Chatbot ROI calculator for website support.

Estimate how many support hours an AI chatbot could free up each month, what that time is worth, and what’s left after software and setup costs. Every number is an assumption you can change.

Made by the team behind Conecto, which sells website chat software. How we keep it useful

Your support volume
per month

Chats, emails and calls that could come through website chat.

minutes

Time a person spends per conversation, including follow-up work.

Share fully resolved by AI

Conversations that end without a person. Compare a cautious and an optimistic case with your selected one.

%

%

%

Costs

Changes the symbol only. Nothing is converted.

$ per hour

Loaded cost: pay plus taxes, benefits and tools.

hours a month

Reading transcripts and updating knowledge.

$ per month

Subscription plus any per-reply AI charges.

$

Writing knowledge, configuration, testing.

Prefilled values are illustrative assumptions. They are not industry benchmarks or Conecto performance figures.

Net hours released per month

13 h

25 h gross, before review time

Monthly value after software

$266

$416 of capacity before software

Modeled first-year value

$1,692

After the one-time setup cost

Setup cost recovered in

5.7 months

At the selected resolution assumption

These estimates describe potential capacity, not guaranteed cash savings.

Compare resolution assumptions

ResultLower (15%)Selected (25%)Higher (35%)
Gross hours released per month 15 h 25 h 35 h
Net capacity change per month 3 h 13 h 23 h
Modeled monthly capacity value $96 $416 $736
Monthly value after software costs −$54 $266 $586
Modeled first-year value −$2,148 $1,692 $5,532
Setup cost recovered in Not recovered 5.7 months 2.6 months

Negative values mean the assumptions don’t cover review time or costs. The calculator never rounds them away.

Want to test these assumptions with real conversations?

How the calculation works

The model is deliberately simple, so every step can be checked. It assumes that each conversation the chatbot fully resolves removes the average handling time you entered, and nothing else.

Gross hours released = monthly conversations × resolution % ÷ 100 × average handling minutes ÷ 60

Net monthly capacity change = gross hours released − monthly review and maintenance hours

Modeled monthly capacity value = net monthly capacity change × hourly support cost

Monthly value after software costs = modeled monthly capacity value − monthly software and usage cost

Modeled first-year value = 12 × monthly value after software costs − one-time setup cost

Setup cost recovered in = one-time setup cost ÷ monthly value after software costs, shown only when that monthly value is positive

Worked example

With the prefilled assumptions: 1,000 conversations a month, 6 minutes each, 25% fully resolved, $32 an hour, 12 review hours, $150 a month in software and $1,500 to set up.

  • 1,000 × 25% × 6 ÷ 60 = 25 h gross hours released.
  • 25 h − 12 review hours = 13 h net.
  • 13 h × $32 = $416 of modeled capacity a month.
  • $416 − $150 software = $266 a month.
  • 12 × $266 − $1,500 setup = $1,692 in the first year, with setup recovered in 5.7 months.

At the lower assumption of 15%, the same inputs give 3 h net and −$54 a month after software: a loss. That is the point of showing three cases. A small change in the resolution rate can decide whether chat pays for itself.

What the model leaves out

  • Partly handled conversations. A chatbot that collects details before handing over can shorten a person’s work. The model counts only conversations that end without a person.
  • Longer or shorter escalations. A good handoff with context can speed up the reply. A poor one can make it slower.
  • Changes in demand. Easier chat can bring more conversations, including ones that would never have been emails.
  • Revenue and satisfaction. Answering a pre-sales question at 10 p.m. may be worth more than the time it saves. The model doesn’t try to price that.
  • Staffing reality. Hours released are capacity. They only become savings if schedules, hiring or overtime actually change.

What a chatbot costs

The calculator uses the costs you enter. It doesn’t guess them, because prices change and depend on volume. The usual parts are:

Software subscription
Charged per seat, per workspace or per plan tier. Check whether the plan you’re pricing includes AI answers.
Usage-based AI charges
Many tools bill AI replies or resolutions separately, through credits or per-conversation fees. Add an estimate to the monthly software cost.
Setup
Writing or cleaning up help content, configuring the widget and handoff rules, and testing. Usually the largest one-time cost for a small team.
Ongoing upkeep
Reviewing answers and updating knowledge when policies change. Enter this as review and maintenance hours, because it recurs every month.
Integrations
Order lookups, bookings or account changes need access to your systems. Custom integration work can cost more than the chatbot itself.

If you’re pricing a custom-built chatbot rather than a chat product, this calculator won’t tell you the build cost. Get quotes, then enter them as setup and monthly costs. For Conecto’s current plans, see Conecto’s pricing page.

How to choose your assumptions

Monthly conversations
Use your help desk, inbox or chat reports. Count the conversations that could reasonably come through website chat.
Average handling time
Include the work after the reply: notes, follow-ups and lookups.
Resolution rate
List your most common questions and check how many have a clear written answer. That gives an upper bound. A pilot on real conversations gives a measured rate. Start with the lower figure.
Hourly support cost
Use the loaded cost: pay plus taxes, benefits, software and overhead.
Review and maintenance
Budget time every month to read transcripts and fix gaps in the knowledge. It is easy to forget and it never goes away.

Formulas and guidance last reviewed on September 23, 2026. Your inputs stay in your browser: nothing is stored or sent.