How to choose an LLMO consultant

I am one of the options on this page, which you should factor in. I have tried to write it as I would explain it to someone who had already decided not to hire me, because that is the only version worth reading.

There are five ways to get this done. They are genuinely different, and the right answer depends far more on your size and your situation than on which is best in the abstract.

What criteria should I use to choose an LLMO consultant?

Eight criteria, in the order I would apply them. The first one eliminates most candidates on its own: ask them to show you what AI systems say about your business before you pay anything. Anyone competent can do that in the first conversation. Anyone who cannot is working from a template.

For each criterion below there is a question to ask, and what the answer sounds like when it is good and when it is not.

1. Will they measure you before you take the meeting?

A good answer sounds like: They show you, unprompted, what ChatGPT or Gemini says today when asked for a business like yours — and which sources it used.

A bad answer sounds like: They ask for your goals and budget first, then propose a package. That is a template, not a diagnosis.

2. Can they tell you where the answer comes from?

A good answer sounds like: They talk about review platforms, directories, trade press and structured data — the sources a model reads. They are specific about which ones matter in your industry.

A bad answer sounds like: They talk only about your website. That is the one source a model trusts least, because it is you describing yourself.

3. Do they refuse to guarantee placement?

A good answer sounds like: “Nobody controls what ChatGPT recommends. Here is what we can influence and here is how we will measure it.”

A bad answer sounds like: Guaranteed AI rankings, guaranteed citations, guaranteed position. Nobody can sell that, so anyone selling it is either mistaken or dishonest.

4. Do they start with consistency rather than content?

A good answer sounds like: The first thing they fix is your name, address, phone number and category being identical everywhere. It is dull, it is cheap, and it outperforms everything else.

A bad answer sounds like: The first deliverable is twelve blog posts a month. That is an SEO retainer with new vocabulary.

5. Do they measure each assistant separately?

A good answer sounds like: They report ChatGPT, Gemini, Perplexity and Copilot individually, because those systems do not share an index and routinely disagree.

A bad answer sounds like: They report a single “AI visibility score.” There is no such number; it has been invented for the deck.

6. Will they tell you what you do not need?

A good answer sounds like: They rule part of their own service out for you, or tell you to do something yourself. That is judgement and it is the whole reason to hire an adviser.

A bad answer sounds like: Everything you asked about is included and everything is recommended.

7. Can they show a before and after on a real client?

A good answer sounds like: A named prompt set, the answer before, the answer after, and the date. Ideally a client who will take your call.

A bad answer sounds like: A logo wall, a traffic chart from a different discipline, or a case study with no prompts in it.

8. Who actually does the work?

A good answer sounds like: The person in the room. If there is a team, you are told who does what and you meet them.

A bad answer sounds like: The senior person sells and a junior delivers. Common, rarely disclosed, and worth asking about directly.

The five kinds of provider

1. A large consultancy

The big firms — the ones whose names you already know — do have serious capability here. They will rebuild your data architecture, integrate your platforms, retrain your team, and treat AI visibility as one workstream inside a larger transformation.

Good for: enterprises that want one accountable partner across strategy and delivery, and that have a procurement process which requires a recognisable name.

The trade: price and pace. These engagements are built for scale rather than speed, and the people who sold the work are usually not the people doing it. For any business under roughly fifteen million in revenue, the economics do not make sense.

2. A specialist LLMO agency

A genuine and growing category. These firms do nothing else, and the good ones have built real tooling — prompt monitoring across models, citation tracking, structured-data pipelines.

Good for: companies with an existing marketing team who need a specialist capability bolted alongside it, and enough volume to justify a monthly retainer.

The trade: the category is eighteen months old, so track records are short and claims are hard to verify. Many are SEO agencies that renamed a service line. Ask how long they have been doing specifically this, and ask to see the measurement they run before and after.

3. Your existing SEO agency, extending into AI

Often the pragmatic choice, and frequently dismissed too quickly. AI visibility and search visibility share most of their foundations: structured data, consistent entity facts, content that answers real questions, corroboration from independent sources. An agency that does those well is already doing much of the work.

Good for: businesses with an agency relationship that is working. Switching costs are real and continuity has value.

The trade: the gap between doing SEO well and understanding retrieval is real but narrow. The test is whether they can explain why a model names one business and not another without falling back on ranking language.

