What an AI marketing consultant actually does
Most people asking this question have already been through a version of it — an audit, a deck, a list of tools someone recommended and then left them to configure. That's advice, and advice is the easy half of the job.
The harder half is doing the work: setting the systems up, running the campaigns, reading what happened, and adjusting on the next round. The work is all still there. What changed is who's needed to do it, because a lot of what used to take a team of specialists can now run under one senior person with a team of specialist agents, someone who knows what good looks like and can check it.
That's the shift behind the term. The tools that write posts have existed for a while, and most of what they produce still needs a senior read before it ships. What's new is that the same person can now own research, content, campaigns and reporting as one connected job, instead of four vendors who never talk to each other.
I sit on the operating side of that line. I set the strategy, I build the systems that carry it out, and I stay through the part where it either works or it doesn't. That's different from most of what gets called "AI marketing consulting," where the person who wrote the plan is rarely the one running the campaign three weeks later.
The work runs in five stages, and each one feeds the next:
reading the market, the competitors and the audience before anything gets built, so the work starts from evidence instead of a hunch.
turning that reading into content, video and campaign material, checked against what was actually found in DETECT.
putting the work out through SEO, paid media and outbound, so the audience that was mapped is the one actually being reached.
feeding what worked, and what didn't, back into the next round of DETECT, so the next pass starts smarter than the last one.
reading what the run actually produced against what it was supposed to move.
Why everyone selling "AI marketing consulting" means something different
For most people selling it, AI marketing is one silo with AI on top: content, or SEO, or campaigns, or the reporting at the end. Someone who runs the whole thing as one multimedia system, technically, end to end, with measurement and analytics built into it, is rare. Twenty years combining generalist strategy for global brands with operating for smaller clients and my own projects is what makes that possible, and it's paying off now. That's the gap the three kinds below all leave open in their own way.
The advisor
Comes in, audits what you have, hands you a roadmap and a tool list, and leaves. The thinking is often genuinely good — the gap is that nobody's there when the roadmap meets your actual team, your actual budget, and your actual week.
The agency
Sells hours across a stack of specialists you mostly never meet. It can work well, especially at scale, but it's built to run many accounts at once, which means the senior attention on any single one of them has a ceiling built into the model.
The software vendor
Writes the guides, runs the webinars, and every piece of content quietly leads back to buying the platform. Useful for learning the space and understanding what the tools can do, less useful when you need someone to actually answer for a campaign that underperformed.
The operator
Someone like me, who builds the system and stays to run it.
What the work looks like →I don't think any of the three is dishonest as a category. Each one is built for a different kind of buyer, and each one has strengths and weaknesses that are worth knowing before you sign anything.
Before paying anyone selling "AI marketing consulting," ask the same three questions regardless of which type they are: who's actually doing the work, what happens the day after the contract ends, and what have they run on their own brand before they ran it on yours.
How I do it: consulting and operating, end to end
I cover the whole thing — analytics set-up first, then creation, campaigns, and the measurement after execution. That's the part most versions of this stop short of, and it's the part that actually tells you whether any of it worked.
Tech set-up comes first: the analytics, the right metrics and the connections between the accounts and the ad platforms, so there's one place the data actually lives instead of a handful of logins nobody ever checks side by side. Creation follows from what that data shows. Then campaigns go live across the channels that make sense for the brief, and every change gets a reason attached to it, so a decision made this month can still be explained months later.
The measurement itself comes after execution, because the only honest read on a campaign is what it actually did once it ran. That reading feeds straight back into what gets learned and what changes on the next round — the same LEARN stage that closes the loop described above.
AI is what makes covering all of that possible for one operator with a team of agents. It used to take a team of people to run research, build content, manage campaigns and report on all of it. Now a lot of that runs on agents I've built myself, on the latest AI stack, which means the person setting the strategy is the same one watching the campaign live and deciding what to change.
I still pull in specialists. They implement the optimizations the system recommends, and in design and creative content ideas a real specialist makes a visible difference, so for bigger clients I'll bring one in when the budget allows — it's always the better option. Depending on the project size, a tech person on the client side is recommended to be in charge of the implementation. For most SMBs that extra spend makes less sense, and the systems are built for exactly that: they pick up what's trending, the zeitgeist, in real time, so the work stays current.
The agents themselves are built as the specialist roles an enterprise campaign needs: research, content, media, PR, analytics, each with its own instructions and its own checks. That's what lets one operator run work at that level without the headcount that used to come with it.
The strategy underneath it is twenty years old — media planning, account leadership, brand strategy, learned inside WPP, Publicis and Dentsu and then tested again running my own company. AI is how I act on that judgment faster, on more fronts, without losing the thread between the plan and the result.
The system I'm building now is the one described above, and I'm implementing it in two places at once: on my own brand, where I can run it in the open, and with clients, where the stakes are real. It's a working system, still being built, maximizing AI tools to get more out of marketing I already know how to run. This whole site runs on it.
