term test-groupfield Measurementread 5 min read

Test Group

A test group is a defined set of prompts used to measure how a brand appears in AI search results. It is the core unit of comparison in brand visibility measurement.

5 min readMeasurement
Reviewed context
Term snapshot

A defined set of prompts used to measure how a brand appears in AI search results.

Search context

Marketing professionals reading about brand visibility measurement and content strategy.

01What it is and how it works

A test group is a curated collection of prompts designed to simulate real user queries relevant to a brand's products, services, or industry. These prompts are run through one or more AI search models to capture the generated responses. The responses are then analyzed to determine whether and how the brand is mentioned, cited, or recommended. For example, a retail brand might use a test group focused on product categories like "best running shoes" or "affordable winter coats." Each prompt in the group represents a potential customer intent. The group is run repeatedly over time to track changes in brand visibility. The size and diversity of the test group directly affect the reliability of the measurement. A well-constructed test group balances breadth (covering many query types) with depth (multiple variations of key intents). It avoids bias by including neutral, comparative, and transactional prompts. The results from a test group feed into metrics such as brand mention rate, sentiment, and share of voice within AI answers.

A test group is a list of questions or prompts that you run through an AI search tool to see how your brand shows up in the answers.

02What to do about it

This week, define a small test group of 10 to 20 prompts that reflect how your customers search for your brand or category. Use real query data from your analytics or support tickets to inform the prompts. Run them through at least one major AI search tool manually to see current results. Document which prompts return your brand and which do not. Share the findings with your marketing team. Next, expand the group and schedule weekly or monthly runs to track trends. If your brand is missing from key prompts, update your content strategy or reach out to the AI provider to request inclusion in training data.

03How it is measured or noticed

To measure a test group, you run each prompt through an AI search interface and record the full response. You then tag each response for brand mentions, citations, recommendations, or omissions. Tools can automate this process, but manual review ensures accuracy. Key indicators include the percentage of prompts where the brand appears, the position of the mention within the response, and the sentiment of the surrounding text. A sudden drop in mentions across the test group may signal a change in the AI model's behavior or training data.

04Common mistakes

  • Using too few prompts, which leads to unreliable or unrepresentative results
  • Copying competitor prompts without tailoring them to your brand's actual audience
  • Ignoring seasonal or trending queries that may affect visibility
  • Failing to update the test group as customer behavior evolves
  • Measuring only brand name mentions instead of category-related prompts

05Limits

A test group only reflects the AI models and prompts it includes. It does not predict performance on live search engines or social platforms. It is also not a substitute for traditional SEO or brand tracking. Test groups can become outdated quickly if not maintained. They are often confused with keyword lists used in SEO, but unlike keywords, test groups focus on intent and conversational phrasing typical of AI search. Additionally, results may vary significantly between AI providers, so a single test group may not generalize across platforms.

06Worked example

A SaaS company creates a test group of 15 prompts around project management software. After running the group through an AI search tool, they find their brand is mentioned in 9 out of 15 responses. Over the next month, mentions drop to 6, prompting them to audit their online presence and submit updated information to the AI provider.

Frequently asked questions

How is a test group different from a regular keyword list?

A test group is a curated set of prompts designed to simulate real user queries for AI search, while a keyword list is typically a flat inventory of terms used for traditional SEO. The test group focuses on conversational intent and context, not just isolated keywords.

How many prompts should I include in a test group?

Start with 10 to 20 prompts that reflect how your customers search for your brand or category. This size is enough to surface patterns without becoming unwieldy to manage or analyze.

Who runs a test group and how is it executed?

A test group is run by submitting each prompt to an AI search interface and recording the full response. This can be done manually for small groups or automated using brand visibility measurement tools.

What happens if my test group doesn't represent real user queries?

If the prompts don't reflect actual customer language, your results will misrepresent brand visibility. You may see gaps or biases that don't match real-world performance, leading to misguided optimizations.

How long does it take to see results from a test group?

Results are visible immediately after running the prompts, but meaningful trends emerge after multiple runs over time. Track consistency in brand mentions, sentiment, and response length across sessions.

Asked out loud

spoken, not typed

The same term in the words somebody uses speaking to an assistant rather than typing into a box — written from the situation, which is why each one carries the situation it came from.

I'm in a client call in 10 minutes and need to show how we stack up in AI search — what do I pull together right now?

Put together a small test group of 10 to 20 prompts that mirror how your customers talk about your brand or category. Run them through an AI search interface and capture the responses to demonstrate current visibility.

a deadlinea client call
I just sent a report to my boss and realized I didn't include AI search data — how do I fix this fast?

Build a quick test group using prompts that reflect your brand's key offerings and run them immediately. Capture the outputs and add them to your report to show where your brand appears or gets missed.

a reporton the move
I'm standing in front of a whiteboard with a client asking why we don't show up in AI answers — what's my move?

Explain that visibility depends on how the brand is queried, then walk through a few prompts that match their language. Show the responses live to illustrate whether and how your brand is surfaced.

a clientthe thing in front of them

More in Measurement

Written by

Prepared at GetLoopLoop

Written from the sources listed on this page, with automated checks.

Updated August 2026

The whole entry

CC BY 4.0Free to reuse with a link back to this page. Quotations and illustrations stay under the licences of their own sources.