4. An independent specialist

One senior person, working directly. No account management layer, no hand-off to a junior team.

Good for: owner-led businesses where the decision-maker wants to talk to the person doing the work, and where the problem is strategy and positioning rather than production volume.

The trade: capacity and continuity. One person cannot run a content operation, and one person can get ill. If you need eight blog posts a month and social management, this is the wrong shape and you should hire an agency.

5. Doing it yourself

Genuinely viable for a lot of small businesses, and almost nobody selling these services will tell you so.

The highest-return work is unglamorous: make your business name, address and phone number identical everywhere they appear; claim and complete your Google Business Profile; get listed on the directories your industry actually uses; add structured data stating plainly what you do and where; and answer the questions your customers ask in the words they use.

If your budget is under a thousand a month, do it yourself. You will get most of the benefit.

The trade: it takes attention you may not have, and you will not know what you are missing. But a business that does those five things properly will beat a competitor who paid for a retainer and never fixed their address.

How to tell competence from vocabulary

The market is full of people who learned the words faster than they learned the work. The phrase is being added to job titles across the industry faster than the skill is being acquired, so the title alone tells you very little. So stop assessing the vocabulary and ask for the receipts.

Ask them to measure you before you pay. Anyone competent can show you, in the first conversation, what assistants currently say when asked for a business like yours. If they cannot, they are working from a template.

Ask what they would not do. A consultant with judgement will tell you which parts of their service you do not need. One selling a package will not.

Ask where the answer comes from. If they talk only about your website, they have misunderstood the mechanism. Assistants build answers mostly from sources you do not own.

Ask for a client you can call. Not a logo wall. A person, on the phone, who will tell you what changed.

What I would tell a friend

The budget question turned out to be the part people quote back most, so it has its own piece now: what to spend on AI visibility, and why budget is only half the question.

Under a thousand a month of available budget: do it yourself, using the list above. You will capture most of the value.

One to five thousand: an independent specialist or a good SEO agency extending into this, depending on whether your problem is thinking or production.

Five to twenty thousand with a marketing team in place: a specialist agency, with a measurement regime you agree before you start.

Above that, with genuine complexity: a large consultancy, and accept the pace.

The worst outcome is not choosing the wrong one of these. It is paying any of them for twelve months without ever having measured whether an assistant knew who you were at the start.

Common questions

What criteria should I use to choose an LLMO consultant?
Eight criteria, in order. Will they measure you before you pay — showing what AI systems say about your business today and which sources they used? Can they tell you where the answer comes from, meaning review platforms, directories and structured data rather than only your website? Do they refuse to guarantee placement, since nobody controls what ChatGPT recommends? Do they start with consistency of name, address and phone number rather than content volume? Do they measure ChatGPT, Gemini, Perplexity and Copilot separately, because those systems disagree? Will they tell you what you do not need? Can they show a before and after on a real client with the actual prompts? And who does the work — the person in the room, or a junior after the sale?

How much should an LLMO consultant cost?
It varies by shape of engagement. Below roughly a thousand a month of available budget, most businesses get more value doing the foundational work themselves. Between one and five thousand, an independent specialist or an SEO agency extending into AI visibility is usually the right fit. Above five thousand with an in-house marketing team, a specialist agency becomes viable. Large consultancies only make economic sense for enterprises.

Can I do LLMO myself instead of hiring a consultant?
Yes, and many small businesses should. The highest-return work is unglamorous: identical name, address and phone number everywhere; a complete Google Business Profile; listings on the directories your industry uses; structured data stating what you do and where; and content answering customer questions in the words customers use. A business that does those properly will outperform one that pays a retainer and never fixes its address.

How can I tell if an LLMO consultant is competent?
Ask them to measure you before you pay. Anyone competent can show you in the first conversation what AI assistants currently say when asked for a business like yours. Ask what they would not do, since a consultant with judgement will tell you which parts you do not need. Ask where the answer comes from, because a consultant who talks only about your website has misunderstood the mechanism. And ask for a client you can telephone.

Is an LLMO agency better than my existing SEO agency?
Not necessarily. AI visibility and search visibility share most of their foundations, so an agency doing SEO well is already doing much of the work. The gap is real but narrow. The test is whether they can explain why a model names one business and not another without falling back on ranking language.