LLM Brand Visibility
Models write a name into the answer. That name can be the legal line you use now, an alias you still own, a product you killed, or a firm you merged last year. We sample that string. We list aliases that are still you. We write a path for names that are not. We do not treat a footnote as this job. We do not treat on-site Organization markup as this job. We do not treat mention hunting as the whole H1.
Position
Buyers paste a prompt. The paragraph that comes back may use last year's trading name, a product line you sold, or a parent you no longer put on the door. Search intent for the live brand still has to match the string in that paragraph. This page works the spoken name, not the cite, not the public panel, and not the Organization block on your site.
AI citation optimization owns eligibility of the cite object. Use this page when the prose uses the wrong name even if no URL is attached. Use that page when the name is fine and the attach is the problem.
Open AI citation optimizationEntity SEO specifies official name, identifiers, markup, and sameAs on your site. A schema H1 mismatch is that page. A model that still says the old name after the site is clean is this page.
Open Entity SEOKnowledge Graph optimization documents what Google already shows in a panel. Assistants can speak a retired name the panel never lists. Panel tickets stay there.
Open Knowledge Graph optimizationAI entity optimization disambiguates two companies the model has merged. A wrong spoken string that is still clearly you is this job. A merge of two legal entities is that job.
Open AI entity optimizationAI entity optimization asks whether the model bound two companies as one node. Entity SEO writes the identity spec on pages you control. Knowledge Graph work starts from facts the panel already shows. AI citation work starts from a claim and a URL. Mention work starts from an empty or wrong attach. This URL is for the string in the prose: official name, alias, misspelling, or retired mark.
We will not sell a share of voice score as the product. We will not treat every surface as this job. We will not correct Wikipedia as a ranking lever from this URL. If legal cannot name the current line, we stop before we sample. If the sample shows the right name and the wrong URL attached, we hand that to citation work.
When this fits
This page fits when you can name the official line and you already see assistants using something else. If you needed markup on the site, a panel pass, or a cite object, start there.
A former trading name, a merged brand, or a product you killed still appears in the prose. We document the current official name and which aliases remain in force. Keyword targeting the old name in titles will not retire it in answers.
An abbreviation you love is also another firm's ticker. Topical relevance of your blog is not a reason to keep shouting the collision. We list which aliases may still be spoken as you.
Hyphen, space, Inc, Ltd. Models pick one. Semantic keywords will not standardize the string. We pick the spoken form you will stand behind and sample against it.
We will say so. Sampling the name line can still be useful. A purchased mention share is not a deliverable. If the unknown is unlinked mentions with no name-line question, that is brand mention optimization.
Process
Four jobs, in that order. Each job produces a note brand can keep. We do not start a cite hunt in the last job. The point is the spoken string, not a footnote package.
We write the official name you will stand behind in answers. Search intent for the live brand has to use that string. A campaign slogan is not the name. A product SKU is not the name unless the question is about that SKU.
Outcome A dated official name line, including what is retired.
We list aliases, abbreviations, and former names that are still you, and those that are not. Topical relevance is not a reason to keep a colliding acronym. We mark which strings may appear in answers as you.
Outcome An alias table: keep, retire, collision.
We run a small set of questions a buyer would ask. We record the exact name string in the prose. We do not score citations here. A URL in a footnote is a side note. The object is the spoken line.
Outcome A name transcript: question, date, string used.
We write what must change on pages and profiles you control so the spoken line can catch up, and what this page cannot force inside a model. Recheck uses the same questions. Entity binding work is a handoff if two firms are fused.
Outcome An alignment note and a recheck date, not a visibility quota.
Deliverables
You leave with a name line, an alias table, a transcript, and an alignment note. You do not leave with a citation package, a panel edit, or a ranking date.
The string you want spoken. Dated. Last year's brand book is not this line if legal moved.
Keep, retire, collision. Abbreviations that match another firm are named.
Questions, dates, and the exact string in the prose. A single screenshot is not a transcript.
Pages and profiles that still teach the retired name, plus what this engagement will not attempt inside a closed model.
When to rerun the same questions. A weekly share-of-voice chart is not a recheck. A promise of Overview placement is not a recheck.
