If you are looking to implement a chatbot for your company, it’s important that you ask the right questions. It might be a stressful decision to evaluate data security, implementation time frames, industry expertise, and overall general selection of vendor criteria. This post shows you some questions that you can ask a chatbot vendors in the market.
Choosing a platform for your business’s chatbot should not be difficult. Request for Proposals (RFP) are a formal process to have a clear understanding of the capabilities of each of the different vendors and what they have to offer. If you don’t issue an RFP and go by demo, you may miss out on some opportunities to ask more in depth questions because the salesperson may have easily hid or talked their way out of answering a question in a video demo or a live demo.
As someone who has not only answered Requests for Proposals for information technology solutions and someone who has solicited Requests for Proposals for chatbot vendors, I think I’ve come up with some of the best questions to ask vendors in this space of emerging technology.
I’ve categories these questions and considerations you should be asking during your information gathering phase. I’m sure that these questions will also help you consider your own needs for a chatbot as well.
Before Requesting a Proposal to a Chatbot Vendor
Requesting a proposal is often a very formal and time consuming process for both you and the vendor. Below are three things that you’ll want to consider before sending an RFP out so that you are not wasting time and effort.
- Understand What Your Company’s Objective for the RFP
- Use a Template to Help Guide the RFP Process
- Ask Deal Breaker Questions Up-Front
I’ll go more in depth in each of the considerations above in a future post. For now, I wanted to at least provide you with some information on the types of questions you should be asking other vendors and maybe even your own organization.
Best Questions to Ask in a Chatbot Request for Proposal
Below are some of my favorite questions to ask when talking to chatbot platform vendors. Some of them may not be pertinent to your workflow, so you can skip asking them. I would recommend asking the ones that are specific to what you are trying to accomplish within your own organization.
During my own personal experience, most vendors couldn’t do all the things that can be presumed from asking the questions, but a majority of them could do at least the important parts when reviewing the list of specific requirements needed for the business use case.
- When was the company founded and when was the chatbot product line developed and deployed?
- Who are your biggest competitors?
- Describe the chatbot use cases that the chatbot market best knows your company for.
- Are there any other customers that you work with in the past that are in our industry?
- What is your platform’s key features and advantages when compared to other chatbot vendors?
- Who is your largest client and what are their number of concurrent users?
- How many of your clients are Fortune 500 companies?
- What is the largest number of intents you have for a single bot?
- How do you define a single bot instance?
- Do you support a universal bot concept (re: master and child bot)? What are the number of child bots that a universal or master bot can support?
- How does your chatbot platform handle global or launch commands when there are multiple instances of bots that are underneath a universal bot?
- Does your chatbot platform support workflow or business processing mapping with the ability to relate conversation dialogues to a BPM?
Chatbot Engagement Channels
- What use cases does your platform support?
- What engagement and digital channels can your platform create platforms for?
- List all the different channels that your platform can support.
- Do you support channel specific conversation dialogues and workflows? Can this be configured in the administration console?
- Can your chatbot deploy through intranet for custom web applications?
Chatbot Platform Language Support
- What text based languages are supported by your platform?
- What voiced based languages are supported by your platform?
- Can your chatbot support enabling multiple languages for a bot without having to rebuild the bot?
- Can your platform create translation steps?
- Can users set their own preferences on what skills can be enabled at their user level?
- Can users set their own preferences on what skills are enabled based on role level?
- Can your platform learn from user interactions and/or roles which users should receive certain skills?
Chatbot User Experience
- Does your chatbot platform have the ability to @ mention users?
- Does your chatbot platform have the ability to use # to mention discussions?
- How does the platform display channel-specific messages to users?
- Does your platform have a built-in spell-check?
- How does your platform handle the ability to display messages?
- How does your platform handle the ability to display video?
- How does your platform handle the ability to display reminders and notes?
- How does your platform handle the ability to handoff to a live agent or involve human in the loop?
- How does your platform handle the ability to manage a chat queue?
- How does your platform handle the ability to ongoing chats in different engagement channels?
- What design features does your platform handle better than your competitors that can improve overall user experience from both the chat user, bot builder, and administrators?
Chatbot Platform Administration Console Configuration Options
- Can user interaction or role based preferences be done programmatically in an administration console with no code?
- Can your platform support bot or skill builders having different views based on their role profile?
- Does your platform include support for development lifecycles across multiple teams?
- Does your platform support bot building governance and approval processes?
- Does your platform have a global management of skills in your platform?
Chatbot Natural Language Processing Engine
- What is the core natural language processing engine (NLP)?
- What is your strategy when it comes to chatbot NLP?
- Does your platform provide conversational templates for our industry’s use case?
- What are the current pre-built entities that you recognize?
- What are the future pre-built entities that you will recognize within the next two years?
- If provided an identity management data source, does your platform’s NLP engine automatically recognize person entities with utterances?
- Does your NLP support fuzzy matching for person name entities with low code or no code?
