Commercial

A while back I sat in on a sales call a friend was taking. An “AI marketing agency” was pitching her. Slick deck, lots of the word “AI,” a $6,000-a-month retainer at the end. I asked one question afterward: what does the AI actually do here that a normal agency wouldn’t? She didn’t know. Neither, it turned out, did they. It was the same content-and-reporting service they’d sold for years, with “AI” stapled to the front.

That’s most of this market right now. So before you spend a euro, here’s the plain version of what you’re actually shopping for.

AI marketing and consulting services are paid help that uses AI to do marketing or business work faster and cheaper than a person doing it by hand. That’s the whole category. Underneath it sit five real things you can buy: an audit (someone tells you where AI would actually help), AI SEO services (content and search work, sped up by AI, plus the new job of getting cited in AI answers), AI marketing services (the wider marketing function, run AI-native), AI automation (someone wires your tools together so the busywork runs itself), and a fractional marketing leader (a senior operator part-time). Each does a different job. Each costs a different amount.

And the costs aren’t a mystery, despite what most pricing pages pretend. A one-off audit runs roughly $5,000 to $15,000. AI SEO and marketing retainers sit around $1,500 to $5,000 a month for a small business. AI automation projects run $1,500 to $15,000 to build, then $2,000 to $10,000 a month to run. A senior AI consultant charges $150 to $500 an hour, with focused projects at $5,000 to $50,000. A fractional marketing lead lands at $5,000 to $10,000 a month for most companies. I’ll break every one of these down below, with the survey data behind it.

The catch is that the price tells you almost nothing about whether you’re getting the real thing or a relabel. That’s the part this guide is actually about.

There’s a number worth sitting with before you spend anything. RAND found that 80% of AI projects fail to deliver what they promised. And McKinsey’s 2025 State of AI survey found that while 88% of organizations now use AI somewhere, only about 6% see real bottom-line impact from it. So most spending on AI is, right now, money lit on fire. Not because the tools are bad. Because the work was scoped wrong, or sold by someone who’d never run the function themselves.

That’s the gap this whole guide is about closing. I’m a growth operator, not an agency. I’ve run marketing for global brands and I build my own AI tools, so I’m telling you this from inside the work, not from a slide. The honest read on buying AI help: the category matters less than the person, the price is more knowable than they let on, and you should almost always start with the smallest possible step.

My take: The single best filter I know costs nothing. Ask the person pitching you, “what would you build, specifically, in the first thirty days, and what would I stop doing once it’s built?” An operator answers in concrete nouns: this workflow, this tool, these four hours a week back. A reseller answers in adjectives. The ones who can’t name what they’d actually do are selling you the word, not the work.

Start with an audit, not a retainer

The move almost nobody sells you is the small one. Before you sign anything ongoing, pay for an honest look at your own setup. An AI audit (or AI readiness assessment, same thing) is someone competent going through how you actually work and telling you, in plain terms, where AI would buy back real hours and where it just wouldn’t.

Why start here? Because the failure data above is mostly a scoping problem. Teams buy a retainer to “do AI” before anyone has decided what AI is even supposed to do. The audit is the cheap insurance against that. A two-to-four-week AI readiness audit runs roughly $5,000 to $15,000 in 2026, and a good one pays for itself by stopping you from buying the wrong $50,000 project.

A real audit checks four things, and most skip at least one: the workflows (where the time actually goes), the tools (what you already have and aren’t using), the data (what you’re sitting on that could feed a better result), and the skills (whether your team can actually run what you’d buy). If you want to see exactly what a thorough one covers, and even score yourself first, the AI audit checklist walks through all four and gives you a 90-day plan to act on it.

My take: A surprising amount of the time, the most honest audit result is “you don’t need to hire anyone yet, go use the $20 tools for three months first.” I’ve told people exactly that. It costs me the bigger engagement and it’s still the right call, because if you’re not getting value from ChatGPT on your own, you’re not ready to pay someone to build you something fancier. An audit that can only ever conclude “yes, hire us” isn’t an audit. It’s a sales call with a price tag.

AI marketing and SEO services

This is where the relabeling is thickest, because SEO and marketing services existed long before AI did. So the buyer’s question isn’t “should I do AI SEO,” it’s “is the AI doing real work here, or is it a sticker?”