Benefits
Brand and legal agree the line. Sampling stops arguing with a slogan.
An acronym that is also a ticker is documented. Teams stop cheering it in prompts.
You see the words, not a blended visibility score that mixes cites and panels.
Someone reruns the questions. Nobody pretends a schema plugin retired a name in answers.
Methodology
This works when legal can confirm the name, and someone can change pages that still teach the old string. If both are missing, we are collecting pretty screenshots. We are not correcting the name line.
We start with the official name, the alias list, and a person who can edit About and product pages. If any of those is missing, we name it first.
We sample published answers. We do not trust a brand deck. Does this paragraph use the name you keep? Is a retired mark still the first noun?
You may rename the roles below. You may not leave the spoken line implied. An implied alias is how two brands share a sentence.
First working session
The first hour is the official name versus the name the assistant just used. It is not a tour of citation products.
We write the official line. We paste one buyer question. We record the name in the answer. If they disagree, that is the job. If two companies are fused in the model, that is AI entity optimization. If Organization markup is the gap, that is Entity SEO. If a panel is the gap, that is Knowledge Graph optimization. If a cite object is the gap, that is AI citation optimization.
Then we list aliases. We mark collisions. The extra or missing strings are usually the job. A Search Everywhere program is not this hour.
We leave with a first name transcript or a stop. A transcript names the string, the question, and who can change official copy. A stop names what is missing: legal sign-off, a job that is actually a merge in the model, or a job that is actually a cite. Both are outcomes. A dashboard shopping list is not.
AI search
A stable name line can help later extraction. This page cannot sell AI Overviews, a chat mention as a trophy, or LLM visibility as a blended score.
People ask whether fixing the spoken name will land the brand in Google AI Overviews (AIO). If the prose still uses a retired mark, generative search has a stale noun to lift. That is ours to tidy as a name line. If the name is already right and models still invent a cite, that is citation work. If they fuse you with a namesake, that is AI entity optimization.
AI search will lift a noun. A slogan is a poor noun. We do not sell Overview presence.
We check the string in the prose. The cite object lives on another URL.
AI entity optimization owns model-side disambiguation. This page owns the spoken line when the firm is already distinct.
Schema
Organization JSON-LD helps a machine read a name the page already states. Depth of identity markup stays on Entity SEO. This page can flag a spoken sample that ignores your markup. It is not a schema campaign as the whole job.
If schema says one legal name and answers say another, we still sample answers here. The spec to align the site lives on Entity SEO.
Dead marks in markup keep teaching the old string. We list which aliases may stay. Fake directories do not go in the array.
Markup will not force generative search to use your current brand. Eligibility of a cite is a different object.
Who we work with
Someone has to sign the official name. We work with that person. We do not replace them with a second brand-visibility retainer.
Confirms the name and aliases. Holds the line when marketing wants a slogan in the sample.
Can retire the old string in copy you control. They do not need to become Wikidata editors.
Can keep the question list. If they try to run this as unlinked mention cleanup, two jobs collide. Mentions stay on brand mention optimization.
Questions
The brand string models write into the answer: the current official name versus a retired legal name, an old trading mark, or a product you killed. A correct name with no footnote is still this job. A footnote on a rival URL is citation work.
No. Entity SEO specifies on-site identity: official name, identifiers, markup, and sameAs you will maintain. This page samples model answers and checks the spoken line. A schema mismatch on your site is that URL. A retired name still spoken by an assistant is this one.
No. Brand mention optimization hunts names with no source or the wrong source. This page is narrower: which name the model used. Mentions can be a later door. They are not the headline here.
No. Knowledge Graph optimization documents public panel facts people already see. Assistants can speak a name the panel never shows. We will not screenshot a panel as the deliverable here.
No. Accurate spoken names can help later. This work does not buy a place in Google AI Overviews (AIO), a generative-search mention, or a citation score. If models still attach the wrong URL after the name is right, that measurement sits on AI citation optimization.
The string in the prose
Write the current legal name, the retired names that must not appear, and two answer samples if you have them. We will say if the spoken string is stale, merged, or a namesake, and whether citation, model binding, or on-site identity should go first.
Current, leftover, transcript, align. Not a mention desk.
