Customer evidence · 4 named accounts

What AI engines said, and what that does not prove

Last updated: October 17, 2026

Every account below is named. Each entry states separately what was observed, and what is not being claimed. Some entries are dated observations of AI answers; others are statements made by the account holder. They are labelled so you can tell which is which.

Run a free visibility scan See published city markets

Named accounts

Four engagements, stated plainly

Dated observations recorded

Kotta Financial Services

Financial services · Durham, North Carolina

Kotta was named in Google AI Mode results for three Durham insurance services on September 6, 2026, and was the first-listed provider in four ChatGPT responses captured on September 8, 2026.

Observed
Recommendation visibility was observed on the dates above for Fixed Indexed Annuities, Indexed Universal Life, and Term Life Insurance queries naming Durham.
Not claimed
These are point-in-time observations, not endorsements by Google or OpenAI, and they do not by themselves establish that Ninar AI caused the placements.
Read the dated Kotta case study

Customer-reported outcome

Dronatize

Drone data and AI services · Raleigh, North Carolina

Dronatize used a local discovery scan to find which Raleigh drone inspection questions returned answers without the company, then published against those gaps.

Observed
The account holder reports appearing in AI recommendations for its target market.
Not claimed
This is the customer’s own statement. No engine transcript for this account is published here, and no ranking position is claimed.
See how the Raleigh market is measured

Customer-reported outcome

Johaus Global

Technology and AI solutions · United States, national prompts

A baseline scan showed the company absent from ChatGPT and Perplexity answers for the custom data and AI consultancy questions its buyers use.

Observed
The account holder reports appearing in AI-generated recommendations five to seven weeks after acting on the scan results.
Not claimed
This is the customer’s own statement. The time frame is theirs and is not a benchmark, a forecast, or a commitment for any other account.
Compare AI visibility platforms

Engagement in progress

CarbonMeter.ai

Carbon accounting · London and UK-wide

Matched London-qualified and UK-wide prompt groups run as separate records so the two markets are never averaged together. The engagement began in September 2026.

Observed
Measurement is running. Each answer is retained with its engine, date, and cited sources.
Not claimed
No result, ranking change, or improvement is claimed for this engagement at this time.
See how the London market is measured

Use cases

Six questions Ninar AI is bought to answer

Each use case names the buyer, the question, a representative prompt, and the evidence retained after the scan. Prompts are examples of shape, not results.

B2B supplier and vendor discovery

B2B teams whose buyers ask an AI engine for a shortlist before they search

Which suppliers does an engine name when a buyer describes our category and never types our brand?

Example promptWhich companies provide industrial water treatment services for manufacturers in Charlotte?

Evidence retained. The full answer text, the engine that produced it, the date, every competitor named ahead of you, and each source the engine cited.

How measurement works

ChatGPT recommendation tracking

Brands that want to know whether ChatGPT recommends them, a competitor, or nobody

Is our brand named in the answer, where in the answer, and which sources were cited?

Example promptWho should I hire for fractional CFO services for a Series A startup?

Evidence retained. Each ChatGPT response is stored on its own, scored on mention, prominence, cited sources, and sentiment, and kept so the same prompt can be rerun later.

Prompt intelligence

Google AI Overviews and AI Mode tracking

Teams losing classic search clicks to AI-generated answers

When Google answers the question directly, are we inside the answer or only below it?

Example promptBest accounting software for a small construction business

Evidence retained. AI Overviews and AI Mode are tracked as separate engines, each with its own retained answer text and cited source list.

How measurement works

D2C product discovery evidence

Direct-to-consumer and ecommerce brands competing against marketplaces in AI answers

When a shopper describes the product instead of naming a brand, which products get named?

Example promptWhat is a good refillable cleaning product subscription that ships to the UK?

Evidence retained. Product and category answers are retained with the retailer or marketplace the engine cited, so you can see whether it linked to you or to a reseller.

D2C AI visibility

City and national measurement side by side

Businesses serving one city that also compete in a national category

Does the engine answer differently when the question names our city?

Example promptWho are the best immigration lawyers in Boston?

Evidence retained. City-qualified and national prompts are stored as separate records. A result in one market is never averaged into or credited to another.

Published city markets

Agency reporting across client accounts

Agencies that need to show a client what changed and why

What did we publish, and did the answers move afterwards?

Example promptWhich marketing agency in Austin handles AI search optimization?

Evidence retained. Every scan is retained, so the before state, the published change, and the after state are separate records a client can inspect.

