Note:
Intent Matching using Text is a Beta feature available as part of Whatfix’s AI Adoption offering. For more information, contact support@whatfix.com or your Customer Success Manager.
What is Intent Matching using Text?
There are two key aspects of enhancing user experience within an application: understanding user intent and delivering context-specific guidance at the right moment. Intent Matching using Text helps address both by using AI to analyze text entered by users in specific fields within an application and determine whether it matches a defined intent or sentiment.
The text being evaluated includes any text-based field in the application, such as feedback forms, support request forms, CRM notes, or comment fields. The intent can represent a sentiment (for example, negative feedback or positive feedback) or a business-specific objective (for example, escalation request, support query, upsell opportunity, or feature usage question).
The evaluation happens in real time during user interaction. Based on the result, Whatfix can display relevant widgets such as Smart Tips, Beacons, Launchers, Pop-ups, and more, helping users receive contextual guidance that matches their situation.
How does the AI evaluate intent?
When a Visibility Rule that uses Intent Matching is triggered, Whatfix sends the selected element's text and the configured intent definition to the underlying AI model for analysis.
The AI compares the content of the element against the intent description provided in the Visibility Rule. It determines whether the text matches the specified sentiment, intent, or meaning and returns a simple yes or no response. This response is then used to evaluate the Visibility Rule and determine whether the associated widget should appear.
The following examples illustrate the prompt structure used internally for analysis.
Negative Customer Feedback (Email)
[SYSTEM]
You are an AI assistant expert in analyzing Sentiment of a text in application for a User's context. Your task is to determine if the given text of an element in a user's context matches the provided sentiment (for example, positive, negative, or a custom definition). Analyze the text and sentiment provided and respond with yes or no based on the analysis.
Important: Do not include any additional text or explanations. Return only 'yes' or 'no'.
[USER]
Given the following text of an element and the input sentiment, determine if the text reflects the sentiment. Respond with 'yes' or 'no'.
Text: "Subject: Extremely disappointed with the recent update
Hi Team, I've been using your product for over a year, but the latest update has made it much harder to complete basic tasks. Features I rely on are now buried behind multiple clicks, and the application feels slower than before. I'm seriously considering switching to a competitor if these issues aren't addressed soon. Regards, John"
Sentiment: "Is the Customer really not happy using this product."
Does the text reflect the sentiment?
At-Risk Sales Opportunity (CRM Record)
[SYSTEM]
You are an AI assistant expert in analyzing Sentiment of a text in application for a User's context. Your task is to determine if the given text of an element in a user's context matches the provided sentiment (e.g., positive, negative, or a custom definition). Analyze the text and sentiment provided and respond with yes or no based on the analysis.
Important: Do not include any additional text or explanations. Return only 'yes' or 'no'.
[USER]
Given the following text of an element and the input sentiment, determine if the text reflects the sentiment. Respond with 'yes' or 'no'.
Text: "Opportunity Name: Acme Corp - Enterprise CRM Renewal
Stage: Negotiation/Review
Latest Customer Update: The customer informed us that they are leaning toward a competitor due to pricing concerns and a better implementation timeline being offered elsewhere. They have postponed the final decision twice and have requested a revised commercial proposal as a last attempt before making their selection.
Risk Indicators: Decision date pushed out twice | Competitor strongly favored | Procurement team challenging pricing | Executive sponsor has become unresponsive.
Sales Rep Notes: I believe there is a high probability that we will lose this opportunity unless we can significantly improve pricing or secure executive alignment within the next week."
Sentiment: "Is the Customer really not happy using this product."
Does the text reflect the sentiment?
What are the capabilities of Intent Matching using Text?
Evaluates the sentiment, intent, or meaning of free-form text entered by users in real time.
Supports both short keywords and full natural language phrases as intent values (for example, "upsell opportunity intent" or "Show when anybody asks a question about feature usage").
Works across a wide range of widget types, including Smart Tips, Beacons, Launchers, Pop-ups, and more.
Returns a simple yes or no decision - no additional explanation or scoring is exposed.
Applies to any text-based element on the application, from feedback forms to CRM fields.
What are the limitations of Intent Matching using Text?
The accuracy of this feature may vary because it relies on the GPT-4o Mini model for sentiment and intent analysis. As a lightweight model optimized for speed and cost efficiency, it may not always interpret nuanced language, complex context, sarcasm, or ambiguous intent as accurately as larger models. Consequently, some classifications may occasionally be incorrect.
Use case
Consider a customer support feedback form where users enter their experience and click Submit to record their feedback. Some might leave negative feedback or express dissatisfaction. Intent Matching using Text detects this and shows a Pop-up with relevant support options, such as troubleshooting articles, live chat, or escalation paths. Without Intent Matching using Text, users might see a generic message such as: “Thanks for your feedback.” However, this doesn’t address their concern or help resolve their issue.
With Intent Matching using Text, Whatfix detects negative sentiment and responds with relevant guidance, such as: “It looks like you’re facing an issue. Here are some resources that might help.”
Steps to show a widget based on user intent
Use the following steps to show a widget based on user intent:
Step 1: Create a Pop-up.
Create a Pop-up based on your requirements. For more information, see Create a Pop-up.
Step 2: Configure the new Visibility Rules.
In the Visibility rules section for your widget, click When does the pop up start and stop appearing?

