FAQ

The trial lasts 7 days on the Pulse plan and includes 500 credits. You add a payment card when you sign up. If you do not cancel before the trial ends, the paid Pulse plan starts after 7 days (€49 a month, excl. VAT). Credit top-ups are not available during the trial.

For companies whose customers ask language models before they choose a supplier or a product: B2B and B2C brands, online shops, agencies and organisations that want to know how models describe their field. Semantio shows whether a model names your brand on its own when it answers typical customer questions, which brands it puts next to yours, how strongly it recommends you and which sources it cites.

With a baseline measurement. You describe your brand: market and language, products, audiences, competitors and key facts. From this, Semantio prepares the questions customers ask at different stages of a decision and puts them to the models. The first result shows where you stand before any changes, and later measurements, on the same questions, show what is changing. The simplest way in is the 7-day trial on the Pulse plan.

Monitoring is self-service work in the app: you set up your brand and questions, and Semantio measures model answers regularly and shows the results in the dashboard and in reports. A study with an expert is run by the Brand Semantics team: they prepare the questions with you, interpret the results, set priorities for change and check the effects in later measurements. Monitoring suits ongoing observation; a study suits strategic decisions and changes to how the brand communicates. To discuss a study, visit brandsemantics.eu.

The plans differ in scale: the number of brands, questions per brand and channels, how often measurements run and which reports are available. Pulse (€49 a month, excl. VAT) is enough to follow one brand once a month across 4 channels. Pro (€199) covers 5 brands, 6 channels, measurements up to once a week, data export and a second report type. Business (€499) gives you 15 brands, all 9 channels, measurements up to once a day and roles for your team and clients. Enterprise has limits agreed individually and API access.

A credit is the unit in which the app accounts for measurements. Each plan has a monthly allowance: Pulse 1,500, Pro 6,000, Business 18,000 credits; for Enterprise the allowance is agreed individually. Reports draw on the same allowance. The credit cost of a question depends on the channel, so the number of questions a plan covers depends on which channels you choose and how many scenarios you run. You see the exact cost in the dashboard before each run.

Yes. You change your plan in the subscription settings. When you move to a lower plan, your account has to fit within its limits, for example the number of brands and scenarios. You can cancel at any time: access stays until the end of the paid period, and until then you can withdraw the cancellation.

As soon as the first measurement has run. You do not have to wait for the schedule: you can run selected scenarios manually in the dashboard, and results appear once the models have answered and the answers have been assessed. Your plan's schedule, from monthly to daily, takes care of the regular measurements that follow.

The basic facts about your brand: market and language, your offer and products, audience groups, main competitors and the information a model should get right, such as the scope of your services, locations or prices. This data is used to prepare the questions and to check the answers. The models being measured never receive it: they answer the questions as they would answer a customer.

Yes. Each client is a separate brand with its own questions, competitors and reports. Pro covers 5 brands and 1 additional team member; Business covers 15 brands, 5 team members and client roles with read-only access. An agency can run the measurements itself and share the results with a client without handing over control of the account.

Semantio prepares strategic reports: Semantic Pulse, a summary of where your brand stands in model answers, on every plan; Strategy Board, areas for action ordered by urgency, from Pro; Deep Dive, an in-depth analysis, from Business; and the Technical GEO Report on Enterprise. From Pro you can export the data to CSV and DOCX files.

On the Enterprise plan we provide a read-only API for results, so you can bring the data into your own analytics tools. On the Pro and Business plans you can move results through export to CSV and DOCX files.

No. Your brand description, facts and list of competitors are used to prepare the questions and to assess the answers. The model being measured receives just the question, as a customer would ask it, with no information about your brand. That way the result shows what the model tells customers, not what it repeats back to you.

Yes. For each brand you choose a market and a language, and the questions are written in the language local customers use. You compare results from different markets separately, because the same model can describe a brand one way in one language and another way in the next. In some markets certain channels are unavailable, and then they are not run.

Pulse costs €49, Pro €199 and Business €499 a month; prices exclude VAT. Enterprise is priced individually. On every plan you pay for a monthly credit allowance and the plan's limits. You can try Pulse first with a 7-day trial.

We do not steer model answers; we measure the effects of changes. Semantio shows what models say about your brand and where the gaps are: questions where competitors win, the sources models cite and information that conflicts with your brand facts. You make the changes to content and sources yourself or with Brand Semantics, and later measurements show whether they worked.

