# Inside Lunasol's GEO and AEO Service for UAE Businesses

By Tamara Karmiczewska


This article is an advertisement for the company that owns this publication. AI Visibility is owned by Lunasol, a Dubai agency, and what follows is a description of Lunasol's own service, written for the owner by a writer who is not independent of it. Every other article in this series is a guide with a promotional section fenced off at the end; this one is the promotional section, labelled as such here so you can decide now whether to keep reading. No price appears anywhere in it, at any level, and every statement about what nobody can achieve is quoted from Google or OpenAI with a link.

Lunasol's GEO and AEO service tracks a fixed list of customer questions across the assistants your buyers use, and records what those assistants say about you and about a named list of competitors. It then builds and fixes the things on your site that carry the facts an assistant reads: the facts themselves, the pages, the markup and the answer blocks. It repeats the measurement on a set cycle so the second reading can be compared with the first. Three levels differ by seven variables, and the quantities below are the scope as currently documented. They are a description, not an offer and not a quotation: the figures that bind any engagement are the ones written into your own proposal. The work is bought by the month, one domain, UAE market, English, with a second language priced separately.

Who this GEO and AEO service is for, and who it is not for

It is for a business in Dubai or elsewhere in the UAE whose customers ask an assistant before they ask a person. Our own service page puts the audience in one line: "Your customers ask ChatGPT, and they ask Google" (Lunasol (https://lunasol.ae/geo-aeo)). That means a business selling something considered rather than impulsive, where a buyer researches before they contact anybody, and where being described wrongly costs more than being described late.

It fits best when three things are true of you. You have a website somebody can change. You can name the competitors you accept being measured against, or you are willing to settle that list in the first conversation and then leave it alone. And you want the reading repeated, because a single measurement of an AI answer tells you very little, as our article on checking what AI says about your company sets out at length.

It is a poor fit in three situations, and saying so is cheaper for both of us than finding out in month three. The first is edit access. Where the website belongs to a franchise, a portal or a head office that will not grant it, that is the thing to settle before anything else, because the site work cannot be done without it. The second is timing. Month one is a baseline by design, the measurement everything after it gets compared against, so it is not where a result shows up. The third is the work itself. Running social channels or ad accounts is a different job and this is not it.

There is also a whole category of business that can run a real share of this in-house, and our article on GEO in-house or with an agency is the honest place to start. If you have somebody who can edit the website, an hour a week and the patience to keep asking the same questions the same way, a good deal of this is yours to do. That article tells you how and it names the nine jobs. This service does that work for you, on a set cycle, at the volumes in the table below and within the boundary the exclusions section sets.

What does Lunasol's GEO and AEO service actually cover?

Two disciplines, sold together, because they run on the same underlying work. GEO, generative engine optimisation, aims at how your business is named and described when an assistant answers a question. AEO, answer engine optimisation, aims at the direct-answer surfaces, the boxes and overviews that sit above the blue links. Our service page carries both under their own taglines, "Generative Engine Optimization" and "Answer Engine Optimization" (Lunasol (https://lunasol.ae/geo-aeo)). Two words recur below. Schema is a standard way of tagging a page's code so that the page states, in a fixed format, what it is and what is on it. An entity, in this field, means the business as a named thing: the name, what it does, where it is, and what it is connected to.

The twelve deliverables are published on that page, and they are reproduced below exactly as they appear there rather than rewritten for this article, so you can check this description against the page it came from. Read them as the names of activities we carry out. What each one amounts to in quantity is in the sections after this one, and the section on what no scope buys sets out what the platforms themselves publish about their own surfaces.

The GEO half covers an "AI visibility audit, what ChatGPT says about you today"; "Entity and brand signal building"; "Content structured for AI citation"; "Presence in sources AI models trust and quote"; "Schema and structured data implementation"; and "Quarterly AI visibility tracking".