In the When does the pop up appear? section, select Something on the application from the What causes the pop up to appear? dropdown.

In the What would you like to identify? dropdown, select Element.

Click Select element using studio.

Enter the URL of the page from where you want to select the element.
.png?sv=2026-02-06&spr=https&st=2026-06-22T03%3A50%3A27Z&se=2026-06-22T04%3A06%3A27Z&sr=c&sp=r&sig=V0EarDpR%2FBU1aWwIS42ddWpDNmeXfQGBpFlEjd2Cz%2BA%3D)
Click Select element.

On the redirected page, use Whatfix Studio to select the desired element.
.gif?sv=2026-02-06&spr=https&st=2026-06-22T03%3A50%3A27Z&se=2026-06-22T04%3A06%3A27Z&sr=c&sp=r&sig=V0EarDpR%2FBU1aWwIS42ddWpDNmeXfQGBpFlEjd2Cz%2BA%3D)
Note:
Click Add Rule icon to add or edit display rules for the selected element. For more information, see Selected Element conditions.

Click the Reselect element icon to reselect the element on the application.

In the Check if element section, choose Exists and compare to show the widget based on some criteria in the underlying application.
.png?sv=2026-02-06&spr=https&st=2026-06-22T03%3A50%3A27Z&se=2026-06-22T04%3A06%3A27Z&sr=c&sp=r&sig=V0EarDpR%2FBU1aWwIS42ddWpDNmeXfQGBpFlEjd2Cz%2BA%3D)
In the Compare to dropdown, select Element text.
.png?sv=2026-02-06&spr=https&st=2026-06-22T03%3A50%3A27Z&se=2026-06-22T04%3A06%3A27Z&sr=c&sp=r&sig=V0EarDpR%2FBU1aWwIS42ddWpDNmeXfQGBpFlEjd2Cz%2BA%3D)
In the Data type dropdown, select Text.
.png?sv=2026-02-06&spr=https&st=2026-06-22T03%3A50%3A27Z&se=2026-06-22T04%3A06%3A27Z&sr=c&sp=r&sig=V0EarDpR%2FBU1aWwIS42ddWpDNmeXfQGBpFlEjd2Cz%2BA%3D)
In the Check if value dropdown, select Match Intent.
.png?sv=2026-02-06&spr=https&st=2026-06-22T03%3A50%3A27Z&se=2026-06-22T04%3A06%3A27Z&sr=c&sp=r&sig=V0EarDpR%2FBU1aWwIS42ddWpDNmeXfQGBpFlEjd2Cz%2BA%3D)
In the Value field, specify the intent value that the selected element’s text should match, for example, Negative feedback or Positive feedback.
.png?sv=2026-02-06&spr=https&st=2026-06-22T03%3A50%3A27Z&se=2026-06-22T04%3A06%3A27Z&sr=c&sp=r&sig=V0EarDpR%2FBU1aWwIS42ddWpDNmeXfQGBpFlEjd2Cz%2BA%3D)
Info:
In the Value field, you can also enter natural language phrases or sentences to describe the intent your text should match. For example, "upsell opportunity intent" or "Show when anybody asks a question about feature usage". This means you can specify intent using natural, conversational language, making it easier to match a variety of user inputs.
Once you have added all the required Visibility Rules, click Save.

Finish creating the widget, and then click Save.
