AI Automation Agency Pricing Rules

Follow me on LinkedIn - AI, GA4, BigQuery

The following are the 14 Rules to pricing automation that I learned from those who came before me (the agency owners making real money from AI automation for years):

Not all of these rules may resonate with you. Take what resonates and leave the rest for others.

Rule 1. Niche down early.

Pick your niche (real estate, dental clinics, hotels, restaurants, etc.) and then stick to it.

It's hard to scale if you keep jumping from niche to niche, as you need deep niche expertise to build automations that solve real-life problems at scale.

Deep niche expertise lets you understand actual pain points, speak the client's language fluently, and build reusable systems.

For example, A real estate voice AI agent has very different qualification flows than a dental clinic booking agent.

Without niche expertise, you're constantly learning on the job, which kills margins and extends delivery timelines.

Niching also dramatically accelerates your sales cycle.

When you can reference three other estate agents you've helped with the same problem, objections dissolve.

You become the obvious specialist rather than a generic "AI automation person."

Additionally, your marketing becomes sharper because you know exactly where your prospects congregate, what language resonates with them, and what specific outcomes matter in their world.

Rule 2. Focus on building a SaaS product instead of a service.

Instead of building custom AI automation systems from scratch for each new client, build one system until it's bulletproof, then sell it repeatedly with minor tweaks.

Same architecture, same playbook, different clients with nearly identical needs.

This will give you more time to sell than to build from scratch for each new client. What that means is not chasing one-off projects.

You can scale a process, funnel, or system only if you are confident it will work 8 out of 10 times. You gain such confidence by building one system that is bulletproof.

Start with custom builds in your niche to learn what actually matters, then extract the common architecture into your productised offering.

My experience with Retell multi-prompt agents versus conversational flow agents is a good example.

I identified that conversation flow agents have too many failure modes, including JSON import failures, 20-node performance limits, and stuck-agent issues.

That learning came from building, not theorising. Once you've solved a problem multiple times, you've earned the right to productise.

Rule 3. Charge for outcomes and not for time, effort or features.

Don’t price your automations based on time or effort spent building the workflow or the features built. Charge for outcomes.

This aligns with well-established value-based pricing principles. Businesses prefer to pay for results, not your hours or the complexity of what you built.

Time-based pricing caps your income at the number of hours you can work. Outcome-based pricing ties your revenue to the value you create, which has no theoretical ceiling.

Features are particularly dangerous to sell because clients may compare your feature list to DIY tools like Zapier or Make and convince themselves they can do it themselves.

Outcomes are harder to commoditise.

Nobody compares "47 qualified leads per month" against a software subscription because they understand the gap between having tools and getting results.

Rule 4. The outcomes you sell should never be about saving time or increasing efficiency.

The outcomes should be either about reducing costs or increasing revenue.

The outcomes should always be tied to monetary value because this is what businesses really understand, and this will help you show clear ROI in monetary terms.

Time savings and efficiency gains sound compelling but create negotiation problems.

For example, “saving 10 hours per week" sounds good until the client asks what those 10 hours are worth.

Then you're negotiating the hourly value of their staff, putting you in a weak position and inviting lowball calculations.

"Reducing cost by 3,000 per month" or "generating 15 additional qualified leads per month" is concrete and inarguable.

Businesses make decisions based on money in and money out.

Time and efficiency are abstractions that require mental translation, so do the translation yourself and present the monetary outcome directly.

This also makes your proposals easier to approve internally because the ROI is obvious to anyone reviewing the expenditure.

When you can clearly explain the ROI in monetary terms, you can easily overcome the price objection.

Calculate and demonstrate both the short-term and the long-term ROI, as the ROI often compounds over time.

Track patterns in client objections so you can address common concerns proactively in future proposals.

Rule 5. Consider delivering at least 100% ROI.

100% ROI means if your client spends X, they earn or save 2X in return.

The appropriate ROI threshold depends on risk distribution.

For one-off projects where the client assumes all future risk and maintenance burden, higher ROI multiples, such as 5x or 10x, make sense because you're not sharing in ongoing success.

For monthly retainers where you're continuously delivering value and bearing operational risk, 100% ROI is fair; they're paying monthly and getting monthly returns.

Per-lead or per-outcome pricing can justify even tighter ROI because you're sharing risk directly. If you only get paid when they get results, tighter margins are warranted.