The AEO half covers "Question research, what your customers actually ask"; "FAQ and answer-block content structure"; "Schema markup, FAQ, HowTo, and more"; "Featured snippet targeting for money queries"; "Optimization for Google AI Overviews"; and "Tracking of answer placements won". One of those names, "Optimization for Google AI Overviews", is the work of making a page clear, correct and well structured, so that it is easy to read and easy to quote. What Google itself publishes about those surfaces is quoted in the section on what no scope buys.

Those are names. Where one of them points at work off your own site, the outreach row in the table below is where that work is counted. The rest of this article is the part the names do not carry: how much of each, on what cycle, and what is not in the box.

The seven variables that set a GEO scope, answered with our numbers

Our article on what determines the scope of a GEO project sets out seven variables that decide the size of any engagement of this kind, and argues that a quote is a scope rather than a price. It also says, in as many words, that a quote without numbers in it is not a quote, it is an unscoped estimate. We publish that, so here are all seven, answered, with our own numbers in them.

One, how many questions get tracked. This service tracks a fixed list of customer questions, the prompts your buyers actually type: 25 of them at Essential, 100 at Growth and 300 at Authority. The list is fixed at the start and then held, because a list that changes quietly makes the second measurement incomparable with the first.

Two, how many competitors you are scored against. A named list of 5, 10 or 15 businesses, agreed at the same point as the question list.

Three, how many answer surfaces are checked. Four, five or six, and they are named rather than counted: ChatGPT, Gemini, Perplexity and Google's AI Overviews at the first level; Copilot added at the second; Claude added at the third.

Four, how many languages. One. The service as scoped is English, one domain, UAE market. A second language is a second job rather than a translation line, for the reasons our article on English and Arabic AI search works through, and it is priced separately.

Five, how much of the site gets changed. Two kinds of work, counted separately. New pages, written and structured to be easy to quote: 4 a month at Essential, 12 at Growth, and a figure set in your proposal at Authority. Existing pages reworked: 2, 4 or 12 a month depending on the level.

Six, who implements. We do. Schema goes into your templates, content goes into your CMS, and the markup is ours to build rather than yours to commission afterwards. This is the variable that decides whether anything changes on your site at all. The volumes and the timetable are in your written agreement.

Seven, how often it repeats. A baseline in month one, then the full re-reading against that baseline every quarter, which is what our published list calls "Quarterly AI visibility tracking". Reporting runs from a report and a call at the first level to a dashboard, an executive report, calls every two weeks and a quarterly review at the third.

Those seven are the questions worth having answered in writing, here or anywhere.

Essential, Growth and Authority: what separates the three levels

Three levels, sold as three packages, Essential, Growth and Authority, in one table of eight rows. The disciplines are the same at every level, so what changes is volume, plus three additions at Authority named under the table. The rows below describe the service as documented on 21 September 2026. Your signed proposal is what binds.

What is counted
Essential
Growth
Authority

Customer questions tracked
25
100
300

Answer surfaces checked
4: ChatGPT, Gemini, Perplexity, Google AI Overviews
5: the above plus Copilot
6: the above plus Claude

Competitors you are scored against
5
10
15

New pages written each month
4
12
set in your proposal

Existing pages reworked each month
2
4
12

Who implements the content, schema and markup
Lunasol
Lunasol
Lunasol

Outreach attempts each month
not included
up to 5
up to 12

Reporting you receive
a report and a thirty-minute call
a dashboard, a report and a call
a dashboard, an executive report, calls every two weeks and a quarterly review

Five notes on that table, because the row headings do not carry all of it, and then the two counts that decide which column is yours.

What counts as one new page. This is the definition that makes the content count mean something, and it is worth reading twice. One item is a new page of up to roughly 1,200 words, and it is scoped to cover the research behind it, an AI draft, an edit, a fact check, the meta title and description, the internal linking, the upload into your CMS and the schema on the page. Where in that process you sign off is set in your agreement. Small edits to an existing FAQ, a meta description or a block of markup do not count as an item, and we draw that line on purpose: a month of small fixes is not counted to you as content delivered.

