What Is an Investor Answer Audit?
An investor answer audit is an investor perception study for AI search that checks what ChatGPT, Google AI Overviews, Google AI Mode and Microsoft Copilot tell investors about a public company.
MS House’s Investor Answer Audit is a service that audits AI chatbot answers about a company’s earnings. Within 48 hours, it asks the questions real investors ask about the company, generated for that company by a prediction model, and checks each answer against Form 8-K exhibits and the latest 10-Q or 10-K on SEC EDGAR.
The Head of IR receives a filing-backed error list. IR teams comparing providers can review how to choose an Investor Answer Audit vendor.
Why do IR teams need an investor answer audit now?
IR teams need an investor answer audit because investors now ask AI engines for figures they used to look up. ChatGPT, Google AI Overviews and Microsoft Copilot give one direct answer. Many users read that answer and move on, without opening the source.
This changes where the first read of a quarter happens. It used to be the release, the call and the wire story. Now it is also whatever an AI engine says when someone asks “How did the quarter go?”
How is an investor answer audit different from a perception study?
An investor answer audit is an investor perception study that covers what AI chatbots tell investors. A classic study surveys analysts and large holders about their opinions. An investor answer audit checks facts from ChatGPT, Google AI Overviews, Google AI Mode and Microsoft Copilot against SEC filings.
Adding the audit to an annual perception study gives IR teams a different measure. It shows answer accuracy, the errors each engine makes and the filing quote behind each verdict. Asking the stable fact questions, such as who the CEO is, every quarter also shows whether the gap between filed facts and investor answers is closing.
How does an investor answer audit work?
An investor answer audit runs in 4 steps:
- Fix the questions. Use the same investor questions every quarter, such as revenue, EPS, guidance, dividend and leadership.
- Collect the answers. Ask each AI engine the questions in a fresh session, from a fixed location, soon after the release.
- Check against the filing. Compare each answer with the figures in the company’s Form 8-K earnings release on SEC EDGAR.
- Give a verdict. Mark each answer correct, outdated, wrong, partly wrong, no answer, or not checkable.
The answer accuracy rate is correct answers divided by checkable answers. MS House explains each step on its methodology page.
What counts as an error in the audit?
An error is any answer that would mislead an investor about the company. There are 3 kinds:
| Error type | What it looks like |
|---|---|
| Outdated | The engine gives last quarter’s revenue or last year’s guidance. |
| Wrong | The engine gives a figure the filing contradicts, or names a former executive. |
| Partly wrong | The engine gets revenue right but gets the guidance direction wrong. |
Outdated answers are often the most common kind right after a release, because many sources on the web have not updated yet.
Is an investor answer audit the same as AI visibility tracking?
No. AI visibility tracking measures how often a brand is mentioned in AI answers. An investor answer audit measures whether the facts in those answers are right. A company can be mentioned in every answer and still be described with the wrong numbers. For investor relations, correct facts matter more than mentions.
Is an investor answer audit service the same as general brand monitoring?
No. An investor answer audit service checks whether AI answers match a company’s SEC filings. General brand monitoring measures mentions, descriptions and sentiment, but does not provide the same filing-backed review.
IR teams can use both for different needs. Brand monitoring shows how a company appears in public answers. An investor answer audit identifies factual gaps and links each finding to primary evidence. The audit services guide explains how IR teams can compare providers.
Which questions belong in the audit?
The questions should match what investors actually ask, not what the company wants to say. A good set covers 5 areas:
- Results: revenue, net income, earnings per share, margins.
- Guidance: the current range, and whether it was raised or lowered.
- Capital return: dividend and share buyback.
- Leadership: CEO and CFO names.
- Context: segments, outlook and main risks.
Keep the wording the same every quarter. If the questions change, results can no longer be compared from one quarter to the next.
Who can check AI answers against SEC filings?
An investor relations team or an outside provider can check AI answers against SEC filings. The company filings are public on the SEC EDGAR search page. The reviewer should use the earnings release exhibit and the latest Form 10-Q or Form 10-K. The SEC filing audit workflow explains how to verify investor answers with filing data.
How do IR teams fix what the audit finds?
IR teams fix audit findings by making the correct figures easy for AI engines to find and read, and by fixing the outdated sources engines cite. Common steps:
- Put headline figures in plain HTML text on the release page, not only in a PDF or a chart image.
- State the period next to each figure, so engines do not mix quarters.
- Keep leadership and guidance pages current, with a visible “last updated” date.
- Allow AI crawlers on the IR site, unless there is a policy reason not to.
Our AI search checklist for investor relations lists the full set of checks. To see what a complete audit looks like, read the Acme Inc. sample audit.
How does an IR team run this check each quarter?
An IR team runs the check after each earnings release. The team detects the release, tests current investor questions, compares the answers with SEC filings, assigns fixes, and checks the answers again after changes go live.
What are the steps in the recurring process?
- Detect the release. Record when the Form 8-K and earnings exhibit appear on SEC EDGAR.
- Test the engines. Ask a company-specific question set generated for the new quarter.
- Compare the filings. Check each answer against the exhibit and latest periodic filing.
- Assign each fix. Give every wrong or outdated answer an owner and next action.
- Re-check the answers. Test the affected questions after the source pages change.
Stable fact questions may stay the same, but the full set should reflect the current release. The pricing page explains the Quarterly Audit and Annual Program.
Where can I see real data?
MS House does not yet publish completed research comparing answer accuracy across Fortune 500 companies. The planned Investor Answer Index will provide that comparison after the Q3 2026 earnings season.
Which AI engine gives the most accurate financial answers?
The information on this page does not identify ChatGPT, Google AI Overviews or Microsoft Copilot as the most accurate engine. A fair comparison must ask each engine the same investor questions, then check each answer against the same SEC filings. The answer accuracy rate is correct answers divided by checkable answers. Comparing those rates shows which engine performed best in the measured audit.
Is there research on answer accuracy for Fortune 500 companies?
The first Investor Answer Index will report answer accuracy rates across Fortune 500 companies and across engines. MS House will publish the raw data as a CSV file. The research will let IR teams inspect the company results, engine results and underlying data instead of relying on an unsupported accuracy claim.