Agency workflow

Six scans, same benchmark

October 9–17 observations using our own account

The examples above describe what is measured. These scans use Ninar AI’s own account because it is the account whose results we can publish without a customer’s permission. Each scan used the same five benchmark groups: category discovery, specific use case, pricing and purchase, workflow integration, and geographic fit.

Scan 467 · Durham · ChatGPT

Completed October 17, 202620 of 20 benchmark prompts completed.

Ninar AI appeared in 1 of 20 answers. The resulting composite visibility score was 3/100.

Scan 473 · Durham · Perplexity

Completed October 17, 202620 of 20 Perplexity API prompts completed.

Ninar AI appeared in 1 of 20 answers. The resulting composite visibility score was 4/100. The only mention came from the prompt that named Ninar AI directly; that branded evaluation scored 87/100, while all 19 unbranded prompts omitted Ninar AI.

Scan 468 · Raleigh · ChatGPT

Completed October 17, 202620 of 20 benchmark prompts completed.

Ninar AI appeared in 1 of 20 answers. The resulting composite visibility score was 4/100.

Scan 469 · Raleigh · Google AI

Completed October 17, 202620 of 20 Gemini Search-grounded prompts completed.

Ninar AI appeared in 1 of 20 answers. The resulting composite visibility score was 3/100.

Scan 471 · London · ChatGPT

Completed October 17, 202620 of 20 benchmark prompts completed.

Ninar AI appeared in 1 of 20 answers. The resulting composite visibility score was 2/100.

Scan 441 · New York · Google AI

Completed October 9, 202620 of 20 Gemini Search-grounded prompts completed.

Ninar AI appeared in 1 of 20 answers. The resulting composite visibility score was 3/100.

What was recorded

For every prompt: the engine, run date, verbatim answer, brand mention, position when named, competitors, and cited sources. Mention, recommendation, ranked inclusion, and citation remain separate fields.

What the rerun established

The relevant product, pricing, documentation, customer, and market pages were live and indexable, but 19 of 20 answers in every scan still omitted Ninar AI. The next work is stronger independent corroboration and more prominent retrieval paths, not duplicate pages.

Observed
Six completed 20-prompt scans from October 9–17, 2026: Durham ChatGPT scored 3/100, Durham Perplexity scored 4/100, Raleigh ChatGPT scored 4/100, Raleigh Google AI scored 3/100, London ChatGPT scored 2/100, and New York Google AI scored 3/100. Ninar AI appeared once in each scan.
Not claimed
No causal effect or improvement is claimed. These are point-in-time observations, and differences between dates, engines, or cities do not establish why an answer included or omitted the brand.

This example is deliberately unflattering. It is published because a page about measurement evidence that only showed favourable numbers would not be evidence. Future comparable reruns will be added whether or not the scores improve.

Published case studies

Full write-ups with dated evidence

Questions about this evidence

What this page does and does not establish

What kinds of businesses use Ninar AI?

Named accounts span financial services, drone and geospatial data, technology consulting, and carbon accounting, across Durham, Raleigh, London, and national prompt sets. The platform is category-agnostic: any business whose buyers ask an AI engine for a recommendation can be measured.

Are these results guarantees of similar outcomes?

No. Some entries on this page are dated observations of AI answers and others are statements made by the account holder. AI answers vary by prompt wording, engine, model version, user context, and time. Nothing here is a forecast or a commitment for another account.

What evidence is retained for each customer?

For every prompt sent to every engine, Ninar AI retains the answer text, the engine name, the timestamp, whether the brand was mentioned, where it appeared in the answer, which competitors were named, the sources the engine cited, and the sentiment. A probe is one prompt sent to one engine.

Do you have third-party review scores?

Not at this time. The feedback on this page comes directly from named accounts and is published without a star rating or an aggregate score, because no verified third-party review corpus exists for Ninar AI yet.

Which engines are these customers measured across?

Ninar AI integrates with 11 AI engines. Each plan selects between 2 and 6 of them per scan, and every engine is scored separately rather than averaged into a single number.

Disclosure

How to read everything above

Ninar AI is a remote software platform headquartered in Durham, North Carolina. AI answers are generated per request and vary by prompt wording, engine, model version, user context, and time. A dated observation records what one engine returned on one date; it is not an endorsement by that engine's operator and does not on its own establish that Ninar AI caused the placement. Statements attributed to an account holder are that account holder's own words. No aggregate rating, star score, or third-party review average is published for Ninar AI, because none has been independently verified.

Read the measurement workflow · Review plans and limits · See every published market · Compare AI visibility tools