Answer Assistant FAQs

No. However, your Account Administrator or Configuration Administrator, must enable and redeploy the feature in your configuration settings to make it active.

Note: Answer Assistant is an opt-in self-service feature. It's available for Hosted Display v2 clients only. 

No. Answer Assistant is not available on Hosted Display v1. If your website is still running on our legacy v1 platform, the AI-generated responses and transparency labels cannot render on your pages. To use this feature, your website must use Bazaarvoice Hosted Display v2.

No. Answer Assistant is currently only available to clients using Bazaarvoice Hosted Display v2.

To ensure you maintain total control over your brand content, Bazaarvoice operates a strict opt-in strategy for AI features. We don't use automatic rollouts to ensure absolute brand safety. You can enable it yourself.

No, Answer Assistant is currently optimized exclusively for Questions & Answers. Extending this functionality to our broader Network Response (Connections) solution is part of our future product roadmap.

Answer Assistant supports English, Spanish, French, and German.

The AI model generates suggested responses by analyzing and synthesizing the most relevant data sources available for each product. To enhance accuracy, the system prioritizes "source-of-truth" content, such as: 

• Product descriptions: Core specifications and features from your catalog.
• Consumer reviews: Real-world feedback from common users.
• PDP / URL context: Live data and layout information from the product detail page.
• Answered questions: Historical data from your Questions & Answers database.

Refer to the Bazaarvoice use of AI article. 

The AI model is only as good as the information available to it. High-quality answers require high-quality data.
Rich product descriptions, with a minimum of 50 words, are foundational to our AI model. 

For the best results, optimize your data based on these benchmarks:

• The 50-word jump: Clients using product descriptions of at least 50 words see the AI model generate drafts for three times as many customer questions.
• The 100+ word ‘Gold Standard’: Product descriptions over 100 words helps us achieve maximum answer coverage.
• Active PDP URLs: Ensure your Product Detail Page (PDP) URLs are active and correct. If website links are broken or outdated, the AI model cannot perform real-time page scanning and will lack the context it needs to generate a response.
• Anti-bot security compliance: If your internal security tools block Answer Assistant from scanning your live PDP, Bazaarvoice will respect that protection. The AI will automatically pivot to using your product catalog and past approved answers to generate the draft.

For more details, refer to our best practice guide for increasing your AI answer coverage.

Not at this time. Answer Assistant is designed to be low-maintenance and relies entirely on data you already provide; your Product Catalog, live Product Detail Page (PDP) URLs, and your historical user-generated content (UGC). 

Our recent research shows that leveraging this existing data provides an average of 55% answer coverage, reaching up to 75% for highly optimized catalogs.    

Refer to our best practice guide Increasing your AI answer coverage.

Not immediately. While the AI does not actively learn from your edits right away, they still help. Every approved Q&A pair is saved to your history, which the AI then uses to give better answers in the future. This leads to more accurate and personalized future drafts. Bazaarvoice also runs continuous backend evaluations to optimize draft quality and tone. 

No. On average, 50–55% of consumer questions receive an AI draft response, effectively cutting your drafting workload in half. Some high-performing catalogs reach 70–75%, while those with poor data health may reach only 20–25%.

Refer to the best practice guide for increasing your AI answer coverage. 

Tips

• Providing detailed product descriptions results in significantly higher answer coverage.
• Make sure your product page links (URLs) are active and working properly. This helps our AI model to seamlessly scan your live pages for extra facts. This can significantly increase the number of answers it can automatically draft for you. 

The current AI framework generates answers using all available data sources. This includes your Bazaarvoice Product Catalog, live Product Detail Page (PDP) URLs, and your historical user-generated content (UGC) . We’re continuously exploring ways to incorporate additional data sources to improve the quality and quantity of answers, especially for products with limited customer feedback.

Refer to our best practice guide Increasing your AI answer coverage.

Answer Assistant generates a unique response draft for every individual question submitted, even if the inquiries are identical or very similar. This ensures your agents can review and tailor the response specifically for each individual consumer's thread.

Yes. The AI model securely scans your live PDPs to enhance accuracy and ensure response drafts are contextually relevant.
 Here's how the page scanning works: 

• Catalog-driven: The AI accesses your live pages using the product URLs provided in your Bazaarvoice Product Catalog. If a URL is missing or broken, the AI cannot scan the page.
• Standard permissions: Certain anti-bot systems can have the side-effect of blocking Answer Assistant from retrieving information live from your PDP. It will only read pages that your IT team has permitted for web crawlers.
• Data synthesis: To generate a draft, the system combines your live PDP context with your catalog product descriptions, approved consumer reviews, and historical Q&A data.