- 01Analytics set-up
- 02Creation
- 03Campaigns
- 04Measurement after execution
- 05Learn
Where the judgment comes from
It started at MindShare, inside WPP, planning media for Unilever brands like Flora and Magnum in 2006 — the kind of job where you learn the mechanics of a media market from the inside, one insertion order at a time. From there to Starcom, Publicis, as Account Supervisor on Samsung Spain in 2008, sitting between the local client and the European region and learning how strategy actually travels between the two.
From there to Carat, part of Dentsu, working up from the team to Account Director across Nokia, Kellogg's and Carlsberg between 2009 and 2016. The wins from that stretch: 2014 Market of the Year with Kellogg's, 2013 Best European Activation for social — both won leading the team, in the room.
After that I built my own company, Universo Miel, growing it on inbound to tens of thousands a month, and closed it on my own terms in 2024. DAIRE Consulting came next, a B2B data-science practice co-founded in 2021 — the bridge between the agency world and what I run now. I went independent to run strategy and execution as one thing, instead of handing the second half of the job to someone else and hoping it survived the handover.
How it lands, operationally
Four samples of how the work lands once it's running, at the level I'd explain them to a client:
SEO, run as a system.
A horsepower analysis of the whole site against proven SEO frameworks and the engines' own guidelines: what to fix, in order. Every page's content read for relevance, then clustered, reordered, relinked or removed so the signals to search and answer engines get stronger instead of diluted. Detailed recommendations on what to modify, add or cut, produced through deterministic steps and local models, so the reports are accurate and don't hallucinate. Then weekly measurement, and the next round of recommendations built on what actually moved.
Paid media, with agents on the wheel and a human on the decisions.
Google, Meta, LinkedIn and X Ads connected to one place. Budgets, bids and creative performance read on a loop, changes proposed with the reason attached, and only the decisions that carry real money reach a person. The account stays legible: a change made in March can still be explained in June.
Partnership decisions before the budget moves.
Before a creator, a publisher or a partner goes into a plan, they get judged against the actual brief: who really watches them, where they're heading, whether their paid work sells. One verdict per name, with the reasoning, in a document the client can forward. The same read I use for my own collaborations.
Consumer intelligence before you spend.
What the market is asking for, read live from the places people actually talk, plus a synthetic focus group built on census-grounded persona data to surface objections and reasons to buy before a euro goes to real respondents. It's a starting point for the creative and the media plan, checked again once real results come in.
Need help with any of this?
Tell me what you're working on. If I don't believe I can help, I'll say so on the first call.
Work with Sonia →Need help? How I work with clients
Two ways in, one system underneath. Either way it's a managed service: you get the result, and a human stays in charge of every call that matters.
- For agencies. I plug the system into your delivery, managed or embedded. You keep the client and the relationship; I bring the engine and the judgment, under your brand if you want it that way.
- For brands. You hire the outcome. I detect the signal, create the work, run and optimize the campaign, and read the result, without you building the AI stack yourself.
What you can hire today
- AI Growth Systems Consulting — the anchor module. This is where most engagements start, and it's the one that covers the full DETECT-to-LEARN loop rather than a slice of it.
- AI-Ready SEO Audit — a read on how the site currently reads to search engines and to the AI answer engines that are starting to matter just as much, with a prioritised list of what to fix first.
- Viral Scripts — short-form video scripts built from what's actually working in a niche right now, ready to shoot.
- Multimedia Personal Brand Content — content built and shipped across a brand's channels, planned and produced as one system instead of one-off pieces.
- AI Avatar Strategy & Creation — a consistent on-camera presence, built once and reused across scenes, so a brand can keep producing video without booking another shoot every time.
- Creator Validator — independent due diligence on a creator before budget goes anywhere near them, with a verdict you can hand straight to a client. See it →
None of these carry a fixed price on this page — scope gets set on a call, against what you actually need. The list keeps growing as I open more modules.
How an engagement starts
- Tell me what you're working on. No form, no gatekeeping — an email describing the actual problem.
- A call. I'll tell you honestly whether I think I can help. If I don't believe I can help you, I'll tell you that on the first call.
- A scoped start. If it fits, we agree what the first piece of work is and what it should produce.
Questions people ask
What does an AI marketing consultant do?
Uses AI across research, content, campaigns and measurement to run marketing work that used to need a bigger team. The version worth hiring builds the system and runs it, rather than handing you a plan and leaving.
How is this different from hiring an agency?
An agency puts hours from a team you mostly don't see against your account. I'm one person who sets the strategy and operates the systems that carry it out, so nothing gets lost between the person who decided and the person who executed.
Do you replace my marketing team?
No. I run the systems and the campaigns that sit on top of what your team already owns, and I plug in where the work needs a deeper hand — research, content production, campaign management, reporting.
What does an engagement cost?
It's scoped on a call, against the module or modules you need. There's no fixed price to quote here because the work ranges from one module to the full loop.
Can you work with my existing tools and data?
Yes. The systems I run are built to connect into what a client already has — analytics, ad accounts, content tools.
Do you work outside Spain?
Yes, worldwide. I'm based in Madrid and I work remotely with agencies and brands wherever they are.