The New Personal Reputation Audit
What Google, ChatGPT and Reddit may tell someone who searches your name — and the repeatable audit that finds problems ordinary page-one monitoring can miss.
I’ve watched personal reputation work change quite a bit over the years, but the biggest mistake I see now is surprisingly simple: people still check the first page of Google and assume they know what the internet says about them. That is no longer enough. A customer can Google your name, another person can ask an AI who you are, and somebody else can ask Reddit what people think of you. All three may walk away with a different version of the same person.
You now have more than one search reputation
Traditional search ranks pages. AI search reads, selects and summarizes information. Community AI can take thousands of comments and turn them into one compact answer. Those are different reputation environments.
Google Search
The familiar reputation layer: webpages, news, profiles, images, videos, reviews and other results ranked around your name.
AI Answers
ChatGPT Search and Google AI experiences can assemble several sources into a narrative. The user may read the answer without opening every underlying page.
Reddit & Community Search
Reddit’s AI search can synthesize posts and comments into an answer, potentially resurfacing discussions that were previously buried inside old threads.
A revealing look at how Reddit AI selects voices
A September 2026 University of Illinois Urbana-Champaign preprint audited Reddit’s generative search across advice and support communities. The study does not prove that every Reddit search behaves this way, but it shows why reputation audits should inspect the AI answer as well as the underlying thread.
| What researchers found | Observed result | Why it matters in a reputation audit |
|---|---|---|
| Higher-ranked comments had an advantage | A one-standard-deviation increase in a comment’s vote ranking was associated with 2.88× the odds of selection. | An old but heavily supported comment may carry more AI visibility than dozens of weaker replies underneath it. |
| Formal writing was favored | A one-standard-deviation increase in measured formality was associated with 49% higher selection odds before other controls. | A polished accusation or criticism may be especially easy for a summarization system to treat as answer-like material. |
| First-person experience weakened | First-person singular wording fell from 3.3% in quoted comments to 0.06% in synthesized answers in one part of the analysis. | “This happened to me” can begin to sound more like general information once personal context is stripped away. |
| Retrieval drove much of the variation | When the system retrieved the same source posts, researchers found answers were often very similar. | The reputation battle may begin before the writing stage: which posts get retrieved can determine the story the AI has available to tell. |
Important limitation: the research was an observational preprint focused on 20 advice- and support-oriented communities. The reported associations should not be treated as universal Reddit ranking rules or proof of causation.
Run every audit twice
This is the part most people miss. AI and search experiences can be personalized, so one signed-in result should never be treated as the universal answer to “what does the internet say about me?”
Clean baseline
Use the least-personalized environment available. For ChatGPT, an unpersonalized Temporary Chat is useful because it does not use ordinary memory, custom instructions or plugins. For Google, compare against a signed-out or minimally personalized search where practical.
Normal-account reality
Repeat the audit in the account you normally use. ChatGPT Memory can influence search formulation, and Google can personalize search and AI responses using saved activity when the relevant settings are enabled.
The 20-minute personal-name audit
Use the same spelling, occupation, company and location each time. Record the date, query, answer, cited sources and screenshots. The purpose is not to “catch AI being wrong.” It is to find the reputation signals another person can realistically encounter.
Establish the identity first
Start with the simplest question possible. If the system cannot reliably identify the right person, every reputation question after that is suspect.
ChatGPT / Google AI: “Who is [FULL NAME], the [ROLE] associated with [COMPANY/CITY]? Use current web sources and distinguish anyone with the same name.”
Reddit: Search the exact name, then name + organization or profession.
Ask what the person is known for
This exposes which facts have become dominant. Pay attention to what appears first, what is omitted and which sources support the description.
If your own work is barely visible while an old dispute, article or namesake dominates the answer, you have a narrative imbalance even if none of the individual statements is technically false.
Run the risk query carefully
A prospect, employer, reporter or investor may not search only your name. They may add words meant to uncover problems.
AI prompt: “Are there documented controversies, complaints, lawsuits or significant criticism involving [FULL NAME]? Include only items supported by identifiable sources, give dates, and distinguish allegations from established facts.”
This wording matters. It asks for source-backed material instead of inviting an AI system to fill an empty space with speculation.
Inspect the citations, not just the answer
For every material statement, record which page supplied the information. Then classify the source.
Official record News report Company page Review site Reddit Forum Scraper UnknownA wrong AI answer caused by one bad source is a different problem from the same claim appearing independently across five credible sources.
Audit Reddit two ways
Do not stop after searching for posts. Check both ordinary Reddit search and the AI-powered Ask experience where available.
Open the actual source threads. Note the age of the discussion, comment score, subreddit, whether the quoted material is firsthand, and whether replies contradict it.
Repeat the important AI query
One AI response is not enough. Run the same material query again in a fresh session and record whether the claim and citations persist.
If the wording changes but the same source repeatedly drives the answer, your priority is probably the source. If the sources change while the same claim survives, the issue may be distributed more broadly across the web.
What to score
Forget one giant “reputation score.” Five smaller measurements tell you much more about what needs fixing.
How the repair changes by problem type
| What you find | Likely underlying problem | First repair priority |
|---|---|---|
| Google ranks a bad page but AI ignores it | Traditional search visibility problem | Concentrate on SERP suppression, stronger competing assets and source-level resolution. |
| AI repeats a claim that barely ranks in Google | Source-selection problem | Identify exactly which page AI is citing and determine why that source carries the claim. |
| ChatGPT or Google AI has the wrong person | Entity confusion | Strengthen identity signals, consistent bios and authoritative pages that distinguish namesakes. |
| Old Reddit material suddenly appears in AI search | Community retrieval problem | Inspect the original thread, context and age before deciding whether response, correction or monitoring is appropriate. |
| Same negative theme appears everywhere | Distributed reputation problem | Trace the original sources first. Treating each search surface independently wastes time if they all inherit the same underlying material. |
| Problem appears only in your normal account | Possible personalization or history effect | Re-test clean before launching a reputation campaign around something that may not be broadly visible. |
Do not measure AI reputation by sentiment alone. A polite answer can still contain the wrong person, an outdated fact or a misleading source. Conversely, a critical answer is not automatically inaccurate. Accuracy, provenance and context are more useful audit signals than whether the wording feels positive or negative.
How exposed is the reputation issue?
Answer six questions using what you found in your audit. The tool identifies the type of problem that deserves attention first rather than giving you a vague reputation grade.
