Clare Videre Documentation

Overview

Clare Videre helps companies acquire customers through better visibility in ChatGPT. It shows whether a company appears when prospective customers ask ChatGPT which products to use, which vendors fit their needs, and how competing products compare.

These answers can shape a buying decision before the customer visits a website or speaks to sales. Clare gives companies a way to observe that part of the customer journey and improve how they are discovered and represented there.

How Clare works

Clare monitors the natural-language questions prospective customers ask while researching a purchase. It repeatedly asks those questions to ChatGPT, measures which brands appear, and records the public sources ChatGPT consults.

This creates a continuing improvement cycle:

1. Define the buyer questions where the company should be visible.
2. Measure how ChatGPT currently answers them.
3. Find prompts where competitors appear but the company does not.
4. Identify the websites, articles, documentation, comparisons, and conversations that inform those answers.
5. Improve the company's product information, website, content, and third-party presence.
6. Continue monitoring to see whether visibility changes.

Clare does not change ChatGPT's answers itself. It provides the measurements, evidence, and priorities a company needs to decide what should change in its marketing, content, documentation, product positioning, or distribution.

What success means

Success means appearing consistently in relevant buyer answers, being represented accurately, and giving prospective customers a reason to consider the company. The goal is not to appear in every answer. It is to be visible for the questions that match the company's real product, audience, and strengths.

Visibility becomes more useful when it is connected to customer intent. A mention in a high-value buying question can matter more than many mentions in questions that do not describe the company's target market.

What Clare can show

Clare can show how often brands appear across repeated answers, how that visibility changes, which competitors occupy a buyer question, and which sources ChatGPT consulted. It can turn these signals into focused tasks for investigation.

Clare cannot guarantee a particular recommendation or prove that one page caused ChatGPT to mention a brand. AI answers vary, and source consultation does not establish causation. Clare reduces that uncertainty through repeated measurement and clear evidence rather than presenting any single answer as definitive.

Overview

A prompt is a natural-language question that Clare Videre asks ChatGPT on behalf of an organization. Prompts represent the questions prospective customers ask while discovering, comparing, and evaluating products - not keywords used to search the organization's own website.

Each prompt is monitored independently. Clare records repeated answers over time so it can measure whether the organization's brand appears, which competitors appear alongside it, and which sources influence the answers.

Choosing prompts

Useful prompts reflect a real buying decision and leave room for ChatGPT to recommend products or vendors. Common prompt families include:

• Category recommendations, such as "best customer support tool for B2B SaaS."
• Direct comparisons, such as "Company A vs Company B."
• Alternative searches, such as "alternatives to Company A."
• Branded evaluations, such as "is Company A good for enterprise teams?"

Audience, use case, industry, or other qualifications belong in a prompt when they can materially change the recommendation. Paraphrases with the same buyer intent usually do not add useful coverage; they divide the sample between equivalent questions.

For unbiased visibility measurement, the organization's brand should not appear in a discovery prompt. A prompt that names the brand is naturally likely to produce an answer that mentions it. Branded evaluations remain useful, but should be treated as a separate kind of question.

Tags and pausing

Tags organize prompts into meaningful groups such as buyer intent, audience, product line, or market. Visibility and source insights can be filtered by tag, allowing unlike prompt families to be analyzed separately. Tags organize the resulting data; they do not alter the question sent to ChatGPT.

Pausing a prompt stops future automatic runs while preserving its configuration and historical answers. Paused prompts are excluded from insights by default, but can be included when reviewing older or temporarily inactive monitoring.

Prompt data

Every completed run is attached to its prompt. Clare uses that history to calculate brand mention rates, identify changes over time, compare the organization with other brands in the answers, and retain the source URLs ChatGPT referenced.

Prompt-level data contributes to the aggregate Visibility, Sources, Reddit, and Tasks insights. Keeping distinct buyer intents in distinct prompts makes those aggregate reports easier to interpret.

Overview

Automatic Runs ask ChatGPT the prompts an organization monitors on a recurring schedule. Each run produces a new answer that Clare Videre records and analyzes, building a history of how ChatGPT presents the organization, its competitors, and the market around them.

All unpaused prompts are included automatically. Pausing a prompt excludes it from future automatic runs without deleting its previous answers or historical data. Prompts can also be run manually at any time.

Repeated sampling

ChatGPT can answer the same question differently from one run to the next. Clare can therefore repeat each prompt several times instead of treating one answer as definitive. A larger sample makes recurring brand mentions, omissions, competitors, and cited sources easier to distinguish from one-off variation.

Clare recommends at least ten runs per prompt per week when a more reliable overview is important. Lower sample sizes still produce usable answers, but percentages and short-term changes will be more sensitive to individual responses.

Data and usage

For every completed run, Clare retains the answer, the brands identified in it, and the sources ChatGPT referenced. This data powers Clare's visibility trends, prompt-level comparisons, total mention counts, source and Reddit conversation analysis, and recommended tasks.

Automatic runs consume usage credits per completed prompt run. Clare shows a monthly estimate based on the current number of unpaused prompts, sampling frequency, and per-run price.

