How much does an AI chatbot cost? A budget for a small online store
Build a realistic chatbot budget from subscription fees, actual credit use, setup and human oversight. Three illustrative workloads show how to size a plan.
An AI chatbot budget has three parts: the software subscription, usage beyond the included allowance, and the time needed to prepare and maintain the service. Breezaro's subscription starts at $8.99 a month; a store that needs product search and human takeover should budget from Pro at $28.99 a month. The right plan depends on features and workload, not traffic alone.
The examples below are illustrative calculations, not measured customer results or forecasts of your store's usage. Prices were checked on 21 October 2026 against Breezaro's pricing. All amounts are in USD; taxes, payment-provider currency conversion and any separately billed external services are excluded.
First, understand what the provider counts
A message, a conversation, a credit and a resolved request describe different things.
| Unit | What to establish before comparing prices |
|---|---|
| Message or reply | Does the allowance count visitor messages, AI replies, or both? |
| Conversation | When does it start, end or count again if a visitor returns? |
| Credit | How does the selected model, a search or an executed action affect usage? |
| Resolution | What qualifies as resolved, and what happens if the customer asks again? |
A visitor might exchange six messages while choosing a product, then ask an operator to check a special delivery request. That is one customer need, several messages and potentially both AI work and human work. A headline such as “1,000 included” is meaningful only when its unit is known.
In Breezaro, ordinary replies on a standard model cost one credit. A reply involving actions can cost more, and premium models have higher rates. Replies, proactive AI communication and executed actions use the shared account balance. Inspect the model's displayed rate and the credit breakdown in conversation history. The credit guide explains the mechanism.
Select the required features before the allowance
Here are the monthly plans used in this example:
| Breezaro plan | Monthly subscription | Included monthly credits | Relevant planning distinction |
|---|---|---|---|
| Basic | $8.99 | 1,000 | One web chatbot for a small knowledge base |
| Standard | $18.99 | 2,500 | Adds WhatsApp, Messenger and Instagram DM |
| Pro | $28.99 | 4,000 | Adds product catalog, Take Control and three Custom Actions |
| Business | $78.99 | 10,000 | Higher capacity; includes removal of the Breezaro label |
Credits are shared across the account's bots. Knowledge-base capacity is a separate limit on processed content. A store with a large catalog may need more capacity even when few customers open the chat. Conversely, a small FAQ can receive many questions.
If your requirement is “an operator must take over a conversation,” Basic is not the correct budget just because the expected usage fits 1,000 credits. Price the workflow you actually need.
Three illustrative monthly workloads
For this calculation only, assume standard-model replies cost one credit and the extra action calls below also each cost one credit. We choose the number of replies and calls for illustration; they are not observed averages. The reserve is a suggested planning margin, not a product requirement.
| Assumption | Small FAQ bot | Store with product help | Busy month with product help |
|---|---|---|---|
| Conversations | 200 | 500 | 1,500 |
| AI replies per conversation | 2 | 3 | 4 |
| Reply credits | 400 | 1,500 | 6,000 |
| Additional action-call credits | 0 | 500 | 1,500 |
| Proactive AI credits | 0 | 0 | 500 |
| Estimated total | 400 | 2,000 | 8,000 |
| Total with 20% reserve | 480 | 2,400 | 9,600 |
| Plan fitting these assumptions and features | Basic | Pro | Business |
| Monthly subscription | $8.99 | $28.99 | $78.99 |
The middle scenario uses Pro because product help requires the catalog, despite its allowance fitting below Standard's credit ceiling. The busy scenario leaves little room if conversations become longer or the model changes. Replace every assumption with your pilot measurements before committing to a budget.
For a mixed workload, calculate each model and action separately, then add the totals. Do not multiply a model's rate by a message count and silently assume searches, actions and proactive messages are free.
Add the work around the chatbot
Someone must check opening hours, delivery information and return instructions, remove contradictory sources, test missing answers, and review conversations after launch. A custom order lookup also needs someone to configure and maintain the connection and its access checks.
For an illustrative internal budget, suppose initial preparation takes six hours and monthly review takes four hours. At a chosen internal rate of $25 an hour:
- One-time preparation: 6 × $25 = $150.
- Monthly review: 4 × $25 = $100.
- Pro subscription plus monthly review: $28.99 + $100 = $128.99 per month.
- First month including preparation: $128.99 + $150 = $278.99.
This is an economic cost estimate. If the owner does the work, it may not create a separate invoice. It still uses time. These hours exclude handling escalated customer requests, development of a custom integration and any existing support-tool costs; add those where relevant.
Plan for peaks and exhausted credits
Estimate an ordinary month and a promotion month separately. Watch the balance during a campaign rather than waiting until the allowance is exhausted. Monthly plan credits reset; unused plan credits do not accumulate. Purchased credits are a separate balance used after the monthly allowance.
The current public offer includes a one-time 10,000-credit pack for $78.99. Compare the actual purchase required with a higher plan's cost and features. Dividing a pack price into a theoretical price per credit does not mean you can buy exactly the small quantity in your spreadsheet. Check the current pack terms before purchase.
Turn the estimate into your own budget
Start with a representative set of questions, including multi-step product selection and a request for a person. Record replies, action calls, total credits and the time spent reviewing answers. Keep failed answers in the sample: they consume effort even when the result is not useful.
Then estimate the month using that workload, account for busy periods and choose the plan with the necessary features. Review the estimate after the first complete billing period. Track useful outcomes alongside spend using the chatbot measurement guide.
You can use Breezaro's 30-day trial, with a 200-credit allowance and no card requirement, for a focused pilot. Its allowance is a testing budget, not a promise that it covers a full month of your store's traffic.