Lettertrace review: pricing, features & alternatives

GitHub README (read 20 October 2026). Each run sends every active prompt to the chosen model, detects your brand and competitors by name and alias, records occurrence count and first position, then asks an LLM to classify sentiment and recommendation for the brands it found. Visibility is the mention rate across prompts; share of voice compares you with tracked competitors.

Gemini, Google AI Overviews and Perplexity need their own keys because, as the README explains, their grounding doesn't survive a router. AI Overviews is described as "a Gemini call plus a forced-search system prompt of ours".

The GitHub API showed the letterstory/lettertrace repository created on 21 July 2026, MIT licensed, with 75 stars, 12 forks and a push on 19 October 2026. The CLI is published on npm as lettertrace and @letterstory/lettertrace.

The hosted lettertrace.com runs the RB2B visitor-identification pixel on its public marketing pages only, according to the README. Self-hosted deployments ship without third-party tracking.

Every row below is a capability we already listed for Lettertrace. It is the published feature set for this page, not a guess at an unpublished demo. Compare it with the Promptwatch checklist that follows.

On this listing, Lettertrace is a AI visibility tracker for Developers and technical marketers who want a free, inspectable tracker they can script from a CLI or coding agent. What we sourced here: Bring-your-own-key for Anthropic, OpenAI, Google and Perplexity, encrypted with AES-256-GCM; Topic-to-prompt variation generation; Visibility, share of voice, prominence and sentiment per run, trended over time; Cited-source capture with per-URL hit rates via target URLs on prompts; Daily, weekly or custom-interval schedules through a cron endpoint.

If you buy Lettertrace for the job above, plan a second tool (or Promptwatch) for the layers this listing does not document. Still missing here: AI crawler logs (Agent Analytics), crawl-to-citation path, bot errors; Visitor analytics: AI-referred sessions and conversions; Content Agents that plan, write, and publish GEO pages to a CMS with a review inbox; Citation analytics at page, domain, Reddit, YouTube, and offsite level; Prompt volumes, difficulty, query fan-outs, personas, city-level geo; Prompt and citation trends (what changed between checks); ChatGPT Shopping tracking and Ads Radar; Agent Chat, Unified Actions, MCP server / REST API on paid plans. Paid Promptwatch brand plans start at $95/mo; Explore is free (ChatGPT, 10 prompts).

Lettertrace's public pages (and this listing) describe a narrower job than Promptwatch. Most rivals answer whether you were mentioned. Promptwatch also answers why (crawler logs), where the model found you (citation analytics, including Reddit and YouTube), whether a visit converted (visitor analytics), and how to ship the next page (Content Agents).

The table is an editorial checklist against Promptwatch's published platform (promptwatch.com/llms.txt, 2026-08-23). It is not a hidden feature list for this vendor. 'Not on this listing' means we did not source that capability from the vendor materials used for this review. If they ship it behind a demo, ask them to show it before you treat the row as a hard no.