Overview

Visibility measures how often an organization's brand appears in ChatGPT's answers to the selected prompts. It answers a direct question: when prospective customers ask these questions, how often is the organization part of the answer?

Clare calculates visibility across repeated prompt runs rather than relying on one answer. This makes it possible to track changes over time and to separate persistent visibility from occasional mentions.

How visibility is calculated

Visibility is the percentage of prompt outputs that mention the organization's brand. If the brand appears in 60 of 100 selected outputs, visibility is 60 percent. An output counts once whether the brand appears once or several times in that answer.

The calculation follows the selected prompt group. Tags can isolate a product, audience, market, or prompt family, and paused prompts are excluded unless the user chooses to include them.

Visibility reports

The Weekly Mention Rate shows the organization's mention rate over time. Prompt Visibility breaks the same measurement down by question, making it clear which buyer intents regularly include the brand and which do not. Clare also identifies prompts that always or never mentioned the brand during the recent period.

Short-term comparisons detect immediate movement, while long-term comparisons use a broader baseline to reveal more persistent changes. Clare warns when either side of a comparison has too few runs to support a reliable conclusion.

Total Mentions ranks every identified brand by the number and share of selected outputs in which it appeared. This provides competitor context alongside the organization's own visibility.

Interpreting visibility

Visibility is a mention metric. It does not claim that ChatGPT selected a winner, ranked the brands by preference, or expressed positive sentiment. An organization can be visible in an answer that recommends it, compares it neutrally, or discusses a limitation.

Discovery prompts that do not name the organization provide the cleanest visibility signal. Branded prompts are naturally likely to mention the named brand, so Clare warns when they are included and recommends separating them with tags.

Overview

Sources shows which public web pages ChatGPT consulted while answering monitored prompts. It helps an organization understand what external material appears around its market, which domains repeatedly inform answers, and whether its own website is part of that source set.

Clare retains the source URLs attached to each completed answer and connects every source back to the prompts and runs in which it appeared.

Web search and trained data

Some ChatGPT answers perform web search and return consulted pages. Other answers rely on the model's trained knowledge without consulting the web. Clare separates these two cases and compares how often the organization's brand appears in each.

For web-search answers, Clare measures how often ChatGPT consulted Reddit, the organization's own domain, and other domains. It also compares brand mentions when the organization's website was consulted with mentions when web search used other sources.

Domains and pages

Domain counts show how many web-search answers consulted each domain. A domain is counted once per answer even when ChatGPT consulted several pages from it. This makes the count represent answer coverage rather than the raw number of links.

Clare separates homepages, other pages, and Reddit conversations. Other pages can include articles, documentation, pricing pages, feature pages, and other URLs with a path. Reddit conversations are restricted to subreddit discussion URLs rather than Reddit search or utility pages.

Each page can be traced back to the prompts that consulted it, including how many runs used the page and what share of that prompt's runs it represents. Pages on the organization's own domain are identified separately.

Interpreting sources

Frequent consultation shows that a source is present in ChatGPT's research path for the selected prompts. It does not by itself prove that the source caused a brand mention or that ChatGPT agreed with the page.

The useful signal is comparative. An organization can see whether its own site participates in the answers, which third-party domains appear more often, which specific pages recur, and whether brand visibility differs between web-search and training-only answers.

Overview

Reddit shows how public Reddit conversations participate in ChatGPT's web research for the selected prompts. It connects the answers Clare monitors with the specific threads those answers cite, revealing how much of ChatGPT's research uses Reddit and whether the organization is present in those discussions.

Clare recognizes subreddit conversation URLs and groups links to the original post and its comments under one thread. Reddit search, profile, and utility pages are not treated as conversations. Clare retains every citation while counting the conversation once, separating the number of different threads quoted from the number of times those threads were quoted.

Reddit insights cover the last 30 days and follow the selected prompt tag and paused-prompt setting. This keeps the analysis aligned with the chosen set of buyer questions.

Reddit in web research

Reddit share of web research is the percentage of web-search answers that cited at least one Reddit conversation. A Reddit-sourced answer may also cite websites, documentation, articles, or other sources; the measurement means Reddit participated in the answer's research, not that Reddit was its only source.

Clare measures how many different threads ChatGPT quoted, how citations are distributed across subreddits, and the age of the quoted conversations. Subreddit share is weighted by citations, so a repeatedly quoted thread contributes more than a thread that appeared once. Thread ages are grouped by month for the first two years and then into a two-years-or-older group.

Together, these measurements show whether ChatGPT repeatedly returns to a narrow set of established discussions or draws from a broader range of communities and newer conversations.

Brand and competitor presence

Clare identifies the brands appearing in answers that cited Reddit and calculates the share of those Reddit-sourced answers that mention each brand. The organization's own brand is measured alongside leading competitors. This is an answer-level comparison: it describes brands in ChatGPT's answers, separately from brand mentions inside the Reddit conversations themselves.

Within the quoted conversations, Clare counts posts and comments that mention the organization. A post or comment counts once even if it repeats the name. Among conversations with complete details, brand coverage is the share containing at least one such mention.

