How would Conversational AI Search work for me?

Search is evolving fast.

People no longer want to guess keywords and trawl through a list of results. They want to ask questions and expect human-like responses.

Organizations need to adapt their digital strategies to meet these changing expectations.

But what exactly is conversational AI search, and how could it work for your organization? Let's dive in.

Skip ahead:

• How is conversational AI search different from traditional website search?
• How does conversational AI search technology work?
• Is Conversational AI Search compatible with existing systems and technologies?
• What are the upfront and ongoing costs?
• What is the expected return on investment?
• Ongoing considerations for success
• Conversational AI Search: Not just better search, better experiences

Think about using a traditional search bar on any website. You test different keywords, hoping to pull up the right content, then browse through a list of links that might contain what you're looking for.

Your website visitors go through this same process when searching your site. Their journey to find an answer or complete an action quickly turns into a research project.

In contrast, conversational search offers an AI-powered experience that allows users to ask questions in their own words. It mimics a human-like interaction, understanding intent and supporting multi-step queries – where users can ask follow-up questions that build on previous answers – to help users get the specific information they need, with specific answers sourced from your content.

The AI-generated answers come directly from your own content (and only the content you wish them to come from – no hallucinations in sight!).

So, while traditional search places a cognitive load on audiences to find their answer, conversational AI search works like a knowledgeable assistant. It understands user intent and instantly delivers the specific information they’re looking for.

Squiz’s Conversational AI Search tool works through a simple but powerful process:

1. Your visitor asks a question in their own words through the search bar.
2. Behind the scenes, Squiz Funnelback Search immediately finds the most relevant content from your approved sources to answer that specific question.
3. Squiz Conversational Search then transforms that content into a clear, direct answer, complete with links to supporting pages so users can complete their task or explore related topics.

In more detail, this process includes 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. It then generates an answer using AI. Finally, it automatically runs that answer through a built-in validation step that checks responses for accuracy against your content and the question asked.

This framework was also built to handle complex questions that include more than one concept, like 'How do I apply for a postgraduate scholarship if I’m an international student?', where the system must search across multiple content sources and combine details into one cohesive answer.

Here’s how it works, step by step:

1. Smart retrieval with Funnelback: When a user submits a query, Funnelback retrieves the most relevant content from your authorized data. This includes structured and unstructured content across your site, intranet, or any other integrated systems.
2. Prompt pairing: The retrieved content is then paired with a prompt template, which controls the tone, format, and fallback behavior of the AI response (e.g., what to say when no answer is found).
3. Answer generation within your environment: The Large Language Model (LLM) selected by Squiz and hosted within Squiz’s secure DXP environment, generates an answer using your content only - never general internet data. (Note: the LLM used cannot be replaced by an alternative - Squiz has picked the best one for the job!)
4. Built-in accuracy checks: Once the system generates the answer, it 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.
5. Response delivery with full visibility: The user receives a direct answer with source attribution and can continue the conversation naturally. These interactions are constantly analyzed for performance, accuracy, and sentiment, giving you insights needed to improve the user experience, with full visibility and control through the Conversational Health Monitor.

Example of this in practice:

The user asks: “How do I apply for a postgraduate scholarship if I’m an international student?”

Behind the scenes, Squiz Conversational Search uses the RAG process:

1. It first retrieves relevant, verified content from your approved data sources (e.g. a postgraduate scholarships page).
2. It then augments that content and generates a plain-language answer using that content only, never from general internet data.
3. It automatically runs the answer through a built-in validation step and checks the answer for accuracy before it’s delivered to the user.

Squiz Conversational Search then responds:

“To apply for a postgraduate scholarship as an international student, you’ll need to complete the online application form by July 31. You must have an offer of admission and meet the eligibility requirements outlined here [link].”

This multi-layered verification process we just described is how Squiz ensures the answer provided is accurate.

One thing many organizations discover too late is that AI-generated answers are only as good as the underlying website search results. Without exceptional search capabilities, even the most advanced AI will deliver disappointing results.

So, our conversational interface is powered by Squiz Funnelback Search for a critical reason: exceptional conversational experiences require exceptional search technology.

Squiz Funnelback Search is designed to integrate seamlessly with any website platform or content management system. So, there's no need to overhaul your entire website or replace your tech stack to start using it.

If you’re already using Squiz Funnelback Search on your website, great! You’ve already got the foundation needed.

If not, you will need to implement Squiz Funnelback Search on the parts of your website(s) you want Conversational AI search on. In this case, we would recommend replacing your current site search completely to avoid juggling multiple search tools.

The Funnelback implementation connects with your existing content sources, authentication systems, and analytics platforms, ensuring comprehensive access to all the information you want it to examine.