Rule 6. Sell outcomes that a business needs all the time and can not live without.

Do you want to run a project shop or run a sustainable automation business?

Selling outcomes that a business needs all the time and cannot live without (such as 24/7 customer support, lead generation, cost reduction, etc.) separates sustainable automation businesses from project shops.

When you sell such outcomes, you can justify monthly retainers, and your clients will never cancel your service (unless you mess up big time).

For example, if you built a lead generation automation workflow, you can charge per lead per month forever.

If your client can pause your service for three months without noticing a meaningful impact, you're selling a nice-to-have.

If pausing means leads stop coming in, customers can't get support, or appointments stop getting booked, you're embedded in their operations.

Essential services survive budget cuts. Nice-to-haves get eliminated first.

This also changes the sales conversation entirely; you're not asking them to try something new, you're offering to handle something they already must do, just better and cheaper.

Rule 7. Always show working demos when submitting a proposal.

Case studies, wireframes, and workflow diagrams require clients to imagine what the solution will do. Most clients are brand new to AI automation and cannot make that imaginative leap. A working demo removes the gap entirely.

This ties back to my productised approach.

If you've built a bulletproof system, spinning up a demo with their business name and sample data takes minimal effort but creates a massive impact.

They experience the outcome rather than imagining it.

Avoid emailing the proposal. Get on a call, present your proposal and close the client right then and there.

Proposals sent via email get shopped around, compared against competitors, delayed for internal review, and forgotten.

A live presentation with an immediate close is the most reliable path to high-value sales. You control the narrative, handle objections in real time, and create urgency.

Rule 8. Use the monthly retainer pricing model from day one.

The setup fee plus the ongoing retainer creates predictable revenue that compounds with every new client. This model also aligns incentives correctly; you succeed when the client succeeds over time, not just at launch.

This monthly retainer includes fees for ongoing optimisation, monitoring and infrastructure maintenance.

This model works particularly well when you've already solved the delivery problem through productisation.

If you're selling essentially the same system with minor tweaks, you can price setup plus ongoing from day one because you already know your delivery costs and timeline.

Regarding the monthly retainer its up to you whether to charge a fixed fee or hourly for ongoing optimisation, new feature requests, revisions, etc. Do what works best for you.

In reality, your monthly fees will never be fixed because you would be charging for usage (API calls) and your own fees.

Don’t take on projects which do not provide scope for ongoing optimisation/maintenance.

One-off builds are time sinks that don't compound into recurring revenue.

Every hour spent on a one-off project is an hour not spent acquiring or serving retainer clients.

Rule 9. You own the infrastructure and all the associated headaches from day one.

The build and transfer consulting model does not generate recurring revenue.

This is the core problem with how most automation consultants structure their business. They treat each project as a deliverable to hand over rather than an asset to operate.

You spend weeks/months building a system. You document it thoroughly because the client "owns" it.

You hand over credentials, workflow exports, prompt libraries, everything.

The client pays your project fee, thanks you, and now has complete independence from you.

Two weeks later, they hired a cheaper freelancer to maintain it. Or their internal team takes over. Or they share your architecture with another vendor who quotes lower for the next project.

Your intellectual property, your hard-won solutions to edge cases, your prompt engineering, all of it walks out the door with the deliverable.

You are not selling the infrastructure (workflows, agent platforms, API access), you are renting it out. What you are selling is the outcomes. However, the client owns all their data.

If your client could manage their infrastructure, they won’t need you, and you don’t want them.

You don’t want DIY clients.

You want clients who want someone to take charge of the complete automation while they do what they do best.

This is a client qualification principle that protects both your margins and your sanity.

Rule 10 - Avoid the DIY clients like the plague.

A typical DIY Client profile:

They ask for workflow exports.
They want admin access to your n8n instance.
They request detailed documentation of every node and every prompt.
They ask questions like "can you teach me how this works" or "can you show me how to modify this myself?"
They want to sit in on technical calls to "learn."

What they're really saying is they view you as a temporary resource, not a permanent partner. They're paying you to accelerate their own learning curve. Once they feel confident enough, they leave.

This client type will drain your resources while convincing you they're a good customer.

They consume disproportionate time.

Every deliverable requires extensive explanation. Every call runs long because they want to understand the technical details.

