How should I price an AI automation project for clients?

Analyze time saved: if client saves 10 hours/week, annual savings = 10h * 50 weeks * R$ 200/h = R$ 100k. You charge 20-30% of year-1 savings (R$ 20-30k). Client happy, you paid well.

Should I charge hourly or fixed price for AI automation?

Hourly: you win if project drags (good for you, bad for client). Fixed: you win if you deliver fast (good for both). AI automation is new, fixed price is risky. Better: hybrid (R$ 8k scope + 10h extras).

Should I charge for consulting before building AI automation?

Yes. 2-4 hours consulting (R$ 2-4k) to map process, validate ROI, write specs. After, execution scope is clear. Client knows costs. Zero surprises. You differentiate from devs who just code.

How do I sell AI automation to tech-resistant clients?

Don't say 'AI'. Say 'time savings'. 'Your team spends 30h/month on repetitive task. Automation cuts it to 1 hour/month. Free up 29h for important work. Investment: R$ 15k. Returns in 2 months.' Client understands.

What AI automations have highest ROI and are easiest to sell?

1. Email automation (lead nurture): ROI 300-500%. 2. Data classification (CRM): ROI 200-400%. 3. Document summarizer: ROI 150-300%. 4. FAQ chatbot: ROI 200-400%. Start with what client already knows hurts.

Pricing AI automation (and how to charge for value, not hours)

When you build a customer service bot, an n8n workflow, or any AI automation, the client always asks: “How much does it cost?”.

Answering “R$5 per message processed” or “R$150/hour of development” is the wrong way. Because you’re pricing by COST, not by VALUE.

Value is when the client saves 20 hours per month. That’s real money.

I’ll show you how I priced 4 different automations I deployed in 2026 and which model worked best.

The problem with cost-based pricing

You spent 40 hours developing a bot. You charge 40h * $30/hour = $1,200.

Client thinks: “$1,200 for a bot? I’ll try someone who charges less”.

Because from his perspective, he’s not paying for your hours. He’s paying for the savings the bot generates (or expects to generate).

If the bot saves 10 hours per month for his support team, over 12 months, that’s 120 hours saved. At $20/hour cost, that’s $2,400 value.

Then $1,200 makes sense. Because he breaks even in 1 month.

But if you just say “$1,200” with no context, it sounds expensive.

Model 1: Anchoring on savings

You start by asking the client:

“How many hours per month does your team spend on [repetitive task]?”

Real examples I got:

Dental clinic: 30h/month answering WhatsApp
E-commerce: 25h/month qualifying leads
Consulting: 15h/month pulling data from spreadsheets for reports

Now you do the math:

Hours saved/month: 30
Hourly cost (salary): $20
Monthly savings: 30 * $20 = $600
Annual savings: $7,200

You charge: 40% of annual savings = $2,880 (one-time)
Or: 10% of monthly savings = $60/month (SaaS)

Why does this math work?

Because:

Client breaks even in about 5 months (R$2,880 / R$600)
After that, everything is profit
You’re capturing a fair fraction of the value you create

The clinic I mentioned: saves $600/month. I charged $2,880 setup + $60/month.

ROI for them: 5 months (then pure profit).

Model 2: Setup + Monthly (for SaaS)

When you’re offering ongoing service (bot running forever, workflow needing maintenance), SaaS is better:

Setup: $800-2,000 (development + integration) Monthly: 10-15% of monthly savings

Real example I did:

Client: dental group wanting screening bot. Savings generated: 20h/month saved ($400/month)

I charged:

Setup: $1,200
Monthly: $60/month (15% of $400)

Client thought:

$1.2k setup is steep, but breaks even in 3 months.
$60/month is peanuts compared to savings.
Deal.

Fast forward 6 months: client saved $2,400. Paid $1,200 + ($60 * 6) = $1,680. Net gain: $720.

Satisfied. Contract renewed for +12 months.

Model 3: One-time ticket (for small automation)

When automation is small (simple bot, basic workflow), you don’t want SaaS. The overhead doesn’t justify it.

Then you charge a one-time ticket based on savings:

Client: small clinic, 5h/month saved. Savings: 5 * $20 = $100/month. Annual savings: $1,200.

You charge: 30-50% of annual savings = $360-600.

Client pays once. You deliver. Done.

If they want updates later, it’s another ticket.

Because SaaS for $100/month doesn’t make sense for them (overhead not worth it).

And for you neither. Maintenance overhead doesn’t justify it.

Model 4: Percentage of savings (for large integrations)

When you integrate AI deeply with client’s system and savings are HUGE, use percentage of savings.

Real example: e-commerce client automated lead qualification.

Before: 25h/month manual lead scoring. After: 90% automated, 2h/month tweaking.

Savings: 23h/month = $460/month = $5,520/year.

I charged: 30% of annual savings = $1,656 + $40/month maintenance.

Client:

Saves $5,520/year.
Pays $1,656 + ($40 * 12) = $2,136/year.
Profit: $3,384/year.
Happy.

Cases where value-based pricing fails

There are clients where you CAN’T use value model. Like:

Very small client

Startup that saves 5h/month, worth $100. You can’t charge $600 because the math doesn’t work.

Then you charge a smaller fixed ticket: $200-300. You eat the loss for portfolio/networking.

