What is a good named rate?

Why an absolute benchmark is misleading

A recommendation answer is a shortlist, not a directory. Where a category has fifteen credible competitors and an answer has room for three, the average named rate across the category is about 20 percent — a shortlist rewards a few and excludes the rest, and most of the fifteen are excluded from most answers. A 40 percent named rate in sunscreen is a much harder result than 40 percent in a category with four players.

Comparing your number to another brand’s is therefore close to meaningless unless you are in the same category, in the same market, measured on the same panel.

What to compare against instead

Your own baseline. The first month’s number is the comparator that always applies. Movement against it is the result.

The competitor set on your own panel. Because the panel records which brands were named instead of you, you get a same-instrument comparison for free. If your named rate is 20 percent and the category leader’s is 65 percent on the same forty questions, that gap is real.

By question type. Rising on brand questions while flat on category questions means you are only being found by people who already know you — the least commercially interesting kind of visibility.

What we scope a baseline against

For a consumer brand entering an export market with no prior English-language corpus, we scope against a baseline in the low single digits, and zero would not surprise us. Searchd has run no client panels, so there is no distribution to publish — that is an expectation, not an observation.

Beyond the baseline, we are not going to publish a trajectory. Searchd launched in 2026 and has not yet run a client panel for six months, so any month-3 and month-6 numbers we printed would be targets dressed as evidence — the exact move this site exists to argue against.

What we will commit to is the shape of the curve, and why. On-site and entity work lands first because it depends on nobody else and needs no time to age. Off-site corroboration lands second and slower, because it requires other people to publish or correct something. When we have real distributions across enough engagements, this page will carry them with the sample size attached.

Related: named rate, prompt panel.