The Customer Fit RFP Checklist
Eight questions that should be in every contact centre technology RFP.
But almost never are. Standard RFP questions get a yes from every vendor. None of these can be answered with a yes.
How to use this
Does the platform support omnichannel? Does it have AI routing? Does it integrate with Salesforce? Those questions were drawn around what vendors can do, not around what your organisation needs.
The questions below require vendors to describe their approach, demonstrate their capability in your specific context, and be honest about what they can and cannot do. They also require your organisation to have done work first. You cannot ask meaningful questions about data foundations without honestly assessing your data. You cannot ask about demographic equity without knowing your customer base.
That foundational work is not the vendor’s job. It’s yours. And it has to happen before the RFP goes out.
Before any AI capability can work as demonstrated, your data needs to be ready for it. Most vendors address this after go-live. You need to ask about it before you sign.
- Does the vendor conduct a data readiness assessment before implementation begins?
- What minimum data quality standards do they require before deployment?
- Who is responsible for data cleansing — your team, the vendor, or a third party?
- How does the platform handle duplicate customer records across systems?
- What happens to AI performance when it encounters conflicting data from multiple systems?
- How is data quality monitored on an ongoing basis post go-live?
Describe your data readiness assessment process. What baseline data quality do you require before deployment — and what is your approach when that baseline isn’t met?
A sentiment score tells you what a customer said in a single interaction. It doesn’t tell you how they feel across their entire journey. Ask for the difference.
- How does the platform measure customer emotion across the full journey, not just within a single interaction?
- Can the platform identify customers who are politely persistent across multiple contacts about the same unresolved issue?
- How is emotional data surfaced for frontline staff in real time — not in a report, in the moment?
- Can non-technical operations managers access and interpret emotional trend data without a data analyst?
- How does the platform distinguish between a customer who was frustrated but left satisfied versus one who appeared satisfied but returned with the same issue?
- How is emotion measured across channels — voice, chat, messaging, email — consistently?
How does your platform measure customer emotion across the full journey, not just within a single interaction? How is that measurement made actionable for frontline staff without requiring data analysis expertise?
Journey analytics is a standard feature claim. The question is who can actually use it — and without booking a data analyst.
- Can a team leader access customer journey data in real time without requesting a report?
- Can an operations manager see where customers are dropping out of self-service flows without a BI tool?
- How does the platform visualise the customer journey across channels in a single view?
- How quickly can non-technical staff identify the cause of a spike in contacts?
- Can journey data be filtered by customer segment, contact reason, or outcome without technical configuration?
- What does the default journey dashboard look like for a team leader — can you show us, not describe it?
Show us how a non-technical operations manager or team leader would access and interpret customer journey data in your platform. Walk us through that experience without a data analyst in the room.
AHT, ASA and abandonment rates have governed contact centres for thirty years. In an AI-enabled environment they are increasingly insufficient. Ask what replaces them.
- Does the platform measure whether a customer’s issue was actually resolved — not just closed?
- Can the platform identify customers who called back within X days about the same issue?
- How is customer effort measured — not just satisfaction?
- How does the platform measure journey completion versus journey abandonment?
- Can the platform surface the relationship between operational metrics (AHT, ASA) and customer outcome metrics (resolution, retention) in a single view?
- What metrics does the vendor recommend for measuring AI performance specifically — not just overall contact centre performance?
- How are quality metrics defined and measured for AI-handled interactions versus human-handled interactions?
What metrics does your platform recommend for measuring genuine customer outcomes — resolution quality, customer effort, journey completion — rather than operational efficiency alone? How are those surfaced alongside traditional KPIs?
RAG, LLMs and vector databases are not plug-and-play. Your environment has to be ready for them. Most RFPs don’t ask whether it is.
- What specific technical prerequisites does the vendor’s AI architecture require from our environment?
- What format does our knowledge base need to be in for the LLM to retrieve from it effectively?
- Who builds and maintains the vector database — our team, the vendor, or a third party?
- What happens to AI performance when the knowledge base is incomplete, outdated, or poorly structured?
- How does the platform handle knowledge base updates — how quickly do changes propagate to the AI?
- What is the vendor’s assessment process for determining whether our environment is genuinely AI-ready?
- What are the ongoing maintenance requirements for the AI architecture post go-live?
What specific technical prerequisites does your AI architecture require from our environment? What does your implementation team do to assess and close those gaps — and what remains our responsibility?
A handoff model is not collaboration. Ask how the platform supports AI and humans working together — not in sequence.
- How does the AI actively support a human agent during a live interaction — not just before or after?
- Can the AI surface relevant knowledge, flag risk signals and suggest next steps in real time while the human maintains control of the conversation?
- What is the mechanism for human agents to flag when the AI’s suggestion was wrong or inappropriate?
- How does agent feedback improve AI performance over time — what is the feedback loop?
- How does the platform handle the transition from AI to human without the customer having to repeat themselves?
- What context does the human agent receive at the point of escalation — specifically what information transfers and in what form?
- Can a human agent override an AI decision mid-interaction — and what happens to that override data?
How does your platform support active AI assistance during a human interaction — not just at handoff? What is the mechanism for human agents to provide feedback that improves AI performance over time?
Every platform has an API. The question is at what points in the interaction lifecycle external systems can actually hook in — and how granular that is.
- At what specific points in the interaction lifecycle are webhook events available?
- Can a webhook fire when customer sentiment drops below a defined threshold mid-interaction?
- Can automation inject context into a live voice interaction from an external system in real time?
- Can a third-party AI model be invoked at a specific point in the conversation flow?
- Is the API event model consistent across channels — voice, chat, messaging, email?
- What is the latency of webhook events — are they real-time or near-real-time?
- What authentication and rate-limiting applies to the API — and how does this affect high-volume automation?
- Can the API read interaction state mid-conversation — not just at start and end?
Map the specific webhook events and API endpoints available at each stage of the interaction lifecycle across all channels. At which points can external systems read interaction state, inject data, or trigger automation in real time?
Your customer base is not homogeneous. Your 55-year-old loyal customer and your 28-year-old digital native have different needs. Ask whether the platform serves both.
- Can the platform report CSAT, resolution rates and channel preference broken down by age group or other demographic segments?
- How does the platform accommodate customers who prefer voice over digital self-service?
- How does the voice AI handle variation in accent, speech pattern and pace across different demographic groups?
- How are self-service flows designed to accommodate customers with lower digital fluency?
- Can the platform identify demographic segments that are systematically underserved by AI interactions?
- How does the platform measure equity of experience across demographic groups — not just overall CSAT?
- What accessibility features are native to the platform for customers with disabilities?
How does your platform measure and report experience quality by customer demographic segment? Can we access resolution rates, satisfaction scores and channel preference data broken down by age group or other demographic markers?
Take it into the room
Get the checklist as a working document.
The A4 worksheet is built to be worked through with your own team before an RFP is drafted — one question per page, tick boxes against each, and room for notes. Tell us where to send it and it’s with you in a minute.
Feature rich doesn’t mean customer fit
Make the gap between a promise and a capability visible before the contract is signed.
If you’d like help mapping your data boundaries, understanding your customer base and building an RFP that finds genuine fit rather than impressive promises — that is exactly what Canzuki does. We don’t sell software and we don’t run vendor demos. We help organisations do the unglamorous, high-impact foundational work required to make AI actually work.