Chatbot ROI Calculator

No black box. Five inputs, four lines of arithmetic. You can reproduce every number below in a spreadsheet, which is exactly the point: a figure you can defend in a budget meeting is worth more than a bigger figure you cannot.

Two derived figures sit alongside them. Plan cost per resolved chat divides your monthly plan price by the conversations the bot handled, which is the cleanest way to compare a flat-rate platform against a per-resolution one. Chats to break even is the plan cost divided by the labour value of one conversation, which is the volume you have to clear before the tool has paid for itself.

The output is only as good as the four numbers you feed it, and two of them are routinely wrong in vendor-supplied calculators.

Resolution rate

We deliberately do not preload an industry benchmark here, because any single number would be misleading. What a bot resolves depends on your content, not on the model behind it. A support site with thorough, current documentation gives a retrieval-based bot a great deal to work with; a five-page marketing site does not. If you have never run one, pick a number you would be comfortable defending, run a trial, then read the real rate off your dashboard and come back.

Handling time

Use the full time a conversation consumes, including after-conversation work: tagging, notes, internal questions, the follow-up email. Teams usually measure only the live portion, which understates what deflection actually gives back.

Fully loaded agent cost

Not the wage. Salary plus employer taxes, benefits, tooling seats, management time and training, divided by hours genuinely worked. This is the number your finance team already uses, and it is higher than the hourly rate on the contract.

Plan cost

Use the price you would actually pay, including per-seat charges and per-resolution fees where the vendor has them. Matram is flat, at $29, $69 or $199 a month with unlimited seats, so the preset buttons are the true figure. For competitors, our chatbot pricing guide lists prices read directly off each vendor's own page.

The calculator counts one thing: support labour you no longer spend. It ignores several effects that are real but harder to price without inventing numbers, so treat the result as a floor rather than a forecast.

• Out-of-hours coverage. A bot answers at 2am. Counting that properly means pricing the conversations you currently lose rather than serve, which is a number only you have.
• Leads captured in chat. Qualified leads have obvious value, but the value varies so much by business that any figure we put here would be decoration. Your close rate and average deal size are the honest inputs.
• Faster first response. Instant answers reduce abandonment. Genuine, but not something we can quantify for you without guessing.
• Deflection on the remaining conversations. Even chats that escalate arrive with context already gathered, which shortens them. The calculator credits nothing for that.

Equally, one cost is omitted: your time. Someone has to curate content, review answers the bot got wrong, and keep sources current. On a platform that crawls your existing site that is modest, but it is not zero. If you are weighing a custom build instead, the chatbot development cost breakdown covers where that time actually goes.

1. Count your monthly conversations. Pull the total number of inbound support conversations for a typical month from your helpdesk or inbox. Include chats, emails and tickets, since a chatbot can absorb repetitive questions arriving through any of them.
2. Estimate the share the bot can resolve. Judge how many of those conversations are answerable from content you already have published or could publish. If you have not run a bot before, start conservative and revise the number once you have real resolution data.
3. Measure your average handling time. Take the average time an agent spends per conversation, including after-conversation work such as tagging, notes and follow-up, not only the minutes spent typing.
4. Work out the fully loaded agent cost. Add salary, employer taxes, benefits, software seats and overhead, then divide by hours actually worked in the year. Use that hourly figure rather than the raw wage.
5. Enter the chatbot plan cost and read the result. Enter the monthly plan price you are comparing against. The calculator returns net monthly savings, annual savings, plan cost per resolved conversation and the break-even conversation count.

Then sanity-check it. If the calculator says you save more than the entire cost of your support team, the resolution rate is too optimistic. If net savings are negative, the plan is priced above the labour it displaces at your volume, which is a genuine result, and usually means you should start on a lower tier rather than not at all.

• AI chatbot pricing guide: what the major platforms actually charge, verified against their own pages.
• Chatbot development cost: build versus buy, including the maintenance nobody quotes for.
• Intercom pricing explained: how per-seat plus per-resolution pricing compounds.
• Matram features: what the bot does once you have decided the numbers work.