Conversational Search

Conversational search is an AI-powered search experience that allows your website visitors to ask questions in their own words, get direct AI-generated answers, and engage in multi-step queries where follow-up questions build on previous responses.

Traditional search requires audiences to use the right keywords in their query, presenting a list of results for them to browse through in the hope it might contain the information they are looking for.

While traditional search places a cognitive load on audiences to find their answer, conversational search works like a knowledgeable assistant that understands user intent and instantly delivers the specific information they need, eliminating the need for users to trawl through search results. 

The key differences between conversational search and traditional search are: 

• Input method: Traditional search relies on keywords; conversational search accepts natural language questions.

• Output format: Traditional search returns links to pages; conversational search provides direct answers from relevant content.

• Context awareness: Traditional search treats each query independently; conversational search maintains context throughout an interaction.

• User experience: Traditional search requires users to hunt through results; conversational search delivers immediate, specific answers.

• User intent: Traditional search matches keywords; conversational search understands the meaning and intent behind questions.

Conversational search can make a significant impact on both user satisfaction and business outcomes: 

• Boost conversions by making things easier: When visitors can simply ask questions and get clear answers right away, they're more likely to take action.

• Reduce support costs: With conversational search handling common customer questions, your team is able to focus on more complex issues that need a human touch.

• Learn what your visitors really want: By seeing what natural language questions users are asking, you get a deeper understanding of their true needs and intentions. These insights highlight content gaps and inform strategic decisions in ways that traditional keyword analysis can’t match.

Choosing the right conversational search solution requires careful evaluation of several key factors. The most important criteria to consider include the underlying search technology, performance standards, how it ensures answer accuracy, and how well the solution integrates with your existing systems.

When evaluating conversational search tools for your organization, be sure to ask:

• How effective is the underlying search technology? The tool's ability to find truly relevant content is essential. No matter how smooth the conversation feels, it fails if the underlying answers are incorrect or unhelpful. Look for proven search technology with a strong track record of surfacing accurate information.

• How well does it deliver on critical requirements: faithfulness to scope, performance, and accuracy?  A superior conversational search excels at all three requirements: strict adherence to your approved sources (ensuring responses reference only your content, not external knowledge), fast response times that meet user expectations, and precise answers that directly address the question asked. Many solutions perform well in only one or two areas, so evaluate thoroughly against all three critical requirements.

• What governance controls are available? Choose a solution that gives you comprehensive oversight of AI-generated answers. You need tools to audit responses, monitor performance, and make adjustments to ensure your AI represents your organization appropriately.

• How is user data handled and protected? Understand how user queries and interactions are stored, who has access to this data, and whether it's used to train models. Ensure the tool complies with relevant regulations and allows you to maintain control over sensitive information.

• How does it address responsible and ethical AI practices? Look for tools that demonstrate ethical AI principles in how they handle your content and generate responses. The best solutions prioritize transparency about AI use, proper attribution of sources, and fair representation of your content without mining or misusing it. Ethical considerations should be built into the system's design rather than added as an afterthought.

• Will it integrate with your existing systems? The tool should connect seamlessly with your existing content sources, authentication systems, and analytics platforms. Consider whether it can access all your relevant information, including protected internal content when appropriate.

• Can you still control and maintain your brand?  Ensure the conversational experience can be adequately monitored and managed to maintain brand voice, guide response formats, and direct users toward specific actions that align with your goals.

• What analytics and insights will you get? Strong reporting features help you understand how users interact with the tool and identify opportunities to improve both the search experience and your underlying content.

Conversational Search by Squiz Funnelback offers superior control, accuracy, integration, and user experience compared to traditional chatbots, making it a more robust and future-proof solution because of these reasons:

• Content control and accuracy: Conversational AI Search – built on Squiz Funnelback – ensures that answers are generated strictly from your approved and up-to-date content. In contrast, many chatbots may require separate training and can become content silos, meaning you have to maintain information in multiple places, increasing the risk of outdated or inconsistent answers.

• No content duplication: With Conversational Search, you don’t need to duplicate your content or maintain separate knowledge bases for the chatbot and your website. This reduces effort and ensures that improvements to your website content benefit all channels, including search and external AI platforms.

• Governance and safety: Conversational Search by Squiz Funnelback includes built-in controls to prevent hallucinations (incorrect or fabricated answers). It also allows you to monitor, curate, and audit responses. For example, the system can suppress or modify answers that don’t meet accuracy thresholds, something most generic chatbots can’t do.

• Integration with search engine: Conversational Search by Squiz Funnelback leverages precision search capabilities (like vector search and retrieval augmented generation) to understand context, synonyms, and user intent, providing more relevant and precise answers than traditional chatbots, which may not be as sophisticated in query understanding.

