Skip to content

Articles / Measurement

Branded vs unbranded questions: what each one measures

Unbranded questions measure whether AI engines bring you up on their own; branded ones measure what they say when asked. Report them apart, never blended.

Published
Reading time
5 min read
By
AmpliRank
Tags
AI Visibility · Question Sets · Branded Queries

Use unbranded questions, the ones that don’t mention you, to measure whether AI engines bring your brand up on their own; use branded questions to check what the engines say about you once a buyer asks. Report the two apart, because a blended rate moves whenever the mix of questions changes, even when nothing about your visibility did.

In short

  • An unbranded question (“best project management software for agencies”) measures discovery: does the engine put you on the shortlist?
  • A branded question (“is Taskmoor good for agencies?”) measures how you are described: facts, comparisons and the sources behind them.
  • Answers to branded questions name you almost by construction, so mixing them into a headline inflates it.
  • A blended rate changes when the share of branded questions changes. In the sample below, the same two rates give 39% or 58% depending on the mix.
  • Keep the headline unbranded, and read branded answers for accuracy rather than as a rate to grow.

Why does the split matter?

The split matters because the two kinds of question answer different business questions. Search marketers already separate them: Google’s Search Console added a branded queries filter in November 2025, because branded and non-branded traffic behave differently and non-branded queries show how new people find a site.

AI answers have the same divide, and it is sharper. When a buyer names you, the engine is answering about you. When the buyer names only the category, the engine decides whether you belong in the answer at all. The pillar guide on measuring AI visibility makes the unbranded rate the headline for that reason.

What does each kind of question measure?

Each kind measures a different stage of the buyer’s path, and each does one job well:

Unbranded questions Branded questions
Example “Best project management software for agencies” “Is Taskmoor good for agencies?”
What it measures Whether the engine names you unprompted What the engine says about you
Named rate Set by the engine; the number to grow High by construction, since the question names you
What to read Who is named instead, the searches run, the sources cited Accuracy of facts, the comparisons drawn, the sources cited
Buying stage Discovering options Checking a name already on the list
Useful as a headline? Yes No

Unbranded questions are where AI visibility is won or lost: a buyer who never sees your name in the category answer may never ask about you. Branded questions matter later, when the buyer checks you out. There, being named is a given; being described correctly, priced correctly and compared fairly is the work.

What goes wrong when you blend them?

A blended rate moves when the question mix moves, so it can rise with no change in how often you’re discovered. Here is a sample measurement for Taskmoor, a fictional brand:

Question set Answers Named Rate 95% range
Unbranded 150 30 20.0% 14.4% to 27.1%
Branded 50 48 96.0% 86.5% to 98.9%
Blended 200 78 39.0% 32.5% to 45.9%

Now keep both rates exactly as they are and change only the mix: 100 unbranded answers named 20 times, and 100 branded answers named 96 times. The blended rate becomes 116 of 200, 58.0% (range 51.1% to 64.6%). A report would show a 19-point rise. Taskmoor’s discovery rate is 20% in both cases.

The same trap catches comparisons. A competitor’s blended rate, measured on questions that name you, compares their discovery with your recall. Measure every brand on the same unbranded answers, so a gap is like for like (how to compare with competitors fairly works through one), and ask any tool you evaluate whether it does: it is one of 8 questions to ask a vendor.

When should you use each?

Use unbranded questions when:

  • you report how visible you are in your category, week to week;
  • you compare yourself with competitors;
  • you decide which pages to write and which third-party sources to earn a mention on.

Use branded questions when:

  • you check that engines state your prices, features and integrations correctly;
  • you want to see how the engines compare you with named rivals (“Taskmoor vs Workloft”);
  • you trace a wrong statement to the source an engine cited, so you can fix or answer it.

Questions that name a competitor but not you (“alternatives to Workloft”) are unbranded for you: they measure whether you’re discovered from a rival’s name, which is often where buyers start.

What does it cost to switch later?

Switching later costs your history. A trend line built on a blended rate can’t be split after the fact unless every answer was tagged by question type, and runs from before and after a change in the mix aren’t comparable.

Tag each question as branded or unbranded when you write it, keep the headline on unbranded questions from the first run, and add branded questions as a separate set. Changing a question set mid-stream breaks comparisons; is that change real? explains why runs must ask the same questions to be compared.

Where to start

Sort your current questions into two lists: those that contain your brand or product names, and those that don’t. If the unbranded list is short, the 60 buyer questions template gives a stage-tagged starting set. Report the unbranded list’s rate with its range and answer count; if it has fewer than about 200 answers, see how many answers you need. Read the branded answers one by one for wrong facts. The method page shows how AmpliRank keeps branded questions out of the headline and reports them separately.

Sources

■ Ask · Measure · Compare · Trace · Fix · Share

Measure what AI says about you.

Cookies