Refer to the Bazaarvoice use of AI article. 

Bazaarvoice completely respects your site's security protocols as a deliberate choice. The AI will use your other available data, like your product catalog and past approved answers, to build the response draft.

If an AI-generated answer is available for a question, you'll see the Generate answer tooltip when you hoover over the AI sparkle icon.  

Yes. Following Bazaarvoice AI trust standards, any response that uses AI, whether fully generated or partially edited, will display an AI disclosure label on the PDP to maintain complete transparency. 

• "AI-generated, Human-approved": The full response was created by the AI and submitted without edits.
• "AI-assisted, Human-approved": The response was generated by the AI but partially modified by a user before submission.
•  No label: No AI disclosure label will display if the answer was written organically by a user before submission.
  

An answer was created entirely by the Answer Assistant based on product content and consumer feedback. This answer is submitted without any manual edits.

An answer is initially generated by the AI but is modified or personalized by a user before it’s submitted. The AI answer is only partially generated by the AI answer assistant.

Yes. Answers published using Answer Assistant are saved as standard Q&A content. If they’re published on your website, they display normally within your consumer-facing Q&A solution. If they’re published using the API, they are automatically available through the Authentic Discovery API and accessible to search crawlers and scrapers. 

No, Answer Assistant will never publishes content automatically. It operates with strict client-side, human-in-the-loop control, meaning a customer care agent on the client side must always review, edit, and verify the accuracy of every single response draft before it goes live.

Our AI model is trained to recognize and intentionally ignore high-risk questions requiring medical, financial, or legal expertise.

The AI prioritizes accuracy over guesswork. If it lacks sufficient context, the draft response box will remain empty. You're be directed to our best practices for increasing your answer coverage. For example, make sure your catalog's product descriptions are at least 50 words long.

All reviews used to generate answers follow our standard moderation process. This includes a review by both our Natural Language Processing (NLP) tools and human moderators designed to ensure the content is appropriate, relevant, authentic, and compliant with all quality standards.

You have 100% control. Answer Assistant handles the initial drafting to give you a head start on consumer inquiries. From there, you must review, modify, or personalize every answer before it goes live on your PDPs.

No. Unlike reviews, which typically undergo rigorous authenticity checks, many answers (especially from clients) bypass standard moderation. Most clients use self-service moderation for Q&A, meaning they’re responsible for overseeing their own content rather than relying on Bazaarvoice’s moderation team. 

You, the client, are responsible for the AI assisted answers submitted and published on your own website 

Yes. Bazaarvoice aligns all of its products with major global frameworks. Because the regulatory landscape evolves rapidly, we treat compliance as a continuous process and a shared responsibility.

Our client-facing portal includes a clear and conspicuous disclosure that the AI-generate and AI-assisted drafts were created by AI. However, these drafts must be reviewed, edited, and approved by you, the client, before being published.

As the final approver and submitter of the answer, you remain responsible for your own regulatory assessments. 

You, the client, retain sole responsibility for all content submitted and published to your live site. Answer Assistant is designed to generate initial drafts; it will not publish answers autonomously. Your customer care team must always review, edit, and verify the accuracy of a draft before submission.

No. Answer Assistant operates strictly behind the scenes within Portal. It’s an internal tool designed to assist your customer care team. It’s not a consumer-facing chatbot. Your consumers will only ever see the finalized answers that your human team explicitly approves and submits.

Based on current Bazaarvoice practices, Answer Assistant does not process personally identifiable information (PII). The system generates responses using only your product catalog data, PDPs, and user-generated content (UGC) like reviews and past questions and answers.

All UGC passes through Bazaarvoice moderation to remove PII before the AI model interacts with it.

The AI prioritizes accuracy over guesswork. If it lacks sufficient context, the draft response box will remain empty and the AI generation available icon will not display.  
To increase your answer coverage, make sure your catalog's product descriptions are at least 50 words long.

No. Our AI model is designed and engineered to recognize and intentionally ignore high-risk questions requiring medical, financial, or legal expertise.

No. To ensure absolute data isolation, Answer Assistant only processes a client's data to generate drafts for that specific client. This real-time processing occurs entirely within our secure private cloud, and submitted answers are never used for cross-client model training or optimization.

Based on current Bazaarvoice practices, our generative AI features are powered by leading foundational models hosted within secure cloud environments. We select our models by evaluation of their speed, performance, and accuracy. To ensure you always receive optimal results, we continuously evaluate the AI landscape and upgrade or change the underlying models as technology evolves.