How to Package and Price AI Automation Services (Freelancer and Agency Guide)

Most freelancers and small agencies who build chatbots and automations make the same pricing mistake: they price the build. They count hours, multiply by a rate and send a quote. The client compares it with a cheaper quote, and the work becomes a race to the bottom.

Clients do not buy a chatbot. They buy faster replies, more bookings, fewer missed leads and fewer hours spent copying data between apps. When you package and price around those outcomes, the conversation changes from “why so expensive?” to “when can we start?”. This guide shows how to define packages, choose a pricing model, run the numbers and protect your scope.

1. Sell the outcome, not the build

Before you write a price, write the outcome in the client's words. “A Messenger chatbot with 12 flows” is a build. “Every inquiry answered within a minute, day or night, and qualified buyers handed to your sales team with their details filled in” is an outcome. The second one can be measured, which means it can be valued.

Three questions get you there in a discovery call:

1. What is the problem costing you now? Missed inquiries, staff hours, slow follow-up, lost bookings.
2. What would “fixed” look like in 90 days? A number they would celebrate.
3. What happens if nothing changes? This is the cost of waiting, and it frames your price.

2. Build a package ladder

Three packages make decisions easier than one custom quote. The client compares your options with each other instead of comparing you with other freelancers. Each tier should solve a bigger problem, not just add more features.

Design the middle tier as the one you expect most clients to choose. The Starter tier makes it look complete, and the Scale tier makes it look reasonably priced.

3. Choose a pricing model

Most automation work fits a fixed project price informed by value: you calculate the value first, then set a fixed package price well below it, and add a retainer for ongoing work.

4. Run the numbers

Use the client's own numbers. Every figure below is an illustrative example for a business that answers inquiries by hand. Replace them with what you learn in discovery.

Annual value = (hours saved per month x hourly staff cost x 12)
             + (extra sales per month x profit per sale x 12)

Price range  = 10% to 30% of annual value
               (lower for new clients and unproven numbers,
                higher when the value is clear and measured)

Then sanity-check the bottom of the range against your costs: your time, platform subscriptions during the build, and the testing you need to do properly. If the low end does not cover them, the client is too small for this package. Offer the Starter tier instead, or walk away.

5. Add a monthly retainer

An AI agent is never finished. Prices change, promos end, new questions appear, and models are updated. A retainer pays you to keep it accurate, and it gives the client a reason to keep you instead of letting the system decay.

• What it includes: a monthly review of real conversations, instruction and knowledge updates, a metrics report, and a fixed number of small changes.
• What it excludes: new channels, new integrations and new flows. Those are new projects.
• What it reports: the metrics from the 5 chatbot ROI metrics, so the value of the retainer is visible every month.

6. Protect your scope

Scope creep is where automation projects lose their margin. Put these items in writing before you start:

SCOPE CHECKLIST
[ ] Channels included (e.g. Messenger, Instagram, WhatsApp)
[ ] Number of flows or agent use cases
[ ] Integrations included (CRM, POS, calendar, sheets)
[ ] Who writes the content: you or the client
[ ] Who provides product lists, prices and policies, and by when
[ ] Number of revision rounds
[ ] Testing plan and acceptance criteria
[ ] Launch date and what counts as “launched”
[ ] Platform subscriptions: who pays, whose account
[ ] Support period after launch, then the retainer terms

Build on the client's own accounts whenever possible. They own their data and subscriptions, you avoid paying their usage, and the handover is clean.

An honest limitation: value-based pricing depends on numbers the client gives you, and those numbers are often estimates. Use conservative inputs, say in the proposal which figures came from the client, and agree on how you will measure the result after launch.

Frequently asked questions

Should I show prices on my website?

Showing a starting price for each package filters out clients who cannot afford you and saves discovery calls. Keep the final price tied to the proposal.

What if a client only wants the cheapest option?

Offer the Starter package with a clear scope. If they want Growth results at Starter price, explain what each tier solves and let them choose. Discounting the same scope teaches clients to negotiate every time.

Who pays for the chatbot platform and AI usage?

The client, on their own account. Include the expected monthly platform and AI costs in your proposal so there are no surprises.

How do I price my first projects without case studies?

Use the lower end of the value range, define the scope tightly, and agree on the metrics you will report. Your first projects become your case studies.

References

1. The 5 Chatbot ROI Metrics That Matter, Dijae Dadula, October 2026.
2. Botcake AI + Pancake Setup Guide, Dijae Dadula, October 2026.

Keep reading

• How Filipino Freelancers Can Get AI and Automation Clients Abroad
• Turn Your Skills Into AI Income
• Don't Automate Chaos
• See all posts on the blog