Conversational AI is a revolutionary technology that makes it possible to interact with computers in a more natural way. From intelligent virtual assistants to intelligent search, conversational AI powers transformative self-service experiences through new and existing digital channels.

As companies look to integrate conversational AI solutions into their existing tech stacks, it can be difficult to understand how to prioritize ever-evolving customer expectations against the needs of the business, as they relate to this technology. It’s hard to determine whether or not the conversational AI solution you’re considering will be able to flex and scale as needed. Read on for tips on how to evaluate vendors’ ability to help your business grow—and what questions you should ask before agreeing on a contract:

How to document business and technical requirements.

As you begin to evaluate conversational AI platform vendors, it is crucial to define the business and technical requirements of your project. The best way to do this is by creating a scoring matrix that can be used across all vendors. This will help you rank each vendor based on how closely they meet your criteria.

Identify Stakeholders

In most organizations, anything that could be viewed as a “digital” solution is highly likely to involve multiple stakeholders. From IT to Marketing, Customer Experience to Legal & Compliance, it’s critical to have all impacted parties come together at project kickoff in order to ensure that everyone’s needs are met.

Set Goals

Before starting on a solution, you want to first define the problem. Doing so will help focus all parties on what’s important and prevent any misunderstandings or miscommunication later in the process (this can happen even within teams.) It will also help set expectations among stakeholders involved in the project as well as ensure everyone starts with a common understanding of what success looks like for the initiative.

Categorize & Score Your Requirements

When you first sit down with stakeholders to walk through their individual and collective wants and needs, you’ll likely end up with an overwhelming list of items. Planning for a way to categorize and assign values to each item can help you better understand the nice to haves and need to haves without too much or not enough detail to be useful.

Why Your Requirements Matter

As we’ve likely all experienced at some point in our careers, not having clearly defined requirements prevents us from achieving our goals efficiently because we’re spending time working around things instead of working through them. Don’t be afraid to include requirements like, “Team is responsive” and other things that will likely have a strong impact on your ability to do business with them long-term.

How do you know if a technology will scale to meet your current and future needs?

While there’s no foolproof way to predict the future, starting with a baseline understanding of your current needs and involving the right stakeholders in a deep discussion and documentation session about projected future needs, is a great start.

Once you have a decent understanding of where your organization is headed, it’s time to engage the prospective vendors to see if they’re headed in a similar direction. For instance, if you’re planning to update your core tech stack, ask about the ability to set up custom configurations (e.g., multi-channel integrations) or configurations that allow you to integrate with planned vendors’ tools? Questions like this may naturally take the conversation from their current capabilities to their planned capabilities.

You’ll want to dig into their product roadmap. Get an understanding of how they prioritize product updates and what the typical timeline for new features is like. Ask them if they can share upcoming planned features and any timing for their releases. Ask them how they measure and react to customer satisfaction (if they’re paying attention to this, it’s a good sign) and ask if they can share their customer retention rate.

If the vendor is able to share their product roadmap, evaluating it alongside your projected future needs and engaging in a discussion with the vendor about any potential gaps, is a helpful exercise.

Look at conversational AI specific needs

When evaluating conversational AI technology, it’s important to look at the specific requirements of your project. How accurate does the model need to be? Which languages will it need to understand? Will there be industry-specific training needs for the model, or is it a generic solution? How much work will be required moving forward to maintain and tune the model as the underlying technology evolves?

You’ll want to ask these questions early on in order to help evaluate if a particular conversational AI solution fits with your needs.

How does the vendor handle security and privacy?

Security and privacy are of utmost importance for any business or individual using technology, especially one that could process (intentionally or not) Personal Identifiable Information (PII). This is no less true when it comes to conversational AI and digital self-service solutions. There are several things you should consider when evaluating the security and privacy practices of a vendor.

• What data do they collect?
• How is this data stored and secured?
• Who has access to this information, and what are their roles within the company?
• What policies are in place regarding how this information can be used or shared with third parties?
• What is their data retention policy?
• Are they compliant with the laws and regulations that your company is governed by?
•  

Vendor questions to gauge flexibility and customer service

In addition to the high-level questions above, there are also a few more specific things to keep in mind when evaluating vendors:

Customer Communication

Many vendors will have separate teams for the sales, onboarding, and ongoing account management portions of the customer relationship. Ask them what the communication looks like throughout each of these phases. Helpful questions can look like:

• How do customers contact you with questions or issues throughout the relationship lifecycle?
• What’s your SLA and response time for issues?
• How do you handle communication around new features releases, bugs, and planned downtime?
•  

Run the numbers and select a vendor

Once you’ve completed the evaluation of the prospective vendors, look at how they each stack up against your requirements list. Regroup with the stakeholders for a final review and then kickoff the contract review process and get to building!

We’re excited to see what the future holds for conversational AI. We hope that this guide makes you feel more confident in your procurement process as you evaluate different conversational AI vendors to support digital self-service.