How Do You Choose a Semantic SEO Agency?
A semantic SEO agency is chosen on three written proofs: a page decision per URL with the reason, a vocabulary document fixed before writing starts, and measurement read per template with pipeline included where a CRM exists. An agency that supplies all three before the contract is a semantic SEO agency. An agency that supplies none is a content agency with a new name.


What three proofs separate a semantic SEO agency from a keyword agency?
Three documents separate the two, and each one exists before the first page is written.
The first is a page decision per URL. Every query variation on the site has a recorded verdict: page, section or sentence, with the reason. The decision is the output of the query deserves a page test. A keyword agency has a keyword list instead, with volumes and no decisions.
The second is a vocabulary document. One name per thing, one verb per relationship, recorded before writing and checked on every page. The document is the source term vector. A keyword agency rotates synonyms for variety and calls it natural writing.
The third is measurement per template. Clicks and query count are read per page type in Google Search Console, and leads are read per cluster where a CRM exists. A keyword agency tracks positions for a keyword set and reports those.
Ask for all three on the first call. An agency doing the work has them in a folder.
Which twelve questions does a buyer ask on the first call?
Twelve questions cover the method, the proof and the contract, and each one has an answer that passes and an answer that fails.
Twelve questions for the first call, each with the answer that passes, the answer that fails, and the document that proves it.
| Question | Answer that passes | Answer that fails | Document that proves it |
|---|---|---|---|
| What is my site about, in one sentence? | The agency names the central entity, the type, the audience and the market in one sentence, on the call. | The agency asks for the keyword list first. | Central entity definition |
| Which of my pages compete with each other? | Two or more URLs named, with the query they split and the merge target. | More pages recommended. | Page decision per URL |
| How do you decide if a query gets its own page? | A named test with checks: demand, entity, similarity, predicate, pattern. | Search volume alone. | Query deserves a page record |
| How is vocabulary kept consistent across writers? | A document with canonical terms, allowed aliases and banned variants, checked per page. | A style guide about tone. | Source term vector |
| What is in a content brief? | Entity, predicate, facts as triples, section order, component per section, links with anchors. | An outline of headings. | A sample brief |
| What order do pages publish in, and why? | A roadmap in dependency order, hardest high-trust topics first. | A calendar with a monthly count. | Publishing roadmap |
| How are internal links decided? | Outer to core, one canonical anchor per URL, recorded next to the URL. | Links added by the writer where they fit. | Internal linking architecture |
| How do you decide where a link from another site is worth placing? | The linking page shares the topic and passes the same page test as the client's own pages. | Domain rating targets and monthly counts. | Placement gate record |
| What do you report, and where does the number come from? | Query count per template and clicks per page from Search Console, leads per cluster from the CRM. | Positions for a keyword set. | A sample report |
| How does the site get cited by ChatGPT and AI Overviews? | Passages that state one fact each, a named author, consensus wording, next-step coverage, measured on a prompt set. | A separate GEO service with a tool subscription. | Prompt-set baseline |
| Who decides the strategy, by name? | One named person owns the map and the vocabulary; another owns what ships. | A team. | Team page |
| What do we own when the contract ends? | The audit, the architecture, the map, the vocabulary and the roadmap stay with the business. | Access ends with the contract. | Service agreement |
A buyer who hears ten passing answers has found a semantic SEO agency. A buyer who hears ten failing answers has found an activity report with a new name.
What does a written quote from a semantic SEO agency contain?
A written quote names the engagement shape, the value used for each scope factor, the deliverables as documents, and the pricing basis.
The shape comes first, because it decides the basis. An audit is a flat fee. An architecture is a fixed-scope project. Production is a monthly program. The factor values follow: site size as indexable page count, query network size, language count, who produces the pages, and the platform. Each value is checkable against the site.
The deliverables are named as documents or pages, never as hours. An architecture quote lists its seven documents and the handoff workshop. A production quote lists the roadmap segment each month covers. An audit quote lists the issue map and the review call.
