instrument, and it will not move the buyer questions.

A single "AI visibility score" for a brand averages across all six and tells you none of this. The cost of that averaging is [measurable and has been measured](https://paralabs.ai/blog/why-ai-visibility-scores-differ-between-tools).

## Check one category before you spend

None of this needs a tool or a subscription. Take the category closest to yours, open its six segment pages, and write down the #1 domain and its source role on each. If the six names are the same, the category has a single incumbent and the strategy is one project. If they are several different names — and in all 13 categories with six published shapes the count runs from three to six — the strategy is that many projects, and the budget should say so.

Then read the engines' own documentation for how each one retrieves before it answers, because the retrieval step is the part any of this can reach: [OpenAI's web search tool](https://platform.openai.com/docs/guides/tools-web-search), [Claude's web search](https://support.anthropic.com/en/articles/11088861-using-web-search-on-claude-ai), [Google's AI features guidance for site owners](https://developers.google.com/search/docs/appearance/ai-features), [Google's AI Overviews help page](https://support.google.com/websearch/answer/13572151) and [Perplexity's prompt guide](https://docs.perplexity.ai/guides/prompt-guide). Six engines, six retrieval paths, and a leaderboard that already tells you which one is citing whom in your market.

## Sources and method

Every figure here is read from the public Machine Relations Index, release `mri_score_v2.0+2026-09-19+0cad03121f60`, on 2026-09-19. Method: all 157 category-and-shape leaderboard pages under `https://machinerelations.ai/index/categories/` were fetched over plain HTTP, all 157 returned 200, and the rank, source, source role and citation-rate columns were read from the page. The 88 pages carrying a `#1 of N` row are the published segments; the other 69 print a collecting notice instead of ranks. Counts of #1 slots by source role are counts over those 88 pages, 81 of them on the six buyer shapes. The machine-readable release is at [machinerelations.ai/data/machine-relations-index.json](https://machinerelations.ai/data/machine-relations-index.json) and every segment page links the domain profile behind each row.

<!-- absence-claim-enumeration: The claim that no domain leads all six buyer shapes in any category is an enumeration over the 13 categories whose six buyer-shape leaderboards all publish in release mri_score_v2.0+2026-09-19+0cad03121f60: ai-infrastructure (4 distinct #1 domains), ai-security-privacy (5), ai-visibility-geo (3), consumer-finance (6), consumer-health (5), consumer-products (4), cybersecurity (6), deep-tech-hardware (5), education-training (3), emergent-prosumer (3), enterprise-software (4), family-software (4), fintech (6); 58 distinct-leader observations over 78 segment pages, minimum 3, none equal to 1. Each is readable at https://machinerelations.ai/index/categories/<category>/<shape> for shapes best_x, how_choose, is_x_worth, problem_first, top_list, x_vs_y. The claim that vendor-owned sources hold #1 in 6 of 81 buyer segments is an enumeration of those 81 pages; the six are cybersecurity/best_x, cybersecurity/how_choose, cybersecurity/is_x_worth, fintech/best_x, fintech/top_list and ai-visibility-geo/x_vs_y. -->

Two limits worth stating. The source-role column is the Index's own deterministic taxonomy, not an editorial judgment: [apple.com](https://machinerelations.ai/index/domains/apple.com) and omnimd.com both carry the label "Editorial publication" at their domain profiles, which anyone can check, so read the column as classification rather than characterisation. And "Other observed source" is the residual class — it holds 23 of the 81 #1 slots, and a class-level reading that ignores it is describing a minority of the index.

Ranks are read from each segment's published leaderboard rather than re-derived, because the release uses competition ranking and re-sorting a stratum moves every tied domain. The same discipline is why the [question-shape swing on a single domain](https://paralax.ai/blog/wikipedia-ai-citation-rate-question-shape-2026) is a measurement and not an impression.

## Links

- [Curated Index](https://authoritytech.io/curated.md)
- [Home](https://authoritytech.io/index.md)