Pricing AI Consulting — Hourly, Project, and Retainer
There are three pricing models for AI consulting, and each one maps to a specific stage of a client relationship. Hourly is for exploration. Project-based is for deliverables. Retainer is for recurring revenue. The trick is knowing when to use which — and how to move clients up the ladder without making the conversation awkward.
Most people starting out default to hourly because it feels safe. You did the work, you send the invoice, the math is transparent. But hourly pricing has a ceiling built into it, and that ceiling is your available time. The entire trajectory of a sustainable consulting practice moves away from hourly and toward retainer — where the client pays for access and ongoing value rather than counting your minutes.
Hourly: The Starting Position
Hourly rates for AI consulting in 2026 sit between $150 and $300 per hour, depending on your niche and geography [VERIFY]. The range is wide because "AI consulting" covers everything from showing a dentist how to use ChatGPT to redesigning a mid-market sales pipeline with custom automations. If you're doing the former, you're at $150. If you're doing the latter, you're closer to $300 — or you should be switching to project pricing.
Use hourly for initial discovery calls, one-off questions from existing contacts, and clients who aren't sure what they need yet. The key constraint is a cap: 10 hours, hard limit. If a client needs more than 10 hours of your time, the engagement has enough shape to become a project. Staying on hourly past that point means you're doing project-level work at exploration-level commitment, and the client has no incentive to scope things properly because the meter is running in small, painless increments.
Setting your hourly rate is straightforward arithmetic, not vibes. Take your annual revenue target, divide by 1,000 billable hours — which is the realistic ceiling for a solo consultant once you subtract sales, admin, marketing, and the dead hours between engagements — and that number is your floor. If you want to earn $200K per year, your hourly rate starts at $200. Going below that means you need more than 1,000 billable hours, and you don't have them. The math is non-negotiable even when the market feels competitive.
The problem with hourly is incentive alignment. You get paid more when things take longer. The client gets more value when things go faster. This tension is tolerable for short engagements, but over weeks or months it erodes trust — even when nobody says it out loud. That misalignment is why hourly should be your shortest-lived pricing model.
Project-Based: Where the Real Money Starts
Project pricing works when the deliverable is clear. "Audit your current workflow across three departments and implement AI tools for the two highest-impact opportunities." Fixed scope, fixed price, fixed timeline. The client knows what they're paying, and you know what you're delivering. Both sides can plan around that.
The range for project-based AI consulting runs from about $3,000 for a focused single-department engagement to $15,000 or more for multi-department implementations with training [VERIFY]. The variation comes from scope complexity, not from your perceived expertise. A $3K project and a $15K project might require the same tools and similar technical skill — the difference is the number of stakeholders, the integration points, and the amount of change management baked in.
Pricing a project is hourly math with a buffer. Estimate the hours — honestly, including the meetings that feel pointless but are necessary — multiply by your hourly rate, then add 30% for scope creep and project management overhead. Round up to the nearest $500. The 30% buffer is not padding. It is the realistic cost of things the client will ask for that weren't in the original brief but are "basically the same thing." Every project has these. If you don't build them into the price, you eat them.
The critical skill with project pricing is scope control. The statement of work needs to say — in plain language — what you will deliver, when, and what counts as done. It also needs to say what you will not do. "This engagement covers departments A, B, and C. Department D is out of scope and would be a separate engagement." Clients don't push scope boundaries because they're sneaky. They push them because they don't know where the boundary is until you draw it.
The profit margin on project work comes from getting faster. Your first workflow audit might take 40 hours. Your tenth takes 20. But the client doesn't know — or care — how long it takes you. They care about the deliverable. As you get faster, your effective hourly rate climbs without the client ever seeing a higher number on the proposal. This is why project pricing beats hourly for anyone who plans to do similar work repeatedly.
Retainer: The Actual Goal
Retainer pricing is the destination. A monthly fee — typically $2,000 to $5,000 for small to mid-market clients — in exchange for ongoing access and support. "I'm your AI person. When something changes, breaks, or when you have a question about a new tool, I handle it." The client gets continuity. You get recurring revenue. Both sides win.
The retainer should never be the first thing you sell. It's the second. The natural path is: do a project (the setup), then offer a retainer (the maintenance). "Now that everything's running, do you want me to keep an eye on it and handle updates as the tools change?" Most clients say yes because they've just seen the value of having someone who understands their setup, and the prospect of maintaining it themselves — keeping up with tool changes, troubleshooting when workflows break, onboarding new team members — sounds exhausting.
Setting retainer prices requires two reference points. The floor is the minimum monthly amount you'd accept to be "on call" for this client. If their retainer takes an average of 8 hours per month of your time, your floor is 8 times your hourly rate. The ceiling is the value the ongoing support provides. If your retainer prevents one bad tool decision per quarter that would have cost the client $10,000 in wasted time and subscription fees, the retainer is worth a significant fraction of that number. Price somewhere between the floor and the ceiling, leaning closer to the value side as trust builds.
The economics of retainer work are what make a solo consulting practice survivable. Five clients at $3,000 per month is $180K per year in predictable revenue before you do a single new project. That base eliminates the feast-or-famine cycle that kills most consulting businesses in year one. You can be selective about new project work because you're not desperate. Selectivity improves your client quality, which improves your outcomes, which improves your referrals. The whole flywheel runs on the retainer base.
The Transition Path
The move from hourly to retainer doesn't happen in one jump. It follows a pattern.
Start with an hourly discovery session — two to four hours, capped — where you understand the client's situation. If there's a clear project, propose it. Do the project at a fixed price. Deliver it well. At the end of the project, during the wrap-up meeting, mention the retainer. "The tools you're using now will update every few months. Your team will have questions. Things will break. I can be your ongoing resource for that." Frame it as risk reduction, not as another expense.
The conversion rate from project to retainer — when you ask — runs around 60-70% [VERIFY]. The key phrase there is "when you ask." Most consultants don't ask. They finish the project, send the final invoice, and disappear. The client goes back to figuring things out alone, hits a wall three months later, and either calls you for another project or — more likely — calls someone else because they forgot about you. The retainer conversation needs to happen while the value is fresh.
What Not to Do
Three pricing mistakes show up constantly in AI consulting.
First, competing on price. If you're the cheapest option, you attract clients who chose you because you're cheap, and they'll leave you the moment someone cheaper appears. The race to the bottom in AI consulting is real — the barrier to entry is zero and there are people on Upwork offering "AI consulting" for $30/hour. You cannot win that race. Don't enter it. Compete on specificity, speed, and documented outcomes instead.
Second, working for free to "build a portfolio." Your portfolio is built by doing paid work well, not by doing free work desperately. Free clients don't value your time, they're harder to manage than paid clients, and the work you do for them rarely translates into case studies worth showing because the engagement had no structure. Charge full price from client one and earn it by delivering exceptional results.
Third, avoiding the money conversation. State your pricing clearly and early. "A project like this typically runs $5,000 to $8,000 depending on scope. Does that fit your budget?" If it doesn't, you've saved both of you weeks of circling. If it does, the rest of the conversation is about the work, not the money. Consultants who delay the price discussion end up doing unpaid discovery work for people who were never going to pay their rates.
This is part of CustomClanker's AI Consulting series — how to be the person they call instead of watching another YouTube video.