Industry, competition and ambition do not appear as price factors. They change the strategy, not the cost of building the documents. A quote that charges more for a competitive niche is pricing fear, not scope. How semantic SEO is priced covers each factor.
How does a buyer check an agency's results before paying?
Results are checked in Google Search Console, with the window named, and then confirmed by a client who speaks.
A screenshot of a rank tracker is the agency's tool talking about the agency's keywords. A Search Console panel is Google's record of clicks, impressions, click-through rate and average position on a property. Ask to see panels. Ask for the window. A 16-month panel shows if the site held through core updates. A 28-day panel hides them.
Then ask to be introduced to a client in a comparable industry, with the agency not on the call. A client who agrees to speak is proof the agency cannot write for itself.
Anonymised panels are normal, because clients do not want competitors reading their traffic. On a call, a live property can be opened with the domain visible. Results, measured in Google Search Console shows the panel format, and client feedback shows reviews tied to the project and the date.
How is an AI search visibility agency vetted differently?
An AI search visibility agency is vetted on the same three proofs plus a prompt-set baseline, because citation is a structural outcome, not a separate service.
Buyers call the work GEO, AEO or LLM visibility. The name does not matter. The method does. Ask how the agency measures citation: a fixed prompt set run on each system, with a sentence-level record of who is cited, re-run monthly. Ask what the agency changes on the page when the record shows a gap: passage independence, one fact per sentence, a named author, consensus wording, coverage of the follow-up questions.
An agency that answers with a tool subscription and a share-of-voice number has a measurement, not a method. The number tells you where you stand. The work that moves the number is editorial and architectural. AI Search Visibility sets out the six requirements and the test for each.
Which answers end the conversation?
Five answers end the conversation, because each one describes an engagement that cannot produce topical authority.
- A guarantee of rankings or citations.Nobody controls what Google or a generative system chooses. A guarantee is a promise the agency cannot keep or a tactic that gets the site hit at the next update.
- A keyword list presented as the strategy.A list has volumes and no decisions. The decisions are the strategy.
- A report with no Search Console panel in it.The agency's tool is reporting the agency's work.
- No named owner of the map.A map owned by a team is owned by nobody, and the vocabulary drifts with every new hire.
- More content as the answer to a traffic drop.A drop after a core update is a structural fault. More pages on the same structure spread the fault.
Each of these is also a question you can ask directly. The answer tells you which kind of agency is on the call.
How does a buyer compare two semantic SEO agency quotes?
Two quotes are compared on the three proofs first and the number second, and only like for like.
Compare an audit with an audit and an architecture with an architecture. A flat-fee audit beside a monthly production program is two answers to two different questions. Inside each pair, check the three proofs: is there a page decision per URL, is the vocabulary fixed in a document, is measurement per template with pipeline where a CRM exists.
A quote that passes all three at a higher number buys a plan. A quote that fails all three at a lower number buys pages without one. The pages without a plan cost more over twelve months, because the merges they create are paid for later.
- Koray Tuğberk Gübür, Search Engine Land: Query templates, expanding the scope of topical authority
- Google Search Central: A guide to Google Search ranking systems
- Google Search Central: Creating helpful, reliable, people-first content
- US 2024/0289407 A1: Search with stateful chat (Google, pending application)
Patents and filings describe systems Google has invented. None is a confirmed ranking factor.
What should you read after how to choose a semantic SEO agency?
How much does semantic SEO costFive engagement shapes and five scope factors. What moves the price and what does not.
Semantic SEO vs traditional SEOWhat an engagement is built from: an entity and a topical map instead of a keyword list.
What is a topical mapA topical map is the plan of every entity, attribute, and query a site covers.
Query Deserves a PageThe test that decides if a query variation gets its own page, a section, or a sentence.
Semantic Content AuditSix checks. Each one has a defined failure and a defined fix.
Semantic SEO ArchitectureSeven documents with a fixed structure. Every engagement produces all seven.
Talk to usMessage the team on WhatsApp
Digital Vikingz