Chatbot and conversational AI ROI calculator

What is chatbot ROI?

The ROI of an AI chatbot, or of conversational AI generally, is the difference between what handling customer questions costs your team today and what it costs once the AI answers them, divided by what the AI costs. Most teams measure it per month, because both salaries and the plan run monthly.

Two numbers decide it: how many questions you get, and what one answer costs in staff time. Everything else, from the number of channels to the quality of the answers, only matters because it moves those two.

A worked example

Say you get 200 customer questions a week. That is about 866 a month. An employee costs around €18.28 an hour including employer costs and spends about 4 minutes per question, so €1.22 per question and €1,055 a month.

On the Heyloha 1,000 plan you pay €339 a month. The calculator then shows a saving of €716 a month, or 68%, around €8,600 a year. Your team gets roughly 10.7 hours a week back.

The break-even point: from 28 questions a week

The smallest plan costs €149 a month. At €1.22 per question it pays for itself from about 122 conversations a month, roughly 28 a week. Below that, the gain is mainly being reachable out of hours, not cost.

Conversational AI ROI across channels

Conversational AI is more than website chat: phone, WhatsApp, email and social messages are answered from the same knowledge base. The logic is the same on every channel, volume against one monthly plan, so the saving grows as you point more channels at it without hiring per channel.

What differs is the cost of not answering. An unanswered chat is a delayed reply; an unanswered call is usually a customer who rings the next company. That is why teams with a busy phone line tend to see the effect first.

Does ROI differ for government, education or enterprise teams?

The formula does not change, the inputs do. Public-sector and education teams usually have a high share of repeat informational questions, about opening hours, deadlines, enrolment or documents, which is exactly the type an assistant resolves on its own. They also have stricter requirements on where data is stored and who can read it, so check EU hosting and a data processing agreement before you compare prices.

Larger organisations tend to have a higher cost per answer and more channels, which increases the saving, but also more systems to connect, which adds setup effort. Start with one channel and the questions you already answer daily, then measure before you widen the scope.

What the calculator leaves out

The calculator uses the cost of every conversation. In practice your team still handles about 20% itself: exceptions, complaints and questions that need a person. Allowing for that, the saving in the example is closer to €505 a month than €716. That is the more honest number to plan with.

On the other hand, the calculator counts no extra revenue. Conversations lost today out of hours or during busy spells do get answered with an AI assistant. For many businesses that outweighs the cost saving, but it varies so much per business that we deliberately leave it out.

ROI beyond cost

Gartner expects AI to resolve around 80% of common customer service issues on its own by 2029, cutting operational costs by 30%. For most businesses the biggest gain is not fewer staff, but not having to hire as volume grows, and no longer missing customers out of hours.

Want to see what customer service costs you today, including hidden costs? Read what customer service costs. For the step-by-step approach: automating customer service.