• Analytics and visibility: Conversational Search provides analytics on what users are asking and how well your content is performing, giving you actionable insights to improve your content. Chatbots may lack this level of transparency and integration with your main content ecosystem.

• User experience: Conversational Search offers a natural language, context-aware experience that aligns with user expectations set by external AI Search tools like ChatGPT and Google, ensuring your site remains competitive and user-friendly.

Search Generative Experience (SGE) refers to AI-powered search tools that provide direct, conversational answers rather than just serving up a list of links. Squiz conversational search is an example of an SGE. 

When you're using a search engine with SGE functionality, you'll see a helpful summary at the top of your results that draws information from multiple sources. These systems typically cite where they got their information from and still give you access to the traditional list of search results below – which can be helpful when you want to dig deeper. 

Squiz conversational search enables users to benefit from a Search Generative Experience (SGE), allowing them to ask questions in their own words and receive instant, accurate answers.

Generative Engine Optimization (GEO) is a marketing approach that focuses on getting your content included in AI-generated search summaries rather than just appearing in regular search results. While traditional SEO helps your website rank higher in search listings, GEO helps ensure your information appears directly in the AI-created answers that users see first. This means creating content that's easy for AI systems to understand and use, organizing your information clearly, and adapting how you write to match how AI systems process information.

Squiz conversational search ensures accurate answers through rigorous quality controls and multiple verification layers designed to reduce errors and deliver reliable responses. Our approach combines superior search technology, content governance tools, and built-in validation steps that work together to maintain consistently high answer quality across all interactions.

Powered by superior search technology: Squiz conversational search is built on Squiz Funnelback Search, our enterprise search solution with 25 years of expertise.

We consistently outperform competitors in retrieving the most relevant information to power your AI answers, through features like:

• Multi-source indexing: Access content from anywhere including third-party platforms and internal systems, not just your primary CMS.

• Search packages: Create purpose-built knowledge bases with granular control over exactly which content sources to include or exclude, ensuring AI draws only from approved sources.

• Tuning: Train the algorithm to prioritize your most authoritative content for AI-generated answers, dramatically improving accuracy and trustworthiness. Our system leverages over 70 ranking factors to automatically evaluate and optimize results based on your training data. 

• Curator: Tailor search results based on user queries, attributes, or behavior to deliver personalized, relevant answers.

This retrieval step forms part of a Retrieval-Augmented Generation (RAG) framework, where AI generates answers strictly based on what’s been retrieved – ensuring content fidelity.

Ensuring content quality for reliable AI responses: Your AI will only be as good as the content that feeds it – this foundation is critical for answer accuracy.

Squiz conversational search comes with a complete set of governance tools for auditing and monitoring content quality. These include: 

• Conversational readiness: Get your content AI-ready before implementation. Simulate how AI responds to common customer questions, identify content gaps, and get actionable insights to improve your content's accuracy.

• Conversational monitoring: In the first iteration of conversational search, you’ll be able to monitor and review conversation history to ensure customers are getting high quality answers. Soon, you’ll also be able to receive alerts when conversations aren’t meeting customer needs, allowing you to make timely improvements. 

• Content health check: Maintain the accuracy of your content with content auditing tools. Soon, you’ll also get proactive alerts about content issues before they impact your customer experience, ensuring your AI continues to perform effectively as your content evolves.

Built-in checks and balances on AI-generated responses: Squiz conversational search uses a two-step verification process to ensure response accuracy. The system first generates an answer based on your content, then automatically runs that answer through a built-in validation step.

During this validation step, the system performs a “faithfulness check” – verifying that the generated response draws only from your approved content sources and doesn’t include any external information. This two-step process ensures that all responses remain accurate and grounded in your organization’s content.

No. If the information isn't in your content, the LLM will not make up an answer. Instead, it will simply let users know it doesn't have enough confidence to answer, or guide them toward using traditional keyword search instead. This ensures users only receive information that actually exists in your content sources.

Squiz maintains essential guardrails to ensure our chosen LLM delivers accurate and appropriate responses. However, you will have full control over exactly what information is used to generate responses, by defining your trusted content sources. You will also be able to manage the style of language, tone of voice and level of detail the LLM uses to generate a response. This gives you meaningful control over your search experience while we handle the technical aspects that keep everything running smoothly.

To set up and launch conversational search on your website, focus on starting small and building on success rather than attempting a complete overhaul all at once. Our recommended steps include: defining a focused use case, evaluating your content quality, connecting to your content sources, deploying the conversational search experience, and then expanding to additional areas – allowing you to implement conversational search in manageable steps while demonstrating value along the way. 

