A client runs a professional services firm with four members of staff. She had been losing leads to voicemail for months when someone suggested she add a chatbot. She went searching, found “virtual assistant” sitting alongside it in the results, and was quoted a different price for each. She wasn’t sure whether the distinction mattered. It does. The two tools suit different jobs, and picking the wrong one costs more than licence fees.
What’s the difference between a chatbot and a virtual assistant?
A chatbot is designed for short, predictable interactions. It works from a scripted flow or a fixed knowledge base, answering the same questions reliably and cheaply. A virtual assistant understands context across several turns of conversation, connects to external systems such as your calendar or CRM, and can complete tasks as well as answer questions: booking an appointment, sending a follow-up, updating a customer record.
In practice, the terms blur considerably in vendor marketing. Many SaaS platforms now bundle both capabilities and market everything as a “virtual assistant.” The distinction worth holding onto is functional: is the tool designed to answer a predictable list of questions, or to take action within a connected set of systems? That question shapes everything that follows.
When a chatbot is the right call
If your inbound queries are short, repetitive, and carry little risk, a chatbot is often the better starting point. UK chatbot deployments suggest 60 to 75 per cent of inbound contact from owner-managed service businesses falls into a handful of repeatable categories: opening hours, service lists, pricing, and basic account queries. When the questions are that predictable, a well-configured chatbot can handle a large share of them cheaply.
Per-interaction costs from UK deployments typically run at £0.40 to £0.55 for a bot, against £4.70 to £9.40 for a human agent once licence, infrastructure, and staff time are factored in. That gap closes fast when scope is wrong, but for genuinely narrow, repetitive work, the economics are hard to argue with.
Well-implemented chatbots can achieve customer satisfaction scores around 4.1 out of 5, with bot-to-human handover satisfaction of around 92.6% when the escalation path is well-designed. Both figures hold when scope is clearly defined and the bot is not overextended into territory it was not built for.
SaaS platforms including Intercom, Zendesk AI, and Tidio can generally be configured by non-technical staff using templates and FAQ uploads. If you run a small team with a manageable volume of predictable queries, starting here is sensible. You can step up later once your needs outgrow it.
When a virtual assistant is worth the extra investment
When the work involves more than one system, a virtual assistant earns its extra cost quickly. Take appointment rescheduling: your customer wants a new slot, which means checking a diary, confirming availability, updating a CRM record, and sending a confirmation. A virtual assistant configured with the right integrations handles the whole sequence. A FAQ-style chatbot stops at the first step and passes the rest back to staff.
NatWest’s Cora virtual assistant handles over 10 million conversations a year across tasks including balance checks, card management, and account changes, with a significant share resolved without human involvement. HSBC’s Amy covers similar transactional ground across digital channels. Both are large-bank examples, but the underlying pattern holds at any scale: wherever a customer’s request is “do something to my booking or account,” a virtual assistant returns more value than a FAQ bot.
Service businesses including clinics, consultancies, and trades firms are increasingly using assistant-style tools as a digital front desk. The tool handles inbound scheduling, follow-up reminders, and initial qualification, all connected across the systems the firm already uses. The staff hours freed up, particularly in administration and first-response handling, tend to be where the return shows up most clearly.
What does it cost to get the choice wrong?
Deploying a chatbot where you needed a virtual assistant produces a recognisable set of problems. Queries the bot cannot resolve pile up as staff escalations. Customers who cannot get a useful answer drop off. Industry contact-centre data links chatbot misuse for complex, variable queries with lower Net Promoter Scores and higher churn in sectors including telecoms and banking. For owner-managed businesses, those consequences are concrete.
The costs tend to arrive in three forms: missed leads when nuanced enquiries bounce; staff time spent fixing errors or following up with customers who dropped off after a poor automated experience; and, in regulated sectors, potential legal exposure. The FCA has fined firms for misleading communications in digital channels, and a chatbot giving unclear or incorrect information about refunds, rights, or financial products can make that risk considerably worse.
UK data protection law adds a further layer. The ICO has been explicit that organisations remain responsible for UK GDPR compliance when using third-party AI tools, including chatbots. If your tool processes personal data without a Data Protection Impact Assessment where one is required, the tool choice becomes a liability rather than an asset.
Going the other direction, deploying a full virtual assistant before your internal processes are consistent enough to automate tends to compound problems rather than remove them. A more capable system replicating messy processes at speed is rarely better than a simpler one.
What to ask before you make the call
The decision between chatbot and virtual assistant starts with one question: what job are you actually hiring this tool to do? If the answer is “answer simple questions so staff can focus elsewhere,” a chatbot is likely sufficient. If the answer is “act like a digital front-desk colleague who can book, reschedule, follow up, and update our systems,” you need a virtual assistant.
Next: what proportion of your inbound is genuinely repetitive? If 60 to 75 per cent of contacts fall into predictable categories and carry low risk, a chatbot can handle a large share safely. If contacts vary and require access to multiple systems or customer context, a virtual assistant is more appropriate.
What personal data will the system see? Map what flows through the tool: names, contact details, payment information, appointment records. Understand where that data is processed, stored, and by whom. Ask vendors specifically about UK GDPR compliance support, sub-processors, and data residency.
How will you handle escalation when something goes wrong? A virtual assistant with a clean handover to a human, with full context passed through, aligns with both ICO guidance and the FCA’s Consumer Duty requirements. A bot that traps customers in loops or forces them to start again is a compliance risk as well as a customer experience problem.
Finally: are your internal processes consistent enough to automate? A virtual assistant adds value when it handles clear, repeatable workflows. If the underlying process is ad hoc, the tool will surface that problem quickly, usually in front of customers.
The technology has moved far enough that the gap between what a chatbot can do and what a virtual assistant can do is real, measurable, and consequential for how your customers experience your business. Getting the choice right at the start is materially cheaper than changing tools six months in. To work through which fits your situation, book a conversation.