The plain version of what’s actually changed: AI makes the execution faster and cheaper. The research, the first drafts, the technical audits, the competitor digging, all of it compresses from days into hours. That’s real, and it should show up in either lower prices or more output for the same money. What hasn’t changed is the strategy layer, deciding what to target and how to stand out, which is still a human call. Google’s own line settles the panic about AI content: they reward helpful content and demote unhelpful content, and don’t care who or what typed it.

There’s a second, newer job stacked on top: AI-citation work. When someone asks ChatGPT or Google’s AI Overview a question, you either get named or you don’t, and that’s a different game from ranking blue links. It’s real but small for now, AI platforms still drive only about 1% of all web traffic per Conductor’s analysis of 13,770 domains. The SEO agency Jeremy Moser put the honest version best: “80% of GEO is good, fundamental SEO. If a GEO service does not openly tell you that, they are selling you snake oil.” Google’s own John Mueller is blunter, warning that the louder someone pushes brand-new AI-search acronyms at you, the more likely they’re “just making spam and scamming.”

Three essays go deep here:

• For the full breakdown of what an SEO engagement actually delivers, the honest timelines, and how to spot the relabels, read AI SEO services: what you’re actually buying.
• If you’re weighing a part-time senior marketer instead of an agency, fractional CMO covers what one really is and the moment you actually need one.
• And because the number is what everyone really wants, fractional CMO cost lays out what you’ll pay in 2026, straight from survey data.

My take: The tell for a real AI SEO provider is that they’ll talk you out of things. They’ll say below a certain budget you’re getting automated reports and cookie-cutter content, so don’t bother. They’ll tell you SEO is a 6-to-12-month game and AI doesn’t change Google’s clock, only the speed of your own work. A relabel promises fast rankings and “AI-powered” everything. The honest one sounds slower and more boring, which is exactly why I’d trust it.

AI consulting and automation services

This is the broadest area, and the names blur together, so let me draw the line cleanly. AI consulting is strategy and judgment: someone figures out where AI fits your business, what to build, and how to get your team to actually use it. AI automation is the building: wiring your tools together so a workflow runs without a human copy-pasting between tabs. Some people do both. Many do only one and imply both.

The strategy half is where the money is usually well spent or wasted, because the hard part of AI isn’t the tech. BCG’s framework says it plainly: AI success is about 10% the algorithms, 20% the data, and 70% the people and process changes. Most pitches lead with the 10% (the part that doesn’t fail) and skip the 70% (the part that does). That’s the whole reason 80% of AI projects fail. A consultant who only sets up tools is working in the wrong 10%.

The automation half is more concrete and easier to price. It’s the boring, valuable stuff: connect your form to your CRM, route leads, draft the follow-ups, pull the weekly report by itself. Real, useful, and often something you can start yourself before you ever hire anyone.

Three essays carry this:

• For the full picture, what an AI consultant does, the types of firms, the real costs, and the red flags, start with what is AI consulting.
• For the building side, what gets wired up and what you can do without hiring anyone, read AI automation agency services.
• And if you’re a smaller shop wondering whether any of this is even for you, AI consulting for small businesses covers what it looks like at your size and what it should cost.

My take: The most useful thing a consultant can tell a small business is “do this part yourself first.” Get a $20 ChatGPT or Claude subscription, use it daily for a month, and you’ll learn more about where AI fits your business than any deck will tell you. The JP Morgan Chase data on 4.6 million small businesses shows the median AI subscription is around $28 a month. The tools are cheap. Outside help earns its price only once you’ve outgrown the “paste it into the chat” stage and need a real, repeatable system across your team. Pay for the system, not the software setup you could’ve done yourself.

The service menu, in one table

You’ve now seen all five. Here’s the whole menu in one place, by the job each one does, so you can find your own row before you look at any price.

Notice that the first row and the last row are the two most people skip. They jump straight to “buy a service” without the audit that tells them which row they’re in, and they reach for an agency when what they actually need is one senior person part-time. Match the row to the real problem first. The price comes second.

What it all actually costs

Nobody likes this section and everybody searches for it. So below is the honest table, pulled from real surveys, not pricing-page fantasy. These are US-dollar figures from the source data. European rates tend to run 20 to 30% lower for equivalent work, so treat the numbers as the high end if you’re hiring outside the US.

A few things worth saying plainly about that table.

