What should I look for in an AI visibility tool?
A buyer's checklist for AI visibility tools: engine coverage, prompt source, repeatability, citations, crawler logs, attribution, action and pricing, with tools compared.
There are now dozens of AI visibility tools, and their feature pages sound alike. The differences that matter show up in a handful of places. Use this checklist to compare them, then see how the main options score.
The checklist
1. Engine coverage, and how each engine is queried
List the engines your buyers use, then ask how each is checked. Live interfaces and API calls can return different answers. Promptwatch monitors ChatGPT (Search), Google AI Overviews, Google AI Mode, Microsoft Copilot, Gemini, Perplexity, Amazon Alexa and Meta AI through the real UI, and tracks Claude, DeepSeek, Mistral and Grok through APIs, so you know which are which. Check plan tiers too. Scrunch AI Core covers four LLMs, and Gauge Growth covers three platforms.
2. Where the prompts come from
Prompts you write, prompts seeded from your Search Console data, prompts the vendor generates and panels of real user prompts are four different things. Ask in writing. Evertune and Profound connect to panel or demand data and are sales-led. Promptwatch's Prompt Explorer suggests prompts from Search Console, ranking keywords and People Also Ask.
3. Repeatability
AI answers vary between runs. Look for daily runs, enough samples per prompt and trend views. Promptwatch runs prompt panels daily per model, country and language. Evertune samples each prompt up to 100 times per model. Weekly-only refreshes, which the catalog notes for Scrunch AI and LLM Pulse by default, can hide a drop for days.
4. Metrics you can interpret
A single "score" is not enough. You want mention rate, share of voice, sentiment and position. Promptwatch reports Visibility Score, Share of Voice, Sentiment (1 to 100 per response, with a weekly AI summary) and Position in Answer, and a Brand Book with aliases to control what counts as a mention.
5. Citation analytics
Knowing you were not named is half a finding. Knowing which pages and domains the engine quoted is the other half. Look for page and domain level data, offsite mentions, and Reddit, YouTube and LinkedIn reports. Promptwatch also classifies citations by content type and source type.
6. Crawler logs
If AI bots cannot reach a page, nothing else helps. Look for IP-verified logs, named agents such as GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot and PerplexityBot, and explained errors. These are rare among trackers. They start on Promptwatch Professional at $245/mo, not Essential.
7. Traffic and conversion attribution
Do AI-referred visits show up, and do they convert? Promptwatch's visitor analytics uses a script, a GTM template or the WordPress plugin. Without this link, the score is hard to defend.
8. Action, not just reporting
The tool should tell you what to do. Options include Promptwatch's Action Board and weekly Discovery agent (up to five evidence-backed actions per run), AthenaHQ's Action Center, and content engines such as Gauge and AirOps.
9. Pricing model
Credit systems can run down fast when you track multiple engines. AthenaHQ Starter has 3,600 credits where one credit equals one AI response. Rankscale uses credits where one engine query typically costs 0.25. Check what a realistic panel costs per month, not the entry price.
10. Data access and trust
Exports, API, MCP and security matter at team scale. Promptwatch has CSV and Excel exports, scoped API keys, a hosted MCP server and REST API, and is ISO/IEC 27001 certified and GDPR compliant.
How the main tools score
| Tool | Engines and method | Citations | Crawler logs | Attribution | Published price point |
|---|---|---|---|---|---|
| Promptwatch | 8 real-UI platforms plus API models | Page, domain, Reddit, YouTube, LinkedIn, offsite | Yes, IP-verified | Yes, with conversions | Essential $95/mo; Professional $245/mo; Business $579/mo |
| AthenaHQ | 11 models on Starter | Citation Engine on Enterprise | No published product | GA4 listed | Starter $295/mo |
| Scrunch AI | 4 LLMs on Core | Not detailed in the catalog | Bot traffic reporting | Not documented | Core $250/mo |
| Profound | Up to nine or ten on Enterprise | Citation share | Agent Analytics | Not documented | Enterprise custom |
| Gauge | 3 platforms on Growth | Not detailed in the catalog | Not documented | GA4 connection | Growth $599/mo |
| LLM Pulse | 5 core engines, others as add-ons | Yes | Not documented | AI traffic analytics | From €49/mo |
| Semrush AI Toolkit | 4 engines | Mentions and links | No | No | $99/mo per domain, unconfirmed |
Red flags
- A vendor cannot say where prompts come from.
- No way to see the raw responses behind a score.
- Results shown only as a single number.
- A promise of a guaranteed recommendation. No tool can deliver this.
- A price that only makes sense for a handful of prompts.
Who should pick which
- You want to measure, diagnose and act in one product: Promptwatch.
- You want a managed action workflow and can pay credit-based Starter prices: AthenaHQ.
- You are enterprise and want a sales-led relationship: Scrunch AI first, or Profound if you want its broader agent contract.
- You already use Semrush: the add-on is convenient, with limits on depth.
A good first test
Take ten real buyer prompts and a competitor. Ask each vendor, live, to show the answers, the cited sources, whether crawlers reach your key pages and what happened to AI-referred visits. The tool that can show all four is the one worth paying for. Promptwatch's free Explore plan (10 prompts, ChatGPT only) is an easy place to begin.