Price an agency chatbot as a managed customer workflow, not as access to a piece of software. A sustainable quote normally separates implementation from recurring operations and makes usage, revisions, support, third-party services, and client responsibilities visible.
There is no responsible universal answer to what every agency should charge. Geography, scope, risk, integration depth, response obligations, and client value differ too much. A pricing system is more useful than a borrowed price list.
Start with your delivery cost floor
Calculate the minimum price that covers the work before adding a margin. Include work that is easy to forget: sales discovery, project management, testing, reporting, revisions, support, payment fees, and the time required when client content is incomplete.
For example, if a project costs your agency 1,000 in its billing currency to deliver and your target gross margin is 50%, the formula produces a 2,000 minimum price. This is a mathematical illustration, not a recommended market rate.
Separate the commercial layers
| Charge | What it covers | Useful boundary |
|---|---|---|
| Discovery or audit | Workflow mapping, source review, risk and feasibility | Defined interviews, pages, and written recommendation |
| Implementation | Training, configuration, branding, installation, and acceptance testing | Named workflow, environments, revisions, and launch criteria |
| Monthly management | Monitoring, approved updates, reporting, and optimization | Review cadence, included changes, and response time |
| Usage or platform | AI messages, voice, phone, dispatch, or connector consumption | Allowance, measurement source, and overage handling |
| Change request | New workflow, integration, language, site, or material scope expansion | Written estimate and approval before work begins |
Three packages that are meaningfully different
Launch
One assistant, one website, approved knowledge sources, branded widget, one conversion path, installation, and acceptance testing. This is for a client that needs a focused first deployment.
Managed
Everything in Launch plus scheduled conversation review, content refreshes, lead or booking workflow monitoring, analytics reporting, dispatch verification, and an agreed change allowance.
Operations
Multiple customer journeys or channels, deeper connector work, voice configuration, documented escalation, higher support expectations, and governance reviews. Price this only after discovery because risk and effort vary substantially.
Use Chirps capabilities where they solve the agreed workflow: website and document training for answers, branded embeds for deployment, lead forms and bookings for conversion, dispatch alerts for follow-up, conversation analytics for review, voice for phone or browser experiences, and human takeover for requests that need judgment.
Price by scope and responsibility, not feature count alone
- Knowledge scope: number and quality of websites, documents, languages, and update owners.
- Action scope: forms, bookings, notifications, connectors, and authenticated operations.
- Risk: regulated topics, sensitive data, financial consequences, and required human approval.
- Service level: support hours, incident response, reporting frequency, and launch deadline.
- Change rate: how often products, policies, staff, pricing, or availability change.
- Deployment surface: number of assistants, client sites, domains, channels, and environments.
Protect the margin with explicit assumptions
A profitable retainer can become unprofitable when every request is treated as included. State how many review meetings, content changes, new intents, languages, and installation environments are covered. Define what pauses the timeline, including missing access or delayed approvals.
For ongoing scope, use the AI chatbot maintenance retainer guide. For implementation inputs, use the client onboarding checklist.
Make value claims verifiable
A proposal can describe the desired outcome, but it should not present forecasts as guaranteed results. The FTC's advertising guidance emphasizes that objective claims need a reasonable basis before publication. Use the client's baseline and report actual outcomes after launch.
| Weak claim | Better commercial language |
|---|---|
| This bot will double your leads | The pilot will measure qualified leads completed through the assistant against the agreed baseline |
| It answers every question | It answers from approved sources and routes unsupported or sensitive requests according to the escalation policy |
| The system replaces support staff | The workflow handles defined repetitive requests while staff retain ownership of exceptions and human conversations |
| Unlimited changes | The retainer includes the listed review cadence and change allowance; additional scope requires approval |
Present the quote so clients can compare it
- State the business problem and the workflow being delivered.
- List implementation deliverables and acceptance tests.
- Separate one-time, recurring, usage-based, and optional charges.
- Explain client dependencies, exclusions, and change control.
- Define the reporting cadence and who owns each metric.
- Add commercial terms, data responsibilities, and a clear next step.
The AI chatbot proposal template turns these decisions into a client-facing structure. If you are still defining your service, start with how to start an AI chatbot agency.
Related agency resources
- AI chat for agencies: one widget, multiple client sites
- Best businesses to sell AI chatbots to
- Current Chirps plans and included usage
- Chirps implementation documentation
This article is general commercial information, not accounting, tax, or legal advice. Validate pricing, contracts, consumer obligations, and taxes with qualified advisers.