LLMO consulting can cover five things: diagnosing your current situation, unifying your company information, revising your copy, implementing structured data, and measuring results. Asking beforehand what will be done, how far it goes, and how it will be verified prevents a gap between expectations and reality. If a proposal guarantees that you will appear in AI answers, ask what that guarantee is based on.
LLMO consulting covers five things, from diagnosis to measuring results
LLMO consulting is support aimed at getting your company name and pages mentioned in the answers given by conversational AI such as ChatGPT and Gemini. The details vary by provider, but most services consist of the following five tasks.
- Diagnosing the current situation. Questions about your company are entered into AI tools to see what answers come back.
- Unifying company information. Your company name, address, phone number and other details are made consistent across your website and external listings.
- Revising the copy. The text is rewritten so AI can extract it easily, for example by stating the conclusion right after each heading and citing sources for figures.
- Implementing structured data. Company information and frequently asked questions are put into a machine-readable form that matches what is shown on screen.
- Measuring results. The same questions are entered regularly and changes in the answers are recorded.
Of these five, items 2 and 3 can be handled in-house. Item 4 requires technical knowledge, so it is the area where hiring a specialist pays off most readily. Items 1 and 5 involve deciding on the questions to ask the AI and repeating them over time; you can start this yourself, but recording and analyzing the results takes time, so many companies outsource it.
10 points to check before you hire
When comparing providers, ask the following 10 questions. The more specific the answers, the clearer the actual work becomes.
- What questions will you enter, and into which AI tools, during the diagnosis?
- In what form will we receive the diagnosis results?
- Which pages will be revised? All pages, or only some?
- Will you write the revised copy yourselves, or only propose a direction?
- Will you implement the structured data yourselves?
- How, and at what intervals, will results be checked?
- How often will reports be delivered, and what will they contain?
- What is the contract term, and what are the conditions for cancelling mid-term?
- What work is included in the fee, and what is not?
- Do you guarantee inclusion or citation? If so, on what grounds?
Point 10 is especially important. The content of AI answers is determined by the companies that operate the AI, and it is not something an outside firm can guarantee. If a proposal offers a guarantee, ask what it is based on, or compare it with other providers.
Warning signs in a proposal
Be cautious of proposals like the following.
- Guarantees such as “you will definitely appear in AI answers” or “you will rank number one”
- No explanation of the actual work, only a claim of a “proprietary method”
- Asking you to sign immediately without showing any diagnosis results
- Assuming a long contract term without stating the conditions for early cancellation
- No explanation of how results will be verified
How AI produces its answers is not publicly disclosed, and the same question will not always return the same answer. On that basis, a trustworthy provider explains honestly what can and cannot be done.
How to think about cost
The cost of LLMO consulting varies widely depending on the provider and the scope of work. When comparing, look not only at the price but also at the following three points.
| What to look at | What to check |
|---|---|
| Scope of work | Diagnosis only, or does it also include copy revisions and implementation? |
| Timeline | A one-off engagement, or an ongoing monthly service? |
| Deliverables | Which do you receive: a diagnostic report, revised copy, implemented data, or progress reports? |
Even at the same price, the value differs if what you receive differs. When requesting a quote, ask for a list of deliverables so you can compare more easily.
Dividing what you do in-house from what you outsource
You do not have to outsource everything. Handling the parts you can do yourself keeps costs down.
The following tasks are easy to handle in-house.
- Compile your basic company information on a single sheet and make the wording consistent across your website, social media and Google Business Profile
- Pick 10 questions you have actually received from customers and add them to your FAQ page
- Write a one-sentence conclusion right after the heading on your key pages
- Enter the same questions into AI tools every month and save the answers
The areas where hiring a specialist pays off most readily are implementing structured data, reviewing the copy across the whole site, and analyzing diagnosis results.
Common mistakes when hiring
Signing a contract without defining your goal
“We want to appear in AI answers” is not enough to judge results by. Decide which questions you want to appear for and in what form your name should come up. For example: “Our company name comes up in answers to people looking for a web design firm.”
Stopping at the diagnosis
Receiving the diagnosis results changes nothing unless the fixes are actually carried out. After the diagnosis, decide who will fix what, and by when.
The website's foundations are not in place
If there are foundational problems such as the site not being registered with search engines, slow loading, or being hard to read on a smartphone, LLMO work is less likely to produce results. Google Search Central explains that there are no special technical requirements for appearing in Google's AI features. Basic search optimization comes first.
How to work together after signing
Once you have signed, settling the following three things in the first month keeps the work moving.
- Appoint one point of contact: someone who can answer the provider's questions quickly
- Confirm who has permission to edit the website. If nobody can log in to the admin screen, copy revisions will stall
- Set a reporting day. Using the same day each month makes it easier to compare recorded changes
Do not leave everything to the provider. Read each report and ask about anything unclear on the spot. AI answers change day by day, so discussing them while looking at the figures and the screens leads to faster decisions.
How to read the diagnosis results
When you commission an LLMO diagnosis, you receive a list of the results of questions put to AI tools. Here is how to read those results.
First, check whether your company name appeared. If it did, note which question it appeared for and in what context. If it did not, look at the companies that are named in response to the same question and examine what is written on their pages.
Next, check whether the information given is correct. If the address, pricing or service details are outdated or wrong, find the page that is the source of that information and fix it. Company information that conflicts between several places is a common cause.
Then look at the pages cited as sources. Check whether your own pages are cited and whether other sites carry information about your company. If information on another site is outdated, you will need to ask that site's operator to correct it.
Diagnosis results cannot be immediately reduced to a score or a ranking. Because answers differ slightly each time, it is normal to enter the same question several times and look at the overall tendency. Check that the diagnostic report states the number of questions, the number of times they were entered, and which AI tools were used.
After reading the results, decide the order of fixes. Inconsistencies in company information can be fixed quickly and have a large impact, so start there. Next, revise the wording of the service pages closest to generating inquiries.
What to ask at the first meeting
At the first meeting, in addition to the 10 checkpoints, asking the following three things reveals the provider's attitude. First, what industries have they supported in the past? If they have experience in an industry close to yours, their question design will be more concrete. Second, what engagements did not go well, and why? A provider who can talk about failures will give you an honest outlook. Third, which parts would be better handled in-house? A provider who does not try to win every piece of the work and tells you what you can do yourself is trustworthy. It also helps to ask to see sample diagnostic reports and sample progress reports.
Copying the answers into the same table for each provider and lining them up reveals differences that the quoted price alone does not show.
Giving the table four columns—scope of diagnosis, reporting frequency, contract term, and cancellation conditions—makes comparison easier.
FAQ
Not necessarily. Unifying your basic company information and rewriting your copy to answer questions directly can be done in-house. One approach is to make progress yourself first and outsource only what is still missing.
It depends on when the AI references the information and on the condition of your pages. No fixed number of days can be given. Ask your provider how and how often results will be checked.
It varies by provider. One option is to commission a one-off diagnosis first and then decide whether to continue based on the results.
The company that built your site understands its structure, so implementation may go more smoothly. For the diagnosis and the proposal, it is fine to compare with other firms before deciding.
There is no need to commission them separately. The basic work overlaps. Check whether the provider will look at both and make proposals accordingly.
We accept consultations on AI search optimization
Merci Ltd. builds structuring optimized for SEO, GEO and LLMO into every new website as standard. We also accept consultations on diagnosing and improving websites that are already live. We will answer all 10 checkpoints in concrete terms.
