Test Gemini Brand Visibility: Clean-Session Matrix

By William Zhu · Cofounder, InfiniSynapse · Published: 2026-09-08 · Last updated: 2026-09-14 · Last verified: 2026-09-14 · Reviewed by the InfiniSynapse Data Team.

Reviewed by: InfiniSynapse Data Team · method review 2026-09-14. Gemini models and privacy Help 200s. No signed-in Gemini chat. No consumer transcript.

Trust / COI: About · Corrections. SEO Health is a commercial product from this team. This page is not a Gemini partner review.

TL;DR

Gemini brand visibility on this desk is two official 200s and zero signed-in chats.

Direct answer: Treat Gemini brand visibility as a new-conversation matrix, not as a blended Google rank. On 2026-09-14, https://ai.google.dev/gemini-api/docs/models answered HTTP 200 (145,896 bytes; title “Models | Gemini API | Google AI for Developers”). Gemini Apps Privacy Hub answered 200 (1,675,861 bytes). This tool URL also answered 200. No Gemini consumer session was opened. The sample cannot establish a mention rate.

A Gemini brand visibility slide that averages branded and unbranded prompts into one percentage is already mixing two jobs. Discovery and verification do not share a denominator.

What you will learn

• Which Gemini-related URLs we actually fetched
• Why a models catalog is not a chat log
• Which session fields a later matrix still needs
• When Privacy Hub copy is not a citation
• Where Perplexity, Claude, and locale hops live

What a clean session actually records

Key definition: Gemini brand visibility is one prompt × one new conversation × one recorded account and locale state. A source counts only when Gemini itself exposes it. A URL found later in Google Search is not that source.

People open Gemini brand visibility after a teammate pastes a chat that already contains the brand name in the prior turn. That chat is contaminated. Start a new conversation. Write the displayed model label, or write “not displayed.”

Gemini Apps Privacy Hub is the privacy boundary this page cites. We fetched it. We did not accept its terms as a substitute for a transcript.

SEO Health stays the deterministic evidence foundation for a public check on this URL. A later signed-in InfiniSynapse Server API long-task workspace can attach SSE /api/ai/events to an aivisibility job; that workspace still does not open Gemini.

Branded and unbranded stay split

A prompt that names the brand tests factual treatment. An unbranded prompt tests discovery. Gemini brand visibility that merges those rows makes prompted repetition look like organic inclusion.

Gemini Apps Help is a help root, not a session archive. Google AI Principles state Google’s published principles. They do not fill an empty matrix.

Account state is a column

Logged-in, logged-out, Workspace, and personal accounts are different rows. Locale and personalization toggles belong in the same record. Gemini brand visibility that hides account state cannot be replayed.

Public pages we could fetch

User-agent InfiniSynapse-SH1-desk/2026-09-14.

Three titles parsed. Three HTTP 200s. Zero chats. Gemini brand visibility still has no answer text to code.

NIST Generative AI Profile is the risk profile we used when deciding a public models page is not a measurement series. OECD AI Principles are a policy frame. They do not name InfiniSynapse in Gemini.

Figure: Every bar is a public page. None is a conversation.

What the models catalog is

The models page lists API model names. A consumer Gemini UI can show a different label. Gemini brand visibility that copies an API model id into a UI row without seeing that id on screen is inventing a field.

W3C language tags is why locale is written as a tag, not as “US English-ish.” Google privacy policy is the general privacy document. It is not a source card.

How to reproduce the three 200s

curl -sS -A 'InfiniSynapse-SH1-desk/2026-09-14' -o /tmp/gemini-models.html \
  -w '%{http_code} %{size_download}\n' \
  https://ai.google.dev/gemini-api/docs/models

curl -sS -A 'InfiniSynapse-SH1-desk/2026-09-14' -o /tmp/gemini-privacy.html \
  -w '%{http_code} %{size_download}\n' \
  https://support.google.com/gemini/answer/13594961

Expect 200 unless Google changes the docs host. Store the title. Do not paste a Gemini answer under this chart. Gemini brand visibility reproduction stops at the public pages until a reviewer actually opens a new chat and keeps the capture offline.

ISO/IEC 42001 overview is an AI management-system standard. It does not certify this matrix.

CLI boundary on the same morning

seo-health aivisibility on the Perplexity sibling printed unauthorized after a page fetch. This Gemini page did not spend that credit either. The command is not a Gemini session. Gemini brand visibility notes that quote scores[] without a taskId from a logged-in run are fiction.

Package pin infinitegrowth@0.1.1; -V printed 0.1.0. Record both.

What a later matrix must keep

When someone does run chats, keep four unbranded discovery prompts and four branded verification prompts. Hash the file. Start each row in a new conversation. Save the visible answer. Save every Gemini-exposed destination. Leave ambiguous aliases unresolved until a second reviewer decides.

Gemini brand visibility fails when a homonym is forced into the mention total.

Citation rule

Count only a source affordance shown in the Gemini interface. Keep the destination. A later Google blue link is a different surface.

Repeat without manufacturing stability

Three nonconsecutive dates describe volatility in a small panel. They do not estimate population share. The sample size here is three public URLs. That cannot establish a weekly Gemini rate, and it does not prove a content edit caused a mention.