What Authority adds beyond volume. Three things that are not on the table because they are not quantities. A brand-truth base, which is the written record of what is true about your business that everything else is checked against. Advanced schema beyond the standard set. And work on AI errors, which is the job of chasing down a wrong fact an assistant is repeating about you and fixing the source it came from.

The Authority page count. The new-page figure at the Authority level is set in your proposal rather than here, and no figure is implied by its position in the column.

The content mix at Growth. At that level, at most one of the twelve pages a month is a comparison, alternatives or best-of piece.

And the boundary around all three. One domain. The UAE market. English. Everything outside that boundary is priced separately and named in the exclusions section below.

A reminder, because this is the part people share: this is an advertisement. AI Visibility is owned by Lunasol and the service described above is Lunasol's own.

Which level fits is usually settled by two counts. How many distinct services or products a customer might ask about, and how many competitors you would accept being measured against. A single service with four obvious rivals does not need three hundred tracked questions, and twenty-five will not cover a group with six service lines across two emirates. Bring both counts to the first conversation and the level usually picks itself.

What happens in the first month, and every month after

Our service page names four steps: Audit, "The AI census"; Signals, "Build what AI trusts"; Content, "Write the way AI reads"; and Track, "Numbers, every quarter" (Lunasol (https://lunasol.ae/geo-aeo)). Here is what each of those means in working time.

The question list and the competitor set. The fixed list of customer questions and the named competitor set are settled at the start of the engagement and then held fixed, because the comparison only holds if the list underneath it stayed still.

The audit. This is the step our published list calls "AI visibility audit, what ChatGPT says about you today". It produces the baseline: a record of what the assistants say about you and about your named competitors, taken at the start and compared against at every re-reading.

The site work. Company facts checked and made consistent wherever they appear, the job our article on whether your website clearly explains who you are and what you do sets out. It is duller than it sounds and it matters more than most of what follows. Schema built into your templates. Pages rewritten and new ones drafted, to the standard our article on writing service pages that answer real customer questions describes.

Every month after that. New pages and reworked pages at the volumes your proposal sets for your level. Outreach attempts at the levels that include them. Reporting at the cadence your level sets.

Every quarter. The full re-reading against the frozen baseline. This comparison is the point of the whole arrangement. Our service page calls this step "Numbers, every quarter" and that is exactly what it is.

What you are handed: the audit, the reporting and the quarterly re-reading

Our article on what an AI search audit should include lists eight things any audit ought to produce, whoever runs it. This service is built to meet that standard.

What you are handed: the audit output, the reporting artefact your level carries, and the quarterly re-reading against the baseline in the same format, so the two can be put side by side.

The audit output lands inside the first month, the reporting on the cadence your level sets, and the re-reading every quarter. What form each of those takes, who owns the content produced and what you keep at the end are set out in your written agreement. That part is worth reading closely: a frozen question list is no use to you if you cannot keep using it afterwards.

What you supply, and what happens if you cannot

What the engagement assumes, and what is settled in the first conversation.

Edit access to the website, or somebody who will act on instructions promptly. This is the one that decides whether the engagement works. If every change queues behind an external developer, the site work will not happen at the pace the table describes, and that is worth agreeing at the start rather than discovering later.

Verified ownership of your Search Console property, held by somebody who is not leaving. Google's rule is that "A property must have at least one verified owner, or no users will have access to the property", and that "You must be a property owner (or an owner of a parent property) to add or remove another user" (Search Console Help (https://support.google.com/webmasters/answer/7687615)). If your only verified owner is a developer who left, that gets solved before anything else does.

Analytics access, at the lowest level that lets the reporting be produced. What level of access the work needs, who holds it, how long anything is held, who else processes it including any reporting tool your level includes, and when access is handed back at the end all belong in your written agreement.