On SEO, Ahrefs polled 439 professionals and found freelancers average about $72 an hour, agencies about $99, and consultants about $171. The realistic monthly range for a small business that wants meaningful results is $1,500 to $5,000. Below $1,000 a month, you’re almost always getting automated reports and cookie-cutter content, not strategy.

On consulting, seniority drives most of the price. Junior consultants run $100 to $150 an hour, mid-level $150 to $300, and senior specialists $300 to $500 or more, with the big firms at $400 to $600. Solo operators usually cost less than firms, because there’s no project manager, account manager, and office to pay for. All the hours go into the actual work.

On fractional leadership, the cleanest data comes from the FRAK State of Fractional report, a survey of 250 fractional professionals: 40% charge $5,000 to $8,000 a month, and most of the market sits in the $5,000 to $10,000 band. For comparison, a full-time CMO costs $350,000 or more a year all-in once you add benefits, bonus, and recruiting. A fractional at $10K a month is $120K a year, roughly a third of the cost.

One pattern runs through all of it: AI is starting to push these prices in two directions at once. The standard, repeatable work is getting cheaper because AI does it faster. The genuinely custom, judgment-heavy work stays expensive, because that’s the part AI can’t do. A provider whose prices are all going up while claiming AI makes them faster is worth a hard question.

My take: Price is a terrible proxy for quality in this market, in both directions. The cheapest option is usually the most expensive one, because the average abandoned AI project burns $7.2 million in sunk cost per Deloitte, and a cheap provider who scopes it wrong puts you on that path. But the most expensive option isn’t safe either. A big firm’s rate is mostly overhead and a junior doing the actual work. Pay for the person who’ll be in the room doing the thing, not for the logo on the deck.

How to tell a real operator from a relabel

This is the question under all the others, so let me make it a checklist you can run on any provider in five minutes.

The relabel tell is vagueness about the work. They sell “AI-powered transformation” and can’t name the specific workflow they’d build or the hours you’d get back. They lead with the tools (the 10%) and go quiet on the people-and-process part (the 70%) that actually decides whether it works.

The operator tells are concrete:

1. They’ve run the function, not just advised on it. They can tell you about a marketing job they actually did, with the boring details, because real results have details and fakes have buzzwords.
2. They’ll talk you out of things. Out of a retainer you’re not ready for, out of a budget tier that won’t work, out of hiring at all if the $20 tools are enough for now.
3. They name what they’d build and what you’d stop doing. Specific nouns, a thirty-day plan, hours bought back.
4. They have case studies with real numbers, the “cut handling time from 4 hours to 20 minutes” kind, not “grew a client 300%.”
5. A senior person does the actual work. If you’re paying senior rates, make sure a senior person, not a junior team behind the logo, is in the room.

This matters more than picking the right category, because the same money buys wildly different things depending on who’s behind it. The market grew faster than its quality controls. The title “fractional CMO” alone went from a couple thousand people on LinkedIn to over a hundred thousand in two years, and some of those are former CMOs who’ve scaled real brands while some are managers who liked the title. Your filter is the work, not the label.

Should you outsource this, or do it yourself?

This is the question that saves people the most money, and almost nobody asks it before buying. Here’s the decision rule I actually use.

The honest default is do-it-yourself first. Most small businesses can get real value from ChatGPT or Claude on their own, and 62% of small businesses that haven’t adopted AI say the blocker is just not knowing where to start. That’s a confidence-and-skills gap, and you close it fastest by using the cheap tools, not by buying an expensive engagement.

Outside help earns its price at the wall: when you need many tools connected with real conditional logic, when the stakes are too high to learn on the job, or when you need a repeatable system across a whole team rather than one person’s browser tabs. Below that wall, a consultant who’s honest will tell you to go do it yourself. Above it, the math flips, because a $10,000 sprint that actually works is cheaper than $30,000 of tools and internal time that goes nowhere.

My take: Outsource the system build, never the judgment. The thing worth paying for is someone setting up a workflow that keeps working after they leave, and training your team to run it. The thing not worth paying for is renting someone’s thinking forever, or paying them to do a tool setup you could’ve done in an afternoon. The whole point of bringing in good help is that the leverage ends up yours, not theirs. If after the engagement you can’t run it without them, you didn’t buy a system. You bought a dependency.

Frequently asked questions

What do AI marketing services actually include?