- Does your platform have built-in support for common misspellings of supported built-in entities?
- What is the minimum proposed user utterances that are needed to train NLP?
- Is your platform NLP system agnostic to allow third party NLP engines to be used?
Chatbot Conversational Capabilities Around Context Switching and Sentiment Analysis
- Does your chatbot platform support context switching?
- Does your platform support conversation memory?
- What types of contexts does your chatbot platform support?
- How does context switching from bot to bot work?
- Can your chatbot control context switching in multi-turn dialogues?
- Does your platform support multiple intent detection with the ability
- Does your platform support batch testing and a detailed view of statistical analysis to gauge the performance of a bot?
- Does your chatbot platform support detecting the sentiment of an interaction?
- Does your chatbot platform detect sentiment of the interaction help programmatically steer the dialogue conversation?
- What are the type of sentiments that your chatbot platform detects?
- How does your chatbot platform score the sentiment analysis?
- Does your platform support developer provided utterances to train the bot?
- Does your platform support real-time data that the platform gathers over time to train the bot?
- Does your platform support chatbot transcripts to help facilitate machine learning?
- Does your platform provide supervised and unsupervised training to expand the language model?
- Does your ML model improve accuracy by programatically excluding failed utterances?
- Does your platform provide a list of identified, unidentified, and missed intents for a human to judge?
- Does your chatbot support greeting topics and common conversation dialogues?
- Does your platform support custom error prompts for each entity?
- Does your platform have the ability to collect user feedback?
- How does your platform provide user feedback?
Chatbot Data and Analytics
- What metrics do you have pre-built to measure chatbot performance?
- How long do you keep data and analytics reporting?
- Do you have data structures for different types of failure and reporting on false positives?
- How do you support data standards for failures, successes, and ability to categorize them within a bot?
- Do you provide the ability to programatically ingest the data in other systems?
- Is there a GUI interface to analyze and review chat sessions?
- Does your platform provide the ability to create custom dashboards?
- What types of reports are available on the platform dashboards available?
- Is there an ability to download full session transcripts?
Chatbot Hosting Architecture
- Does your platform support uploading documents in conversation flows? If so, what types are supported?
- Is your platform a SAAS platform? If so, where is it hosted.
- Is your platform hosted in a private or public cloud? What region is it hosted in?
- If the platform is in the public cloud, does each customer have their own tenet?
- What are the supported identity authorization models?
Chatbot Deployment and Bot Styling Toolkits
- How are deployments handled to ensure minimal downtime?
- How does your platform support low code or codeless bot development?
- Does your platform provide command line interfaces and/or SDKs?
- Does your platform provide extensibility?
- Does your platform support a customs script files to be reused?
- Does your platform have an SDK for web applications, Android, and iOS?
- Does your platform have the ability to have global variables that are used across bots to avoid hard coding?
- Can your platform support reusable components if custom nodes or blocks are built for a bot or a conversation?
- Can your bot be deployed under multiple personas based on channel or audience?
Chatbot Security and Privacy Considerations
- Is your platform GDPR compliant?
- Is your platform HIPAA compliant?
- Do you have any active customers using your platform with PHI today?
- Do you have the ability to redact and detect any PII in your retained data and analytics products?
- How are chatbot transcripts encrypted?
- How do you store customer data and ensure it’s secured and isolated?
- What’s your data and retention policy?
Chatbot Implementation and Consulting
- Describe the implementation process.
- What is the average length of time for a typical implementation/setup?
- What level of expertise and resources are required to manage the software?
- What training programs are provided initially and for the life of the product?
- Does your team provide consultation services to optimize and make recommendations to improve chatbot performance, ROI, and internal efficiency?
- What is the scope of the technical service support model?
- What is the Service Level Agreement of Support for your platform?
- What is your incident response or incident escalation plan?
- What is the preferred method to contact support personnel in the event of an emergency?
- What training materials and user manuals will be provided to train the trainer and trainers?
- How many employees are needed to implement a bot or a skill?
- What is your session definition for a chatbot session billing?
- Is pricing structured at the bot level, session level, or number of users?
- How does your company do licensing of a chatbot platform?
- Is your platform pricing agent based, skill based, transaction based, or user based?
Summary of Chatbot Request for Proposal Templates and Questions
I hope this list of questions helps you as you progress on your journey of implementing a chatbot for your own organization. I know this is a large list, but you might consider doing a Request for Information (RFI) if you didn’t want to go through a formal process to talk through all of these questions. RFIs are generally less informal and take up less time.
However, if you are on a journey to choosing a platform, these platforms can be expensive. Making sure you ask the right questions to some of these vendors will make all the difference. For any demo or proof of concept, your sales team can make things automagically work (we like to call it vaperware). So ensuring that you have a Request for Proposal (a formal one) tied to a contract is important to ensuring that your implementation is a success or you at least have a mitigation plan if things go awry. You’ll at least have something to go back on during the implementation process to review what was promised or stated as being available.