As of 2026 this page is not a customer Gemini export.

Failure modes that invent a chat

Docs 200 as a mention. A models catalog is not “Gemini named us.”

Privacy Hub as a citation. The Help article is Google’s policy page.

Prior-turn contamination. The brand is already in the thread.

Blended rate. Branded and unbranded rows share a percentage.

ISO badge as a Gemini score. 42001 is a management standard.

Unauthorized CLI as invisibility. Login failed. Gemini was not asked.

A Gemini brand visibility memo that skips the new-conversation rule is already late.

How a reviewer should read the log

Start with the three 200s. Confirm the models title still says “Models | Gemini API.” Confirm Privacy Hub still says Privacy Hub. Confirm this tool URL still answers 200. Only then open a new Gemini conversation, if the ticket actually needs one. Gemini brand visibility that opens the chat first will contaminate the record with whatever the reviewer already believed.

Write the displayed model string on its own line. If the UI shows a marketing name and the models catalog shows an API id, keep both. Do not pick the prettier one. A later analyst will need to know which string was on screen.

Save the full visible answer before you click a source chip. Clicking can change what the interface still shows. Gemini brand visibility that follows links first is measuring the destination page, not Gemini.

If the answer names a competitor and not the brand, keep the row. Zero is a result. Deleting quiet rows inflates the next average. If the answer names a homonym, leave mention = unresolved until a second person reads the span.

Locale belongs next to account state. A US personal account and a Workspace account in another country are not one cell. W3C language tags already told us to write the tag. Do that.

When the reviewer is done, the folder should contain: the dated prompt file, the three public fetch logs, the conversation captures, and a one-page decision. The decision may be “no action,” “fix a page,” or “fix the alias ledger.” Gemini brand visibility is not required to produce a campaign.

Two people repeating the three curls should see the same class of 200 unless Google moved a host. Two people opening chats will not see identical prose. That difference is why the public 200s are the desk and the chats are a later, labeled addendum.

Do not upload a Gemini transcript to a public ticket to prove you worked. Redact. Store a hash of the allowed excerpt. Gemini brand visibility that publishes raw chats to look empirical is leaking context, not proving a rate.

If the only artifact you have is this article’s chart, say so. The chart is HTTP 200 versus title present. It is a useful stop rule. It is not a Gemini mention series. Teams that need a series must cut the addendum and keep it off this URL until the captures exist.

What stays on disk after this desk

Keep the two curl logs, the tool-URL status line, and a one-sentence note that no chat was opened. That folder is the 2026-09-14 Gemini brand visibility packet. If a later teammate adds chats, they go in a dated subfolder with the prompt hash from the locked-set sibling, not into this chart.

Print the package pin on the same page as the empty transcript note: infinitegrowth@0.1.1 and -V = 0.1.0. Those strings do not prove Gemini behavior. They prove which CLI we could have used and did not authorize.

A reviewer who needs a punch list can still run seo-health check on this tool URL. Expect on-page lights, not mention bits. Gemini brand visibility does not promote those lights into a Gemini score.

If legal asks whether we “tested Gemini,” the accurate answer is: we fetched Google’s public models and privacy pages and we did not open a consumer session. Anything stronger is a different desk with captures.

Store the user-agent InfiniSynapse-SH1-desk/2026-09-14. A later fetch with a browser UA can receive a different shell. That is expected. Write the new status beside the old one. Do not silently replace the 200s.

The decision this page supports is narrow: refuse a blended Gemini rate until a new conversation exists. Gemini brand visibility does not decide titles, products, or budgets. Those decisions need the addendum folder.

When the addendum folder is still empty next quarter, republish the same three 200s with a new date if you re-fetched them. Do not backfill chats to make the quarter look complete. Empty is a legal state for Gemini brand visibility. A quarter with three dated 200s and a still-empty transcript folder is a complete desk, not an unfinished slide.

Nearby guides

The Perplexity source-card hub owns the 403 Help fetch. Do not copy that table here.

The AI visibility hub still owns the broader product story. Gemini brand visibility stays on the clean-session job.

Frequently Asked Questions

Did we open a Gemini consumer chat?

No. The desk fetched public pages. Gemini brand visibility still has no answer text to code.

Should branded and unbranded prompts share one rate?

No. Combining them makes prompted repetition look like discovery. Split the columns.

What counts as a Gemini citation?

Only a source Gemini itself showed. A URL found later in Search is not that evidence.

Does a models-page 200 prove API coverage in the consumer UI?

No. API names and UI labels can differ. Write the displayed string.

May we publish a mention percentage from this article’s chart?

No. The chart is HTTP 200 versus title present. It cannot establish a mention rate.

Conclusion

Use Gemini brand visibility when the job is to refuse a blended rate until a new conversation can be stored. This 2026-09-14 pass recorded three public 200s and zero chats. Stop when the models title, the Privacy Hub title, and the empty transcript folder remain visible. A bounded docs log beats an unsupported Gemini percentage.

Do not paste a Gemini transcript into the public comments of this page. Keep session notes in the matrix file, or do not claim Gemini brand visibility at all.