Decisions. Somebody in your business needs the authority to sign off the question list and the competitor set, and to say no when a colleague wants them reopened in month four.

What is not included, and what is priced separately

A package at any of the three levels buys work on your own property and a capped number of outreach attempts. Everything outside that is either excluded or quoted separately, and both lists are below.

Not included at any level, and not sold by us as part of this service: digital PR, publisher fees, and sponsored articles. No placement in any AI answer is sold or committed to at any level, which is not a commercial choice but a fact about those surfaces, and the section below covers it. No placement in any publication is committed to either, because that decision belongs to the publisher.

Priced separately: extra content items beyond the monthly count, extra questions beyond the tracked list, an additional language or an additional market, extra outreach attempts beyond the cap, Arabic content, digital PR, and development work. Where a publication charges for placement, that is the publisher's fee plus our fee, and it is worth asking how the publisher will label it before anything is commissioned.

What an outreach attempt is. Outreach at the levels that include it is counted in attempts and capped, so the most you are buying is a number you can see before you sign. Each attempt is a real piece of work. What a publisher then decides is the publisher's.

One more, and it comes from our own evidence rather than our own document. When we asked an AI assistant what a GEO agency does for a Dubai business, one of the five activities it listed was executing "targeted digital PR and secure mentions in trusted regional publications, industry directories, and business indexes". That is a real activity. It sits outside the monthly package by design, and where you want digital PR it is quoted separately, as the list above says. We are telling you that because you would otherwise find it out from an assistant and wonder why it was missing from our list.

What a buyer is told when they ask an assistant what a GEO agency does

We ran two questions through Google's AI Mode on one day, under a protocol written and frozen before either run, and we are reporting what came back whether or not it flatters us. One run each, one tool, one day. An illustration of what a buyer is told, not a survey of anything.

The first answer was confident, well organised, and contained no numbers at all. Asked what a GEO agency actually does for a business in Dubai, the assistant produced five activity groups, each split into an explanation and a line beginning "What the agency does": schema and entity work, restructuring content for language models, local and bilingual signals, off-site authority and digital PR, and tracking what it called "citation share". It distinguished carefully between what the models do and what an agency does, which is better than the pricing answers we logged for our article on what determines the scope of a GEO project. And across the whole answer there was not one quantity: no count of questions, surfaces, competitors, pages or attempts, no cadence and no timeframe.

It cited thirty-two outbound links across thirteen hosts, and not one of those hosts was a search engine's or an assistant maker's own documentation. We read the counts from the rendered page and opened none of them, so this describes what was cited rather than what those pages say. Two of the answer's statements carried no citation in the answer, and neither is supported by anything in this article's sources. They may be true. We do not know, and we are not going to repeat them here as though we did.

The second question, asked with no marketing context, was answered as a geography question. The assistant explained that the acronym "typically refers to a Geography (or Geospatial) Research Proposal" and gave eight sections of an academic research proposal. That was not an error or a truncation. It was a complete answer to a different question, and it is reported here as such.

That second run is the more useful of the two. The label is not settled. A buyer typing the acronym without context is not reliably shown this industry at all. So do not buy anything on the strength of a three-letter word. Ask a supplier for counted rows instead, of the kind set out in the table above.

What no GEO scope can buy, and what we say on day one

Here is the thing we tell every client in the first conversation, before anybody signs anything.

Nobody can make an assistant recommend your business, at any level and at any budget. That is not a statement about this service; it is what the platforms publish about their own surfaces, and here it is in their words. Google states plainly that "No one can guarantee a #1 ranking on Google" and lists among its warning signs providers who "claim to guarantee rankings, allege a 'special relationship' with Google, or advertise a 'priority submit' to Google" (Google Search Central (https://developers.google.com/search/docs/fundamentals/do-i-need-seo)). For its own AI surfaces Google says that "There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary" and that "You don't need to create new machine readable files, AI text files, or markup to appear in these features" (Google Search Central (https://developers.google.com/search/docs/appearance/ai-features)). That is why no scope here buys placement, and why the schema and the answer blocks are not sold as a way in: they are the work of making a page clear, correct and well structured, so that it is easy to read and easy to quote. OpenAI states that "Placement is not guaranteed" and that "Search results and citations can be incomplete, outdated, or incorrect" (OpenAI Help Center (https://help.openai.com/en/articles/9237897-chatgpt-search)). Our article on whether anyone can guarantee a ChatGPT recommendation goes through the whole position.