They ask follow-up questions via email, via WhatsApp, via voice notes. They want screen recordings of how you built things.

They request documentation that no normal client would ever read. A client who should take 5 hours monthly takes 15.

They negotiate everything.

Because they "understand" automation, they question your pricing.
They know that n8n can be used as open source. So they may force you to self-host.
They know Retell charges per minute.
They've seen Make tutorials on YouTube.
So they mentally subtract your "markup" and resent paying for expertise they believe they almost have.
Every invoice invites scrutiny.

They churn predictably.

The relationship has a built-in expiration date. They're not buying a permanent solution; they're buying temporary help until they feel ready to take over.

Six months in, they announce they're "bringing it in-house" or they've "hired someone junior to manage it."

Your recurring revenue disappears, and they take your methods with them.

They often leave bad reviews when things break.

Because they've been tinkering, modifying prompts, adjusting workflows, and changing settings. Then something fails. They blame you.

They don't mention their modifications. Your reputation suffers from problems they created.

Here is how you can filter out DIY clients during the discovery call.

When a prospect asks too many "how" questions rather than "what" questions, that's a signal.
When they ask about exporting workflows or getting trained on the tools, that's a signal.
When they mention they've "played around with Zapier" or "watched some n8n tutorials," that's a signal.

Politely decline or price them out.

Your business model depends on clients who want outcomes, not education.

Rule-11: Your ideal client profile for ongoing revenue.

They have little to no interest in how exactly your automation system works.
They care only about whether it works.
They ask questions like "how many leads did we get this month" or "what was the booking rate this week."
They never request access to your tools.
They view automation the same way they view electricity, something that should just work, delivered by someone else, billed monthly.

These clients respect expertise boundaries.

They don't try to manage their own accounts payable software; they hire an accountant.
They don't try to fix their own plumbing; they call a plumber.
They won't try to manage their own automation infrastructure; they pay you.

Rule 12. Your fees depend on your industry and the type of client you serve.

Your fees mainly depend on the industry you work in and the type of clients you serve, rather than purely on the outcomes you sell.

For example,

If you are targeting real estate, you can easily charge 4 figures per month for your automation.

If you are targeting high-end hotels or enterprises, you can charge even 5 or 6 figure monthly fees.

You can sell the exact same automation to two different clients, says a barber and a solicitor. The latter would likely pay you a lot more for the same booking automation workflow than the former.

The same product or service has different value to different buyers based on their ability to pay and the relative impact on their business.

If you want to charge more, then target clients who can afford to spend more. So choose your target audience wisely from day one.

Stay away from DIY clients who want to hire you to learn automation.

Often, such clients are difficult to work with, question your every step, are very demanding and above all, won’t stick around for very long.

Rule 13. Prioritise what works for you: fewer, higher-value clients or many smaller ones.

I won't tell you to prioritise fewer, higher value clients over many smaller ones or vice versa.

The typical recommendation to pursue fewer, larger clients assumes that client management overhead scales linearly with client count. But productisation changes this equation.

If your system is productised and bulletproof, the marginal cost of adding client number 50 is minimal, mostly onboarding and minor configuration.

In that model, volume wins.

The common advice to consolidate into fewer large clients applies to custom consulting, where each client requires significant personal attention, custom builds, and ongoing strategic input.

This is another reason why rule 2, building a productised system, is foundational.

Rule 14. Report on business outcomes to retain customers.

Don’t report on system performance, bug fixes or health check. Your client doesn’t care whether workflows worked with 100% uptime last month. They only care about how much money they made or saved last month.

A monthly report showing "your workflows had 99.8% uptime" means nothing to a business owner.

A report showing "we generated 47 qualified leads this month, 12 converted to viewings, estimated commission value of 18,000" is undeniable proof of value.

Outcome reporting also gives you ammunition for retainer increases.

When the numbers show you're delivering 5x what you're charging, the conversation about raising fees becomes easy.

Additionally, outcome tracking protects you during client budget reviews.

When finance asks why they're paying you $2,000 per month, your client can point to documented ROI rather than trying to explain what an n8n workflow does.

These rules are harder to execute on day one, but create a much more valuable and defensible business once implemented.

The main prerequisite is having done enough custom work to know what a bulletproof system looks like in your chosen niche.