GEO is the work of shaping how a brand appears in answers generated by language models and by search engines that reply with ready-made text, such as ChatGPT, Gemini or Google AI Overview. In SEO you compete for a place in a list of links; in GEO you work on whether the model mentions the brand at all, which brands it puts next to it, whether it recommends it and which sources it cites. GEO starts with measurement, because without it you do not know what to improve.

LLMO (Large Language Model Optimization) is the work of organising knowledge about a brand so that language models understand it correctly: what the brand does, which category it belongs to and which facts apply to it. GEO takes a wider view, looking at whole answers, including in search engines that reply with ready-made text, and at how the brand compares with its competitors. In practice the two terms describe the same direction of work and are often used interchangeably.

No. Search engines still bring traffic, and many answers generated with search use pages found by a search engine. What changes is the measure of success: beyond rankings and clicks, it matters whether the model mentions and recommends the brand in an answer after which the customer often clicks nothing at all. That is why SEO and GEO are worth running together and measuring separately.

Not always. A model may know your brand from its training data differently from what current search results suggest, and with search it picks a few sources and builds one answer from them. A page ranked first can be left out, while a brand further down can make it into the answer if it fits the question better. The way to check is to measure the answers.

From two places. The first is the data the model was trained on: knowledge from the internet up to the end of training, fixed in the model. The second is search: some models and search engines check current pages before answering and cite them as sources. So what a model says about your brand depends both on what the internet has said about it for years and on what can be found about it today.

An answer from memory relies on what the model learned in training, so it may not know about recent changes to your offer. An answer with search combines that knowledge with current pages and usually cites sources. The same brand can come out differently in the two modes. From the Business plan, Semantio measures the search mode separately, so you can see whether the issue lies in the model's memory or in the sources available today.

They are two ways in which Google answers with ready-made text. Google AI Overview is a summary shown above the regular search results for some queries. Google AI Mode is a separate conversational mode of the search engine, with longer answers and follow-up questions. Semantio measures the two channels separately, because a brand can appear in one and not in the other. Which channels are available depends on the plan.

Because that is how a customer who has not yet decided asks: “which system for…”, “who in the UK does…”. If the brand name is in the question, the model will usually write about it, so the result tells you little. The real test is a question without the name: does the model bring up the brand on its own, and does it recommend it? Semantio prepares questions for different stages of a decision, and at the search stage the questions do not include the brand name.

No. A model can list the brand, describe it neutrally or present it as the first choice; it can also add a caveat that puts the customer off. That is why, beyond the mention itself, Semantio shows the tone of the answer and how strongly the model recommends the brand compared with competitors. For sales, what counts is the recommendation, not just being on the list.

First, establish the scale: is it a single answer or a pattern that repeats across several channels? Semantio flags possible conflicts with the facts you entered about your brand. The next step is the sources: completing and aligning the facts on your own website, in your brand profiles and on the sites models cite. Changes do not work overnight, so you check their effect in later measurements.

Models with search cite sources: company websites, industry portals, comparison sites, encyclopaedias and forums. Semantio shows which sources appear in answers about your category, whether your own domain is among them and how strong those domains are. That tells you where the facts about your brand should be present and up to date, so a model can find and use them.

They help organise knowledge about your brand. Structured data (schema.org) describes the company, products and people unambiguously, and the E-E-A-T principles (experience, expertise, authoritativeness, trustworthiness) strengthen content that search engines and models treat as reliable sources. They are not a switch, though: markup alone will not make a model recommend your brand. It is worth checking the effect of each such change with a measurement before and after.

Providers release new model versions regularly, and sources on the internet change, as does what appears in search results. The same brand can therefore come out differently a few weeks later without having changed anything itself. That is why measurement runs in cycles, on the same questions, and Semantio records the measurement date and the model version with every result. How often to measure depends on how fast your industry moves: from monthly to daily.

No. Some services, such as ChatGPT, show ads alongside answers. A paid placement is not the same as a recommendation from the model, so Semantio does not count ads as recommendations. We measure the content of the answer, that is, what the model itself says about the brand.

Like any change: a starting point, an action, a new measurement. First a baseline on a fixed set of questions, then changes to content and sources, then further measurements on the same questions in the same channels. You compare whether the model mentions the brand more often, how strongly it recommends it, which brands it puts next to it and which sources it cites. A single result can fluctuate, so conclusions come from several measurements, not one.

Because a model answers a question, not a demographic. A purchasing director comparing suppliers asks differently from a small business owner looking for a first solution. That is why a persona in Semantio is a role and the decision that person makes, paired with a specific product. That pairing produces the questions customers actually ask.