AI marketing services use AI to do marketing work faster and cheaper than doing it by hand. In practice that’s content (research, drafts, repurposing), SEO and search work, the newer job of getting cited inside AI answers like ChatGPT and Google’s AI Overviews, and sometimes paid ads, email, and reporting. The genuinely useful versions use AI to speed up execution so you get more output or a lower price, while keeping a human on the strategy, what to target and how to stand out. The relabeled versions are the same service that existed before AI, with “AI” added to the name and the price unchanged. The test is whether they can name what the AI specifically does that a normal service wouldn’t.

How much does AI consulting or a fractional CMO cost?

A senior AI consultant charges $150 to $500 an hour, with focused projects at $5,000 to $50,000 and monthly retainers at $2,000 to $10,000, based on multiple 2026 pricing surveys. A fractional marketing leader runs $5,000 to $10,000 a month for most companies, where about 40% of the market sits according to the FRAK survey of 250 professionals. For comparison, a full-time CMO costs $350,000 or more a year all-in, so a fractional at $10,000 a month is roughly a third of the cost. European rates tend to run 20 to 30% lower than the US figures. The price is driven mostly by seniority and scope, and a solo operator usually costs less than a firm because there’s no overhead for project managers and offices.

What’s the difference between an AI consultant and an AI automation agency?

An AI consultant gives you strategy and judgment: they figure out where AI fits your business, what to build, and how to get your team to actually use it. An AI automation agency does the building: wiring your tools together so a workflow runs without a human copy-pasting between tabs. Consulting answers “what should we do and why.” Automation answers “build the thing that does it.” Some providers do both, many do only one and imply both. The reason the distinction matters: BCG’s research says AI success is about 70% people and process and only 10% the tools, so an automation shop that skips the strategy and adoption work often builds something technically fine that nobody uses.

How do you tell a real AI operator from a relabeled agency?

Ask what they’d build specifically in the first thirty days, and what you’d stop doing once it’s built. A real operator answers in concrete nouns, this workflow, this tool, these hours back, because they’ve run the function themselves and real work has details. A reseller answers in adjectives like “AI-powered transformation” and goes vague on specifics. Other tells of the real thing: they’ll talk you out of buying something you’re not ready for, they have case studies with real numbers rather than “grew 300%,” and a senior person does the actual work instead of a junior team behind the logo. The fastest single filter is to ask what they built at their last three jobs.

Should you outsource AI work or do it yourself?

Do it yourself until you hit a wall. Most small businesses can get real value from ChatGPT or Claude on their own at around $20 a month, and the most common blocker isn’t cost, it’s just not knowing where to start. Begin with the cheap tools daily for a month, automate one simple workflow, and learn by doing. Bring in outside help once you’ve outgrown the “paste it in the chat” stage, when you need many systems wired together with real logic, when the stakes are too high to learn on the job (compliance, deliverability, revenue), or when you need a repeatable system across a whole team rather than one person’s browser tabs. The rule: outsource the system build, never the judgment.

Where should you start if you’re not ready for a retainer?

Start with an audit, not a retainer. An AI audit or readiness assessment is a one-off engagement, roughly $5,000 to $15,000, where someone competent looks at how you actually work and tells you where AI would buy back real hours and where it wouldn’t. It’s the cheapest way to avoid the 80% of AI projects that fail, because most of those failures are scoping problems, buying a service before deciding what it’s for. A good audit checks your workflows, tools, data, and skills, and gives you a prioritized plan. The most honest result it can return is sometimes “go use the $20 tools for three months first, you’re not ready to hire yet,” and a real one is willing to say that.

Want to know what’s actually worth doing?

If you’ve read this far, you’re trying to spend money well, not just spend it. The outcome worth aiming for is simple: a clear picture of what’s worth doing in your business, what it’d realistically cost, and what you should just do yourself. Not a retainer you’ll wonder about in three months.

So the smallest sensible first step is the audit, not the contract. I’ll take an honest look at where AI would actually buy you back hours and where it wouldn’t, and tell you straight, including if the answer is “you don’t need to hire anyone yet.” I’m a growth operator who builds these systems, not an agency reselling them, so the read you get is from inside the work.

The first conversation is a free 15-minute spar. No pitch, no slides, just a straight take on what your setup actually needs right now. Come grab a slot and bring the one thing you’re trying to figure out whether to buy.