Client wants to “try it”

“Run the bot for free so I can see if it works”.

Never do this. Without skin in the game, client doesn’t commit. Failed deployment.

What I do: “I’ll run a 2-week pilot for $200. If it works, we do the full contract”.

Client has lower budget than ROI

I had a client saving $2,000/month, but budget was $800. Incompatible.

Then you either:

Reduce scope (50% automation instead of 100%)
Propose SaaS instead of ticket (spread the cost)
Decline and find another client

I declined that one. Because you can’t sell automation to a client without budget.

How you calculate savings

This is critical. If you get savings estimation wrong, the whole model breaks.

Technique 1: Ask them

“How many hours per month does your team spend on [task]?”

They answer, usually over or underestimated. So you:

“Let’s measure it properly for 1 week. Then we calculate for real”.

Technique 2: Track during pilot

If you’re doing a pilot, run bot/workflow for 2 weeks. Compare:

Time before: 30h Time after: 5h (still has adjustments) Savings: 25h (once stabilized, goes to 27h)

Technique 3: Look at similar data

If you’ve done 3 similar bots, you know you save “around” X hours.

New client: “Experience with 3 similar clinics: average 18-22h/month saved. Your estimate?”

My pricing framework in 2026

Here’s my process:

Step 1: Estimate savings with client (Q: hours/month?) Step 2: Calculate annual value (savings/month * 12 * hourly cost) Step 3: Choose model:

If savings < $800/year: small one-time ($120-200)
If savings $1.2-3.2k/year: medium one-time ($480-1,000) or SaaS
If savings > $3.2k/year: SaaS (setup + 10-15% monthly) or percentage (25-35%)

Step 4: Proposal includes:

Estimated savings
What they pay
ROI (when they break even)
Terms (setup vs monthly, contract length)

Real example: proposal for clinic

CLIENT: Dental Group, 8 dentists

PROBLEM: Receives 150 WhatsApp messages daily. Support team spends 30h/month answering simple questions.

SOLUTION: Bot with Claude API + n8n answers basic questions, qualifies leads.

ESTIMATED SAVINGS:

Hours saved/month: 20 (out of 30, 10 stay manual)
Hourly cost: $20
Monthly savings: $400
Annual savings: $4,800

PRICING:

Setup + integration: $1,200 (40 hours)
Monthly: $60/month (15% of savings)

PROPOSAL: “The bot saves you $4.8k per year. You invest $1.2k + $60/month. Break even in 3 months. Then pure profit”.

Client: “Makes sense. Let’s do it?”

Deal.

Mistakes I made (and the lessons)

Mistake 1: Charged hourly when it was automation

First bot I built, charged $200/hour * 35 hours = $7,000. Client nearly fainted. Because he was thinking economy (how much do I save), I was thinking cost (how much it costs).

Should’ve calculated savings (20h/month = $400/month = $4,800/year) and charged $3,500. Client would’ve accepted because he breaks even in 3 months, then it’s pure profit.

Today I NEVER quote automation by the hour. Always by savings.

Mistake 2: Underestimated savings

Client said they’d save 10h/month with data analysis bot. I built it, works perfectly. Turned out 25h/month in practice. But I charged based on 10h, so I left $30k of value on the table.

Should’ve done a 2-week pilot first to measure real vs estimated.

Today I ALWAYS pilot for 2 weeks to measure real savings.

Mistake 3: Infinite SaaS for finite savings

Charged $120/month for client saving $80/month. The math looked fine: $120 is 150% of savings, they’ll love it. But here’s what happened: the client could cut the bot and save $80, or keep paying $120 and net only $-40. Month 3, they canceled.

The problem: client didn’t FEEL the

The meta-game of pricing

Pricing isn’t just math. It signals value. A R$ 2,000 system feels cheap. A R$ 8,000 system feels professional and robust.

High pricing attracts serious clients. Low pricing attracts tire-kickers who will argue about every invoice.

Don’t underprice to win deals. You’ll regret it. Price based on value delivered. If you’re transferring R$ 100k of value, charging R$ 5k is underpricing.

The client that pays R$ 5k expects results. The client that pays R$ 20k expects amazing results and gets better support from you because they’re more serious.

Your price is a filter. Use it wisely.

Portfolio effect

When you build multiple automation systems, each one teaches you something. Your second system costs less to build than your first (you reuse patterns).

So your pricing shouldn’t stay flat. First automation: R$ 10k. Second: R$ 8k (faster build). Third: R$ 6k (reusable templates).

But you can actually maintain high pricing on all three because each saves the client money from day one. You’re not getting cheaper. The client is getting better value per dollar.

The key: don’t drop price because “I’ve built this before”. Drop price only if the client is price-sensitive. If they’re quality-focused, keep price high and deliver faster. Win-win.

Seasonal pricing adjustments

AI automation demand changes seasonally. In Q4 (October-December), companies have budget to spend. In Q1, budgets are depleted.

Smart move: price higher in Q4. Price lower in Q1 and offer longer payment terms to make it easier.

Or: create packages. “Quick automation”: R$ 5k. “Full system”: R$ 15k. Let clients self-select based on budget.

Tiering beats single-price. Tiering lets different segments pay different amounts while feeling the pricing is fair.