A word on speed, because it is the question everybody asks and the published sources say less about it than people expect. Google's own guidance on its search documentation says only that "Some changes might take effect in a few hours, others could take several months" (Google Search Central (https://developers.google.com/search/docs/fundamentals/seo-starter-guide)). Search Console says of its own data that "Normally, however, collected data should be available in 2-3 days" (Search Console Help (https://support.google.com/webmasters/answer/96568)). No published source gives a date for an AI answer.

The reporting has limits too, and they are published ones. Search Console says "To protect user privacy, the Performance report doesn't show all data" (Search Console Help (https://support.google.com/webmasters/answer/96568)). Analytics only looks back so far, with a thirty-day default for acquisition key events and ninety for the rest (Google Analytics Help (https://support.google.com/analytics/answer/10597962)). And a single reading tells you about that run and nothing more.

What you are buying is a quantity of work on the things that are yours to change, measured the same way each time so you can see whether it moved. Every row in the table above counts work done, not results achieved.

Common questions

How do I know which level I need? The two counts that settle it are set out under the table above: your services, and the rivals you accept being measured against. Bring both numbers to the first conversation rather than a guess at a level.

Can I start at one level and move? The thing worth knowing before you choose is that a question list that grows keeps its comparison with your baseline, and one that shrinks does not. How a change of level is handled is settled in your written proposal.

What if I already have an SEO agency? Read the scope you signed against the seven variables in our article on what determines the scope of a GEO project, and see which of them your existing contract answers. A good deal of this work overlaps with technical SEO, and our article on the difference between SEO, AEO and GEO sets out where. If it is already in a scope you are paying for, do not buy it twice. Any variables it does not answer are the ones worth bringing to a first conversation.

Do you write the content with AI? The definition of a content item above says so in as many words: research, an AI draft, an edit, a fact check, then meta, linking, upload and schema. We publish that plainly, for the same reason this whole article publishes its quantities. What you are paying for is the research, the editing, the fact checking and the structure, which is where the time goes and where a page becomes quotable or does not. The fact check is a step in the process, not a warranty of accuracy, and your sign-off comes before anything goes onto your domain.

What happens if an assistant says something false about my business? The audit records what the assistants say about you, so a false fact shows up in that record. Chasing a wrong fact down and fixing the source it came from is named work at the Authority level; the fix is applied to the source rather than to the assistant, because the source is the only part anybody can change. What the platforms publish about their own outputs is quoted in the section above.

How long before I see something? You will see the baseline inside the first month, because that is a measurement of what exists today rather than a result. Anything beyond that depends on surfaces nobody controls, and the section above quotes Google and OpenAI rather than guessing on their behalf.

What is the shortest sensible commitment? Long enough to produce a second reading you can compare with the first. One baseline on its own tells you where you started and nothing else.

How to start, and what the first conversation covers

One message. Message us on WhatsApp or send an email, both on lunasol.ae/geo-aeo (https://lunasol.ae/geo-aeo). There is no form to fill in and no gate to get past.

Send four things if you have them, and say so if you do not, because they can be built in the first conversation instead.

Ten to twenty questions a customer might type, in their words.

Three to five competitors you accept being measured against.

Which assistants your buyers use, if you know.

Whether you need one language or two.

Those four decide the level and most of the scope, and the first conversation is built around them. If you send nothing else, send those.

