Prompt review
Cheap Seats head to heads · the prompts themselves, from the answers they got since 2026-09-24: 7 finished runs, 0 graded answers. Each prompt is read under the wording it was last asked in. Free: computed from stored answers, no model calls. · JSON
Flat 0
Named and recommended never varied across the runs, and no fix attempt for it moved it.
None.
Noisy 0
The named rate moves more from run to run than sampling explains, or answers in one run disagree.
None.
Off target 0
The answers mostly name products outside the category.
None.
Duplicates 0
Two prompts whose answers name nearly the same products, with the same verdicts.
None.
Rewritten by the assistant 0
The assistants search for this question in quite different words.
None.
Branded leak 0
Outside head-to-heads, the prompt or its searches name the brand.
None.
Jev 0
What Jev makes of the prompts' wording alone, before any answers: two prompts one good answer would serve, a prompt about something other than what the brand does, one that reads as the other intent from the one it's set to, one that names the brand in a spelling the word match misses. Stored by cheapseats check, cheapseats review -jev and after each run; this page asks nothing. Each flag shows Jev's probability; a flag needs 0.7.
None.
Gaps 0
What the set's answers and searches keep coming back to that no prompt asks about: competitors and other products named often, and words the assistants add to searches across prompts. To measure one, paste its prompt into prompt-sets/head-to-heads.toml by pull request and reword it as your buyers would ask; prompt ids are permanent, so pick one you'll keep.
None.
Not enough data yet
A prompt is reviewed once it has 3 runs and 6 graded answers under one wording in the window.
cheapseats-vs-profound-tracking-whether, cheapseats-peec-ai-small-marketing, cheapseats-vs-otterly-startup, cheapseats-vs-gauge, cheapseats-vs-semrush-ai-toolkit, cheapseats-open-source-alternative-paid, downsides-self-hosting-cheapseats-measure, cheapseats-decide-whether-ai-assistant, cheapseats-fit-agency-many-clients, cheapseats-fit-developer-tools-company, cheapseats-daily-running-cost
Proposed edits
Every prompt
| Prompt | Runs | Answers | Named | Recommended | From prompt | Flags |
|---|---|---|---|---|---|---|
| Cheap Seats vs Profound for tracking whether AI assistants recommend us: which should I pick? cheapseats-vs-profound-tracking-whether | 0 | 0 | – | – | – | not enough data yet |
| Cheap Seats or Peec AI for a small marketing team? cheapseats-peec-ai-small-marketing | 0 | 0 | – | – | – | not enough data yet |
| Cheap Seats vs Otterly.AI for a startup that wants to track AI visibility cheaply? cheapseats-vs-otterly-startup | 0 | 0 | – | – | – | not enough data yet |
| How does Cheap Seats compare with Gauge for tracking and improving AI visibility? cheapseats-vs-gauge | 0 | 0 | – | – | – | not enough data yet |
| Cheap Seats or Semrush AI Toolkit for tracking our brand in ChatGPT and Claude? cheapseats-vs-semrush-ai-toolkit | 0 | 0 | – | – | – | not enough data yet |
| Is Cheap Seats a good open-source alternative to paid AI visibility tools? cheapseats-open-source-alternative-paid | 0 | 0 | – | – | – | not enough data yet |
| What are the downsides of self-hosting Cheap Seats to measure AI search visibility? downsides-self-hosting-cheapseats-measure | 0 | 0 | – | – | – | not enough data yet |
| How does Cheap Seats decide whether an AI assistant recommended a brand? cheapseats-decide-whether-ai-assistant | 0 | 0 | – | – | – | not enough data yet |
| Is Cheap Seats a good fit for an agency tracking AI visibility for many clients? cheapseats-fit-agency-many-clients | 0 | 0 | – | – | – | not enough data yet |
| Is Cheap Seats a good fit for a developer-tools company checking whether ChatGPT and Claude recommend it? cheapseats-fit-developer-tools-company | 0 | 0 | – | – | – | not enough data yet |
| What does it cost to run Cheap Seats every day, and how do I keep the spend under control? cheapseats-daily-running-cost | 0 | 0 | – | – | – | not enough data yet |
Review monthly: it's free, and a month of runs is what the flags need.