AI Conversation Sentiment Analysis

Conversation Sentiment Analysis shows how AI platforms present Your Brands in monitored responses over time.

It helps you distinguish being mentioned from being presented favourably. A brand can appear frequently while receiving neutral descriptions, mixed recommendations or negative comparisons.

What the chart measures

For each monitoring date, Wheremore groups qualifying responses into five sentiment classes:

Positive — Your Brands is clearly recommended or presented favourably.
Neutral — Your Brands is mentioned without a clear judgement or recommendation.
Mixed — Your Brands is presented with caveats, trade-offs or suitability for only some use cases.
Negative — Your Brands is clearly not recommended or presented unfavourably.
Unknown — the response does not contain a supported sentiment value.

The chart is a 100% stacked area chart. Each measured date shows the percentage distribution of these sentiment classes among the qualifying responses for that date.

How the percentages are calculated

Each monitored response contributes no more than one sentiment classification for Your Brands.

For each date:

responses in one sentiment class ÷ all qualifying Your Brands responses × 100

For example, if 100 qualifying responses mention Your Brands and 70 are positive, 15 neutral, 10 mixed, 3 negative and 2 unknown, the chart displays 70%, 15%, 10%, 3% and 2%.

Hover over a measured date to see:

the percentage for each sentiment;
the underlying response count; and
the total qualifying responses used for that date.

For example:

Positive: 42.9% · 3 of 7 responses

How Your Brands are identified

The chart uses the company and product names saved under Your Brands in AI Visibility Settings.

All saved aliases represent the same monitored organisation. If several aliases appear as structured providers in one response, Wheremore counts the response once and uses the sentiment associated with the best provider rank.

Provider rank 1 is better than rank 2.

If different aliases have conflicting sentiment at the same best rank, Wheremore resolves the conflict conservatively in this order:

Negative
Mixed
Neutral
Positive
Unknown

This prevents one monitored response from inflating the result through several aliases.

Use specific aliases that match the names AI platforms return. Avoid broad terms that could refer to unrelated providers.

Before data appears

The selected project needs:

An AI Visibility monitoring configuration.
Monitoring questions assigned to journey phases.
One or more selected AI platforms.
Completed and processed monitoring responses.
At least one company or product name under Your Brands in AI Visibility Settings.
Structured provider records containing one of those saved names.

Sentiment is read from the structured provider data saved with each processed response. The chart does not infer sentiment again from the answer text when it loads.

Quick start

Open the AI Visibility Dashboard.
Select a project.
Choose a date range.
Leave Journey Phase, Configuration and Platform set to all values for the broadest view.
Set Provider Position to Mentioned Anywhere.
Review the balance between Positive, Neutral, Mixed, Negative and Unknown.
Hover over a measured date to see percentages and response counts.
Narrow the filters to investigate a journey phase, audience configuration or AI platform.
Compare Mentioned Anywhere with Top 3 and #1 Recommendation.
Open AI Platform Conversation Responses to read the answers behind an important change.

Apply the dashboard filters

Project

Project selects the monitoring configurations, processed responses and Your Brands settings used by the chart.

Projects are limited to those available in the active workspace.

Date Range

Choose a preset period or a custom start and end date.

Dates are interpreted as UTC calendar dates and both custom boundaries are included. Last 30 Days means today and the preceding 29 UTC dates.

Journey Phase

Choose all phases or restrict the chart to:

Awareness;
Consideration; or
Decision.

This shows whether AI platforms describe the brand differently as buyers discover a problem, compare solutions or approach a final choice.

Configuration

Choose all configurations or one monitoring configuration.

Use this to isolate a particular audience, market context, product or question set. Disabling a configuration stops future scheduling but does not remove its completed historical responses from this chart.

Platform

Choose all platforms or one platform.

Use this to compare sentiment across ChatGPT, Claude, Gemini, Perplexity and their supported search-enabled variants.

Provider Position

Provider Position restricts which Your Brands provider records qualify:

Mentioned Anywhere — accepts Your Brands at any valid provider rank.
Top 3 Recommendations — accepts ranks 1, 2 and 3.
#1 Recommendation — accepts only rank 1.

This filter changes the population being measured. It does not merely zoom the existing chart.

For example, Top 3 answers the question: “Among monitored responses where Your Brands appears in the first three structured provider positions, how is it presented?”

Start with Mentioned Anywhere, then compare Top 3 and #1 to see whether prominent recommendations carry a different sentiment mix.

Read the chart

Each sentiment row uses this format:

Positive: 42.9% · 3 of 7 responses

The hard numbers explain the percentage:

3 is the number of qualifying responses classified as Positive on that date.
7 is the total number of qualifying Your Brands responses across all sentiment classes on that date.
42.9% is 3 divided by 7, expressed as a percentage.

The tooltip also provides a short explanation of each sentiment class. All sentiment rows for the same measured date use the same response total.