Handling all types of content

Squiz Conversational Search can process both structured and unstructured data across your organization.

What does this mean? Many of your most valuable information resources don't fit neatly into traditional databases with defined fields and formats. Think of PDFs, long-form articles, social media feeds or documents with varying layouts – this is unstructured data.

Squiz Funnelback Search, the search engine that powers Conversational AI Search, was specifically built to handle this challenge. It can analyze and extract meaningful information from these diverse content sources to power accurate AI responses.

Introducing Funnelback’s new Conversational AI Search functionality involves two key types of upfront investment: technology and its related preparation work.

The other consideration at this early stage is whether your content is AI-ready. The good news is the Squiz team is here to help.

Technology investment

AI search pricing varies significantly across providers, with most basing licensing and subscription fees on usage metrics such as the number of search queries, data volume and indexing, and the use of any advanced features.

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.

Our pricing model is different because we built it around the metrics that matter. The content you're searching and the conversations you're having with customers.

Unlike other models, our business-centric pricing model means finance teams can easily understand and forecast costs based on real business drivers, not technical complexities. This enables you to plan your search investment with the same clarity you use for other business-critical tools.

Content preparation

Tools like ChatGPT and Gemini are already scraping and interpreting your website content to answer user questions, whether you've optimized for AI or not. By proactively auditing your content for accuracy, comprehensiveness, and clarity, you're not only preparing for your Conversational AI Search implementation but also ensuring that external AI tools provide correct information about your organization.

This preparation typically involves reviewing your existing content to identify gaps, inconsistencies, or outdated information 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.

The benefit extends beyond just powering Conversational AI Search. It improves your overall content quality, which enhances user experience and helps you maintain control over how your information is presented across all AI platforms.

Implementation costs

Getting your Conversational AI Search up and running is straightforward and flexible, with options to suit your organization's resources and capabilities:

• Internal team collaboration: If you have technical resources available, your developers can handle the integration with guidance from Squiz, allowing for valuable knowledge transfer and skill development within your team.
• Fully managed implementation: For organizations focusing on their core business, Squiz's professional services team can manage the entire process from start to finish, with tailored solutions that align with your specific requirements and timeline.

Many organizations find that a hybrid approach works best. Your team handles content-related tasks where they have the most expertise, while Squiz manages the technical configuration. This partnership approach optimizes resource allocation while ensuring a smooth implementation.

Conversational AI Search typically delivers significant returns through:

• Improved user engagement and reduced bounce rates as users are more likely to stay on your site and take action rather than leaving without answers.
• Increased conversion rates as users find what they need faster.
• Reduced support costs with Conversational AI Search handling common customer questions, allowing your team to focus on more complex issues.
• Deeper user insights by seeing what natural language questions users are asking, you get a better understanding of their true needs and intentions.
• Highlighting content gaps to inform strategic decisions in ways that traditional keyword analysis can't match.

Most organizations find that the efficiency gains and conversion improvements of Conversational AI Search create substantial ROI that quickly outweighs the initial investment and ongoing costs.

Maximizing the value of your Conversational AI Search involves these important ongoing considerations for the organizations that have implemented it:

Content quality management

Conversational AI search responses will only ever be as good as the content that the organization indexes as a source for answers. This foundation is important for answer accuracy. You may need to allocate resources within the organization for:

• Regular content audits to identify and address information gaps
• Updating content that becomes outdated over time
• Creating new content based on conversation patterns to fill identified needs

These content activities require resources but deliver benefits across your entire digital presence, not just for Conversational AI Search.

Governance and monitoring

As an organizational user of conversational AI search you can ensure it continues to perform well by:

• Reviewing conversation history to identify successful and unsuccessful interactions
• Analyzing performance metrics to spot trends and opportunities
• Adjusting search configurations based on real-world usage patterns

Please note, while you, as a website owner, will be able to monitor and review conversation history to ensure customers are getting high-quality answers, this is not a feature that is available to the customer (end user) at this time.

Training and skill development

Your team may need to develop new skills to effectively manage Conversational AI Search, including:

• Learning to structure content in ways that work well for both humans and AI
• Understanding conversation analytics and what they reveal about user needs
• Making informed decisions about content strategy based on AI interaction data

While these activities require planning and resources, they represent valuable capabilities that strengthen your overall digital strategy and user experience.

Search is changing fundamentally. As users increasingly turn to conversational interfaces for answers, the organizations that adapt will be the ones that stay relevant and accessible.

The question isn't whether Conversational AI Search will become the standard – it's already happening. The question is whether your organization will adapt.

If you are interested in seeing how conversational search could work for you, just book a call and we can show you it in action.