Here's a breakdown of the steps to implement conversational search: 

1. Define an initial use case: Begin with a focused use case that presents a low risk, high confidence, and high impact starting point – for example, your staff directory. This lets you learn and demonstrate value quickly.

2. Analyze your current search data: Look at your existing search analytics to see what questions users are asking today. 

3. Evaluate your content quality: Use tools like the conversational readiness tool, and run a content audit to identify content gaps, inconsistent terminology, and duplicate information that might affect AI performance.

4. Update your content where needed: Address the issues found in your audit to ensure your AI has quality information to work with.

5. Set up your first implementation: Deploy conversational search for your chosen use case, connecting it to your relevant content sources.

6. Use monitoring tools to refine performance and content: Review and monitor conversation history to ensure they are meeting customer needs. Use content auditing tools to catch any content quality issues early.

7. Expand to additional areas: Once your initial project is working well, gradually apply conversational search to other parts of your digital presence.

This practical approach helps you implement AI search in manageable steps while demonstrating value along the way.

If you’re already using Squiz Funnelback Search for your site search, you’ll be able to use that as your foundation for conversational search – so you won’t need to replace anything. If you’re using a different site search, you can selectively implement Squiz Funnelback Search on specific sections of your website(s) where you want conversational search to work. However, in that scenario, we would recommend replacing your current site search with Funnelback rather than having to juggle two search tools.

When budgeting for conversational search, you'll need to account for both the technology investment and related preparation work.

Technology investment: For Squiz conversational search, the subscription fee is based on the number of indexed documents and the number of conversations (i.e. Q&A-style flows with site visitors) per year. 

Content preparation costs: Before implementing AI search, you'll need to ensure your content is ready. This means auditing existing content for accuracy, comprehensiveness, and gaps that might affect AI performance. Depending on your content volume and current quality, this may require time from your internal marketing and content teams or support from external consultants.

Implementation costs: Technical integration will also require some resource allocation. This will typically involve:

• Internal team time: If using your existing staff, allocate time for developers to handle the integration with your website systems

• External support: For organizations without in-house capacity, Squiz's professional services team can manage the entire implementation process, with pricing options tailored to your specific requirements

Many organizations find that a hybrid approach works best, with your team handling content-related tasks while Squiz experts manage the technical configuration, minimizing both costs and demands on internal resources.

Return on investment: Conversational search typically delivers significant returns through:

• Increased conversion rates as users find what they need faster

• Reduced support ticket volume as common questions get answered automatically

• Time savings for both users and internal teams

Most organizations find that the efficiency gains and conversion improvements from conversational search implementations create substantial ROI that quickly outweighs the initial investment and ongoing costs.

Squiz Funnelback delivers the most advanced and secure enterprise search platform with AI-powered conversational search capabilities. According to Squiz clients, we are the intelligent search vendor of choice because of:

• Complete content control: Squiz Conversational AI Search uses a Retrieval-Augmented Generation (RAG) framework. It retrieves content using Squiz Funnelback’s enterprise-grade search engine that ensures only the content you want to crawl is included in the response. No external sources, no surprises, just accurate responses based on what you've already published and defined as accurate.

• Enterprise-ready search governance: Built-in AI safeguards automatically verify answer accuracy and flag problematic queries before they reach users. You maintain full control over what gets displayed with Squiz Funnelback.

• Actionable search analytics and intelligence: See exactly what your visitors are asking and how well your content performs. These insights help you optimize your site and prepare for the AI-driven future of search.

• Scalable search deployment: Have the flexibility to roll out a rapid pilot with specific content areas, before scaling to a full implementation. Squiz has pre-built search templates to deploy fast.

• Inclusive and accessible search: Accessibility-compliant search interfaces with screen reader support and potential for future voice interface readiness.

No, Squiz Conversational Search does not support multiple Large Language Models (LLMs). One LLM has been chosen to power the answers generated by Squiz Conversational Search. This has been identified by Squiz as the best LLM for the job and cannot be replaced by alternative options.

We recognize finding truly experienced people can be difficult, so you can partner with us to implement Conversational Search on your website. Our services team has extensive experience with Conversational Search implementations tailored to your specific needs, with a proven process designed for optimal performance and ongoing support.

That said, if you have an existing relationship with an implementation partner who has experience with optimizing content for generative engines, we’re happy to work with them to set up conversational search on your website. We can explore collaboration opportunities on a case-by-case basis, though we recommend discussing your requirements with our team first to determine the best approach for your project.

Yes, you can find technical documentation that is specific to Conversational Search in the Squiz Help Center, here: https://docs.squiz.net/funnelback/docs/latest/conversational-search/index.html