Measure your clients' sites in AI answers, one by one

Tasadr is a measurement tool, not an agency. We ask AI engines the questions your client's customers ask, in their words and their dialect, and record for every answer whether your client's site is cited with a link, mentioned without one, absent, or whether the engine wrote no answer at all. From those outcomes comes the Tasadr score out of 100, for each site on its own.

The work that raises that score is called optimising for AI answers (GEO): getting an engine to choose your client's site as a source when it writes its answer. The agency does that work; Tasadr measures where the site stands before and after it. So we built it for the people doing the work: the text of every answer as evidence, who got there ahead of your client, and what to change first.

SEO and digital marketing agencies in the Gulf now hear a new question from their clients: does ChatGPT mention us when someone asks? Some call this "AI visibility". A keyword ranking report cannot answer it, because what the customer reads is not a list of links but one paragraph that credits two or three sources. Tasadr gives you the answer written down and dated, question by question and engine by engine.

3 sites on Pro, 10 sites on Business. Each site is measured on its own, with its own score, questions and rivals, so one client's results never mix with another's, and you move between your clients' sites inside the same account.

The question balance belongs to the account, not to each site: 100 questions on Pro, 150 questions on Business, spread across your clients' sites as each contract needs. A large client takes a bigger share; a new client starts with a few questions and adds more after the first measurement.

Engines and cadence follow the plan, not the site. Pro measures your sites on 5 engines (ChatGPT, Gemini, Perplexity, Google AI Overviews and Google AI Mode), with new results every week. Business measures them on 8 engines, adding Copilot, Claude and Grok, with new results every day. What sets each engine apart is in the eight engines we measure.

Each site has its own country and dialect, set before the first measurement. The engines are asked about your client's clinic in Riyadh the way people in Riyadh ask, about your other client's store in Dubai the way people in Dubai ask, and the suggested questions for each site follow its country and dialect.

That changes the result, not only the wording. Someone looking for a dentist in Riyadh types the question in Saudi dialect and names a Riyadh district; someone in Dubai types it in Emirati dialect and names a Dubai one. The engine searches with the words it heard, and reaches different sources. So each site's questions are written in its market's dialect and Google's answers are read as they appear in its country: what you show your client is their own market, not a generic version of someone else's.

Each site's questions come from three places: what your client hears from customers in sales and support, real searches from Search Console if you link a Google account with access to your client's site, and the site's own pages. How to gather questions worth measuring is in choosing the questions worth measuring. Three sectors agencies often serve have full guides: clinics in AI answers, real-estate projects in AI answers and online stores recommended by ChatGPT.

Evidence they can read, not a number they have to take on trust. Under every verdict we keep the answer text as the engine wrote it, the links it cited, the date of the measurement and the engine that wrote it. When your client asks "where does this number come from?", you open the answer itself.

• The Tasadr score out of 100 per site, with its range whenever a measurement spans one, so your client knows how much the number swings before acting on it.
• The outcome for every question on every engine: cited, mentioned, absent or no answer. An engine that wrote no answer is never counted against your client's site.
• Who was cited instead: on which question, on which engine, and in what words.
• An analysis that says why the rival was cited: their page and your client's side by side, and what their page answers that your client's does not.
• An action plan ranked by priority: what to change first, on which page, and why.

With a measurement before the work and one after it, on the same questions. The first measurement is the baseline; without it, neither you nor your client can tell whether an improvement came from your work or from the engine moving. So measure a new client's site before you touch a page, and make the date of that measurement the starting point you count your results from.

1. 1Measure before you start: the first measurement of the client's site is the reference for everything after it.
2. 2Change one page at a time: edit ten pages in one day and you will not know which edit moved the number.
3. 3Run an on-demand measurement after the edit: 10 on-demand runs a month on Pro, 30 on-demand runs on Business, started whenever you like instead of waiting for the schedule.
4. 4Trust only a confirmed change: one answer that moved is not news. An alert reaches you in your dashboard once a change is confirmed, whether a new rival, a lost citation or a new win, and anything unconfirmed stays in the history without an alert.
5. 5Show your client the difference in words: the answer before the edit and after it, and who was cited then and who is cited now.

For every question we record who was cited instead of your client's site, and how often. That list is the map of their rivals inside the answer, and it can look nothing like their rivals in the search rankings: a local directory or a comparison site with one precise page can beat a big brand whose pages are general.

The analysis then compares your client's page with the rival's: what the rival's page answers that your client's does not, and which sub-question it covered and your client skipped. You get 50 competitor analyses a month on Pro, with no ceiling on Business. From the gaps that repeat, an action plan ranked by priority is built, one that can serve as your team's work plan for that client. How to read that comparison and follow a rival month by month is in which competitor ChatGPT recommends instead of you.

Most of that plan is edits to page text, not code: the question as a heading, the answer in two lines, the price, time or city written out. How one engine reads a page and chooses from it is walked through in how ChatGPT picks its Arabic sources, and Google's own surface has its guide in AI Overviews in Saudi Arabia.

We say it before you subscribe, because whoever buys today buys what exists today:

• No PDF export. Results are read inside Tasadr; no ready-to-send report comes out of it.
• No team seats. An account has one user today, so there is no separate login for your team or your clients.
• No white-label. The interface carries the Tasadr name, and your agency's name or logo cannot replace it.
• No API. Results cannot be pulled automatically into your agency's dashboard or tools.

Choose by the number of clients you follow, then by the number of questions each client needs. Every price is in Saudi riyal and fixed, with no meter per question or per request; the full detail of each plan, including the extra question packs, is in Tasadr plans and prices. If you are comparing measurement tools first, the eight questions to put to any tool are in how to choose a GEO tool.

Write to support@tasadr.ai with three things: how many client sites you follow, how many questions you want measured, and the markets and dialects you work in. We reply within one working day with a written quotation for Business, starting at 4,999 SAR a month. Every other way to reach us is on contact the Tasadr team.

Six steps that give you something to show your client at your next meeting:

1. 1Create a free account and add the site of the client whose market you know best.
2. 2Set the site's country and dialect before the first measurement.
3. 3Write five buying questions your client's customers ask, in their words rather than your pitch deck's.
4. 4Run the measurement, then read each answer and who was cited in it.
5. 5Show your client the baseline and agree on the first page to edit.
6. 6Move to Pro when you add the second and third client.