A growing Positive area — a larger share of qualifying responses presents Your Brands favourably.
A growing Neutral area — visibility is increasing without a clear recommendation.
A growing Mixed area — AI platforms are attaching more caveats or trade-offs.
A growing Negative area — unfavourable comparisons or recommendations are becoming more common.
A growing Unknown area — more processed provider rows lack a supported sentiment classification.
Positive at Mentioned Anywhere but Mixed at #1 — general mentions are favourable, but leading recommendations carry more caveats.
Different results by journey phase — positioning may work during discovery but weaken during evaluation or decision.
Different results by platform — model sources, retrieval behaviour or response framing may be affecting presentation.

Always consider the response count in the tooltip. A large percentage movement based on a few responses is weaker evidence than a smaller movement supported by many.

Monitoring days and missing dates

Monitoring may run only on selected weekdays.

A day without qualifying responses is not a 0% sentiment result. Wheremore omits unmeasured dates from the plotted series and joins the nearest measured dates to keep the trend readable.

The connecting area is visual continuity only. It does not mean responses were measured on the intervening dates.

There is an important difference:

0% for a sentiment — qualifying responses existed, but none belonged to that class.
No observation — no qualifying Your Brands responses existed under the selected filters.

How settings changes affect history

The chart applies the current Your Brands aliases to stored provider records.

Adding, removing or correcting an alias can therefore change which historical responses qualify. It can also change which alias supplies the best-ranked sentiment in a response.

Use an annotation to record material alias, configuration, question, platform or processing-schedule changes. Keep settings stable when evaluating a continuous trend.

Add annotations

Select a date inside the chart to add a short annotation.

Useful annotations include:

changes to Your Brands aliases;
edited questions or audience context;
a launch or positioning change;
an important website or content update;
campaign activity;
a change to selected AI platforms; or
a processing-day change.

Select an annotation label to delete it.

Annotations provide context only. They do not alter responses or sentiment calculations.

What the chart does not measure

Conversation Sentiment Analysis is not:

Share of Voice;
Competitive Win Rate;
an overall mention rate;
citation rate;
conventional web sentiment analysis across social media or reviews;
a new interpretation of the full answer text at dashboard load;
proof that a content, campaign or product change caused a movement;
search traffic, rankings, revenue or conversions; or
a statistically weighted market-opinion survey.

It describes the structured sentiment assigned to Your Brands in the monitored AI responses that qualify under the current filters.

What should I do next?

Open AI Platform Conversation Responses to read the underlying wording and recommendations.
Use Customer Journey Phase Progression to compare visibility across buyer stages.
Use Share of Voice to compare relative appearance volume.
Use Competitive Win Rate to see who ranks better during head-to-head responses.
Use Citation Rate Analysis to see whether answers provide sources and cite your domain.
Review Key Item Tracker for recurring product, feature and positioning terms.
Review AI Visibility Settings if aliases are missing or incorrectly grouped.
Review Manage AI Visibility Tracking if questions, platforms, phases or processing days need changing.

Data timing

The chart is not real time.

Monitoring runs on each configuration’s selected processing days. Results appear after provider calls finish and raw responses have been processed into structured runs, providers and sentiment values.

The newest scheduled date can be incomplete while jobs remain pending, running, saved as raw responses or waiting to be processed.

Google AI Mode uses the same Your Brands alias matching and stored structured sentiment as other platforms. Answer-level sources are not sentiment labels. A response without usable provider sentiment can remain visible in the response viewer without contributing to this chart.

Troubleshooting

Select a project appears

Select a project in the global dashboard controls.
Confirm the active workspace is correct.
Confirm you have access to the project.
Allow the project selector to finish loading.

Add at least one of your brand names appears

Select Configure on the AI Visibility Dashboard.
Add the required company and product names under Your Brands.
Save the settings.
Reload the chart if it does not refresh automatically.

No qualifying sentiment observations were found

Increase the date range.
Return Journey Phase, Configuration and Platform to their all-values options.
Set Provider Position to Mentioned Anywhere.
Confirm the selected configuration has completed monitoring runs.
Confirm saved aliases match structured provider names.
Open AI Platform Conversation Responses to confirm the brand appears.

Unknown is unexpectedly high

Inspect the affected responses and provider records.
Confirm the response processor saved a supported sentiment value.
Check that the provider output followed the required sentiment contract.
Keep Unknown visible rather than treating missing classifications as Neutral.

A percentage changed sharply

Hover over both dates and compare their total qualifying responses.
One additional classification can create a large movement when the daily total is small.
Confirm the project, filters, Provider Position and Your Brands aliases are unchanged.

The chart connects across days without monitoring

This is expected. The connection keeps measured weekday observations readable and does not create intermediate measurements.

The chart changed after editing Your Brands

This is expected because current aliases are matched against historical structured provider records. Add an annotation at the settings-change date.

The reports measure different things:

Conversation Sentiment measures how Your Brands is presented.
Share of Voice measures relative response-level appearance.
Competitive Win Rate measures structured rank outcomes during head-to-head encounters.

Match the project, date range, journey phase, configuration, platform and Provider Position before comparing them.

Conversation Sentiment could not be loaded

Select Retry.
Refresh the page if the session expired.
Confirm the project belongs to the active workspace.
Check whether other AI Visibility charts load.
Ask an administrator to inspect the application log if the error persists.