The first conversation covers the same ground as this article, in your specifics rather than in general: the question list, the competitor set, the surfaces, which pages are in scope, who holds the access, and which of the three levels matches what you just described. A confidentiality undertaking can be put in writing before you send anything; ask for one. The proposal follows, with your own figures in it.

Method, sources and notes

What this article is. An advertisement for Lunasol's own GEO and AEO service, published by a publication Lunasol owns, and labelled as such in its first line and again here. It is general information about a service and not a quotation, an offer or a contract. The scope quantities describe the service as currently documented, they may change, and any engagement would be governed by a separate written agreement in which the figures quoted to you are the ones that bind. Nothing here is an offer capable of acceptance, and no engagement arises until a written agreement is signed by both parties. Nothing here is legal, tax, financial or other professional advice. The quantities in this article describe the service as documented on 21 September 2026. Whether monthly quantities carry forward is set in your agreement, as is what happens when the access, approvals and decisions listed under what you supply are not available when needed. Readers act on this at their own discretion. To the fullest extent permitted by applicable law, neither this publication nor its owner accepts liability for any decision taken on the basis of it, and nothing here excludes or limits any liability that cannot lawfully be excluded or limited.

Who publishes this. AI Visibility is owned by Lunasol. The author writes for this publication and is not independent of its owner. This article makes no claim about any other provider, named or unnamed, in any direction: no comparison, no characterisation, no market description. Where it refers to how proposals in this field are written, it does so in general terms and about no identified provider. Where the live check produced claims about what agencies do as a category, we reported that it produced them and, in the two cases where they carried no source, declined to adopt them. Questions about this article can be sent to the contact address on lunasol.ae/geo-aeo.

On the numbers. The quantities in this article come from our own internal service documentation rather than from the public service page, which publishes deliverable names and no figures. Every quantity here is stated conservatively. These quantities are the service's current published scope, not a guarantee or a minimum, and your written proposal is what confirms them. One row, the new-page count at the Authority level, carries no figure at all.

No price appears anywhere in this article, in any currency, at any level, including for the items listed as priced separately. That is deliberate. What a piece of this work costs depends on how much of it there is, and our article on what determines the scope of a GEO project sets out the seven variables that decide that. Nothing here should be read as a price, a rate, a range or a guide to one, and none can be inferred from it.

No client is named, no case study appears, no result is claimed and no testimonial is reproduced in this article.

The live check. Two questions, one run each, one tool, one day, one connection, in Google's AI Mode, under a protocol frozen before either run. Neither run failed. The second run answered a different question from the one asked, which is reported in full rather than discarded. Link counts were read from the rendered page and we opened no cited page, so what we say about sources describes what was cited and not what those pages say. We name no company, agency, publication, product, host or domain from either answer. Assistant answers vary by account, location and time, and quoted documentation changes without notice, so check the current version of any page quoted here before relying on it.

Sources. The Lunasol GEO and AEO page, quoted only for what it publishes; Google Search Central, "Do you need an SEO?", the "SEO Starter Guide" and "AI features and your website"; Search Console Help on managing owners, users and permissions and on Search Console data; Google Analytics Help on attribution settings; and the OpenAI Help Center on searching the web with ChatGPT.

Nothing here makes a business shown, named, cited or recommended in any AI product, and nobody can promise that it will. What the platforms themselves publish about their own surfaces is quoted above, in their own words.

Product and company names are used here for identification only. Neither this publication nor its owner is affiliated with, endorsed by or sponsored by any of them, nothing here is an endorsement by us of any of them, and none of them has reviewed, approved or contributed to this article. ChatGPT is a product of OpenAI, and the OpenAI Help Center is a publication of OpenAI. Google Search, Google Search Central, Google Search Console, Google Analytics, Gemini, AI Mode and AI Overviews are products or publications of Google. Copilot is a product of Microsoft. Claude is a product of Anthropic. Perplexity is a product of Perplexity AI. All are trademarks of their respective owners.

AI Visibility is owned by Lunasol.