Clare also shows the complementary number and share of conversations that do not mention the organization. This distinguishes a broadly represented brand from one that appears in ChatGPT's answer while remaining absent from the discussions ChatGPT cited.

Conversation analysis

For each quoted conversation, Clare brings together its title, subreddit, age, last activity, citation count, comment count excluding the original post, posts made by configured team Reddit accounts, and number of posts or comments mentioning the organization. The evidence can be ordered by any of these measurements, making frequently cited conversations with little or no brand presence visible as distinct opportunities.

Every conversation remains connected to the prompts that quoted it. Clare shows the total citations, the number of prompts involved, and the citation count and share for each prompt. This separates sources that recur across several buyer questions from conversations associated with one specific intent.

The same evidence powers Post on Reddit tasks. When ChatGPT recently quoted a conversation and no available post or comment mentions the organization, Clare can surface the thread as a focused place to consider contributing relevant expertise.

Interpreting Reddit signals

A citation shows that a Reddit conversation appeared in ChatGPT's research path. It does not prove that the conversation caused a brand mention, that ChatGPT agreed with it, or that the discussion expressed positive sentiment. Brand presence measures whether a name appears, not whether a commenter represents the organization.

Very old or unavailable conversations can have incomplete details. Clare retains their citation and subreddit information where possible, uses the Reddit URL when a title is unavailable, and marks unknown ages, comment counts, or mention counts as unavailable rather than treating missing data as zero.

Overview

Tasks turns Clare's prompt history into a focused list of visibility gaps, source problems, and conversation opportunities worth investigating. The tasks are generated from measured run data, not from a generic checklist or an unsupported claim about what will influence ChatGPT.

Each task is tied to a specific monitored prompt or recently quoted source and includes the evidence that triggered it. Tasks update as new prompt runs change the underlying signal.

Establish Visibility

An Establish Visibility task appears when a prompt has recent answers but none of them mention the organization. Clare shows the competing brands that do appear, giving the organization a concrete view of who currently occupies that buyer question.

The absence may reveal a genuine visibility gap, or it may reveal that the prompt does not match the organization's intended market. Clare therefore also makes it possible to remove an irrelevant prompt rather than treating every zero-mention question as a problem to solve.

Strengthen Sources

A Strengthen Sources task appears for a prompt where the organization is mentioned, but pages from its own domain are absent or appear behind the most frequently consulted pages. Clare identifies the organization's strongest existing page and the third-party domains that appear ahead of it.

This is evidence that ChatGPT often researches the topic elsewhere. It can guide investigation into whether the organization's product pages, comparisons, documentation, or articles explain the relevant subject clearly enough to become useful sources.

Post on Reddit

A Post on Reddit task appears when ChatGPT quoted a Reddit conversation during the last 30 days but no available post or comment mentions the organization. These conversations offer a concrete place to contribute relevant expertise where ChatGPT is already finding source material.

Recover Visibility

A Recover Visibility task appears when the organization's recent mention rate for a prompt falls substantially below its longer baseline. Clare includes the current and previous mention counts so the decline can be evaluated in context.

These tasks require enough runs in both periods before Clare treats the change as meaningful. A decline is a signal to investigate changed answers, competitors, and sources; it is not proof of a specific cause.

Overview

Clare exposes a streamable HTTP MCP endpoint at https://clarevidere.com/mcp. AI tools that support MCP can use it to investigate visibility, prompt health, competitors, cited sources, Reddit opportunities, and optimization tasks using the organization's stored Clare data.

MCP access uses the Clare account's personal API key as a bearer token. Treat that key like a password and do not share configuration prompts that contain it. You can copy or regenerate the key in Setup → Account; regenerating it immediately invalidates the previous key.

Setup

After signing in, open Insights → MCP for ready-to-copy Codex, Claude Code, and VS Code configurations containing your API key.

Tools

The read-only tools return account and prompt configuration, prompt health and stored answers, competitor performance, cited Reddit conversations without a brand mention, combined optimization tasks, and the current number of pending prompt runs. Website-changing optimization tasks exclude snoozed prompts by default; clients can request the snoozed prompts and their expiration times. Reddit tasks are never snoozed.

The write tools can snooze all website-changing optimization tasks for one prompt for two weeks, unsnooze all snoozed prompts, add a monitored prompt, or queue new runs of an existing prompt. Organization viewers can use the read tools but cannot make these changes.

Approvals and costs

Fetching stored Clare data does not change the account and does not require approval. Adding a prompt changes the organization's monitoring configuration and may increase future automatic-run costs. Running a prompt immediately spends usage credits.

Clare uses MCP form elicitation to enforce approval for every write tool. Before making the change, Clare shows the affected task or prompt, organization, schedule or run count, and estimated cost where applicable. The action runs only after you explicitly confirm it. If a client does not support form elicitation, it can still use every read-only tool, but Clare refuses write requests.

The endpoint uses MCP protocol 2025-11-25. Some clients also show their own tool confirmation before Clare's approval dialog; that additional confirmation is controlled by the client.

The MCP payload formats are intended for automation, but they are not version-locked. Contact us before depending on an individual field never changing.