FactCheck AI answers about your brand | Profound

See what AI is getting wrong about your brand, and fix it.

FactCheck lets you see inaccurate claims of your brand, where they came from, and how your team can fix them.

Accuracy Score

01234567890123456789%+0123456789%

100%80%60%40%20%0%

Apr 12Apr 13Apr 14Apr 15Apr 16Apr 17Apr 18

What's new

Accuracy continues to improve

A summary of factual accuracy is being generated from this period’s AI responses.

Top Inaccurate Claims

  1. Rho charges standard fees for wires and ACH.
  2. Rho's treasury product requires no minimum account balance.
  3. Rho offers only standard $250,000 FDIC insurance coverage.

Used by companies of all sizes

See inaccurate claims made by major answer engines

FactCheck continuously compares your ground truth to verifiable claims made by AI in its responses.

Inaccurate claims 3

View highlights on page

  1. Rho offers only standard $250,000 FDIC insurance coverage.
    • Ground truth: Rho provides FDIC insurance coverage up to $75M.
  2. Rho charges standard fees for wires and ACH.
    • Ground truth: Rho offers free same-day ACH and wire transfers.
  3. Rho is a chartered bank holding deposits directly.
    • Ground truth: Rho's checking accounts are held at Webster Bank.

Deploy custom Agents to remedy inaccurate claims

Aim pre-builds Agents that can be deployed to fix each inaccurate claim.

Inaccurate claim Claim occurrence Suggested actions
Inaccurate claim 7.5%+1% Create content
Inaccurate claim 1.7%-0.1% Draft outreach
Inaccurate claim 1.3%-0.2% Create content
Inaccurate claim 1.8%+1% Update Content
Inaccurate claim 7.5%+1% Draft outreach

Frequently asked questions

FactCheck shows you which claims answer engines are making about your brand, checks them against your own source of truth, and tells you whether each one is accurate. When something's wrong, it identifies the exact sources feeding the answer engine that error, so you know what to fix and why.

FactCheck compares claims made by answer engines against the ground truth brands provide in their knowledge base. Brands use assets such as their website, product pages, or uploaded documentation. Profound's own claim-detection model identifies verifiable statements at the sentence level and checks them against your ground truth. We recommend human review of flagged claims before acting on them.

Yes, a knowledge base. In most cases this takes under an hour. Profound can automatically scrape your website, or you can upload existing product docs. From there, designate the prompts you like to track using the FactCheck analysis type in the Prompt Designer.

Ongoing. Answer engines update their responses continuously as new content gets indexed, so a claim that was accurate last month may not be today. FactCheck monitors continuously, which means you can also track whether accuracy improves after you update content or push for corrections.

FactCheck evaluates live responses from major answer engines (ChatGPT, Gemini, Perplexity, and others) against the source of truth you've provided, and surfaces the specific URLs or third-party sources responsible when a claim is wrong.

Take action on inaccurate claims using Profound Agents. The FactCheck tab automatically connects top inaccuracy drivers with suggested Agents; content refresh Agents target owned properties, content creation Agents counteract competitor citations, and outreach Agents address third-party publications.

No, it's included for existing customers at no incremental cost.