Sagentel Basics: How to Add Your Project and Start Tracking Visibility

This walkthrough takes you from an empty dashboard to a live project collecting data. It covers every field in the setup form, what to put in each one, and the choices that are worth thinking about rather than clicking through.

Setup takes about ten minutes. Your first results arrive shortly after that.

Before you start

Have three things ready. It makes the process faster and the data better.

Your brand name exactly as customers write it. Not your legal entity name unless that is also what people call you.

A list of alternate names. Abbreviations, common misspellings, your old name if you rebranded, and any product names people use interchangeably with the company name.

Five to ten real questions your buyers ask. Pull them from sales calls or support tickets. You can start without these and use suggested prompts instead, but your own questions will always produce more useful data.

Step 1: Create your project

After registration you land on an empty projects screen. Click Add project.

Brand name

Enter the name people actually use in conversation. This is what Sagentel looks for inside AI responses, so it should match the way your brand appears in the wild rather than the way it appears on your incorporation documents.

Brand synonyms

This field catches the variations. AI responses do not always use your exact brand name, and a mention you fail to detect is a mention missing from your score.

Worth adding here:

• Abbreviations and acronyms
• Common misspellings, especially if your name is hard to spell
• Your previous name if you have rebranded, since older sources still use it
• Product names that get used in place of the company name
• Your name with and without a legal suffix such as Inc or Ltd

If you are not sure what to add, leave the field blank. It will be handled by AI when needed. You can always come back and add synonyms later once you see how models actually refer to you.

Domain

Add the domain you want to track. This is how Sagentel identifies when your own site is being used as a source, which is separate from your brand being mentioned by name. Both matter, and they are different problems with different fixes.

Industry

Choose the industry that matches your category. This shapes context for the project, including the prompts the tool suggests later. Pick the category your buyers would use to describe you, not the aspirational one.

Country

Select the country you want to track. This matters more than most people expect. AI answers differ by market, and local competitors that never appear in one country often dominate the same question in another.

If you sell in several markets, start with your most important one. You can expand later rather than diluting your first read across regions you are not focused on.

Suggested prompts

The last option is whether to enable suggested prompts. If you do not have specific prompts ready, turn this on and the tool will automatically generate a list you can review and choose from.

A reasonable approach either way: enable it, then treat the suggestions as a starting draft rather than a finished set. Keep the ones that sound like real buyer questions, discard the generic ones, and add your own alongside them.

Click Create project and you are set up.

Step 2: Add your prompts

With the project created, the next step is adding what you want to track. Click Add prompt.

Choose your AI models

Select which models to run against. Starting with ChatGPT, Gemini and Copilot gives you good coverage across different source pools, since each platform reads different things and returns different brand sets for the same question.

Tracking more than one from the start is worth doing. Checking a single platform is the most common measurement mistake in AI visibility, because being strong in one tells you nothing about the others.

Choose the country

Select where you want the prompts to run. Keep this consistent with your project country unless you are deliberately testing a second market.

Add the prompts

You can add prompts one at a time, or paste up to ten at once.

Some guidance on what to write:

Use natural language, not keywords. Buyers type full questions with context. “Best CRM” is a keyword. “We're a 15 person agency looking for a CRM that integrates with Gmail, what should we use?” is a prompt, and the two return different answers.

Cover more than one intent. A useful starting set includes category questions (“what are the best tools for X”), comparison questions (“X vs Y for a company like ours”), and brand questions (“is X any good”, “X pricing”). Brand questions are uncomfortable to track and often the most revealing.

Include your competitors by name in at least a few comparison prompts. If you never track head to head questions, you will not know how you are positioned against the shortlist you actually compete on.

Start with ten to twenty. Enough to see a pattern, small enough to review properly. Expand once you know what the data looks like.

Add tags

You can add tags now or after the prompts are in. Either works, but tagging is the difference between a flat list and a report you can read.

Tag by buying stage, product line, persona or competitor. Once you have thirty or more prompts, filtering by tag is how you find the actual insight, which is usually not your overall score but something like strong presence on brand questions and near zero on comparison questions.

Step 3: Let the analysis run

Once your prompts are added, the analysis process starts automatically. You do not need to do anything to trigger it. While it runs, there are two useful things to do.

Step 4: Generate more suggested prompts

Click Suggested prompts, then Generate prompts. In a few minutes you will receive a new set of suggestions. This is worth doing even if you arrived with your own list. Suggestions surface phrasings and angles you would not have thought of, particularly around problem-stage questions where buyers do not yet know what category of solution they need. Review them the same way as before. Keep what sounds like a real person, drop what sounds like a keyword tool.

Step 5: Tag your prompts while you wait

If you skipped tagging during setup, do it now. It takes a few minutes and it saves you from a dashboard that shows one number and no explanation.

Step 6: Read your first results

Once responses come in, you have a baseline. A few things to look at first:

Where you appear and where you do not. Sort by prompt and look for the pattern rather than the total.

Which competitors show up. Expect at least one you were not tracking. Smaller, content-heavy players often appear in AI answers well above their market position.

Whether the description is accurate. Check pricing, features and positioning. Errors here are common, they persist quietly, and they are fixable once you know they exist.

Differences between platforms. A gap between two models is a diagnosis. Strong in one and absent from another usually points at a source problem rather than a content problem.

Setup checklist

• Brand name matches how customers write it
• Synonyms cover abbreviations, misspellings and any former name
• Domain added for source tracking
• Industry and country reflect your primary market
• At least three AI models selected
• Ten to twenty prompts covering category, comparison and brand intent
• Tags applied so you can filter later

Common setup mistakes

• Entering only the legal company name and missing how people actually refer to you
• Tracking a single AI model
• Writing keywords instead of conversational prompts
• Skipping comparison prompts because they name competitors
• Leaving prompts untagged, which makes the data hard to act on
• Changing the prompt list every week, which destroys your ability to see a trend

Lock your core prompt set once you are happy with it. Consistency over time is what turns this from a snapshot into a measurement.

That is your project set up. Once responses come in, you can start analysing the data and building a strategy around it. Start at sagentel.ai.