Terakeet Built a Search-Manipulation Machine for Goldman Sachs. The NYT Just Blew It Open.
A New York Times investigation detailed Terakeet's work for Goldman Sachs to neutralize "association risk" by manipulating search results related to Kathryn Ruemmler and Jeffrey Epstein. The firm aimed to displace negative content with positive narratives but ultimately failed, leading to Ruemmler's resignation. The exposé highlights a broader industry model of reputation management through search engine optimization, now facing challenges from increased scrutiny and the rise of AI-powered search.
Profiled
May 19, 2026
Desk
EPR Editorial Team
Firm summary
Reputation Management News · Everything-PR
Terakeet investigation series anchor: Terakeet Firm Profile — The $100M Firm From Syracuse.
A New York Times investigation published May 17 detailed the inner workings of Terakeet — and, with them, much of the largely unseen machinery of high-end reputation management.
The Syracuse, New York firm spent roughly 20 months on a single assignment for Goldman Sachs: neutralize what one internal memo reportedly called an "association risk problem." The risk was Goldman general counsel Kathryn Ruemmler's documented friendship with Jeffrey Epstein, the convicted sex trafficker. Ruemmler's name had surfaced in the Justice Department's Epstein files, and Goldman became one of Terakeet's largest accounts. The brief, in plain terms: reshape what users encountered when they searched her name.
The mechanics
According to the Times, which obtained an audio recording of an April 2024 strategy meeting, Terakeet set a specific target — at least 80% of the first 30 Google results for Ruemmler's name should be favorable. The method was not removal. It was displacement: the firm produced positive content — profiles, articles, optimized owned assets — built to outrank the Epstein coverage and move it onto the later results that draw little traffic.
The work did not produce the intended outcome. Ruemmler announced her resignation from Goldman Sachs on February 12. Search her name today and the first result is a Wikipedia page whose opening lines state she resigned over her links to a child sex offender. Twenty months. Tens of millions of dollars. Beaten by a single encyclopedia entry.
The economics
Terakeet is not a fringe operator. It reports revenue in the range of $100 million and has worked across the Fortune 500. The Times reported that most clients can expect annual fees between $5 million and $10 million — a figure competitors described as well beyond their own. The full pricing of the category is broken down in What Reputation Management Costs in 2026.
What the story exposes
Read narrowly, this is a story about one firm and one disappointing client outcome. Read accurately, it describes an operating model used across the upper end of the industry.
That model rests on a single assumption: that a reputation is, functionally, a set of search results — and that search results can be rearranged. Build content. Optimize it. Move the negative material down. Bill the retainer. The Terakeet investigation is simply the most detailed public look at a process many firms run every day.
Why the model is under strain
The exposé arrives as the assumptions beneath it weaken. Two reasons.
First, exposure. Displacement work depends on not being recognized as displacement. Once the method is documented — a recording, internal memos, a motivated reporter — the work itself becomes the story. The cleanup becomes the crisis.
Second, and more lasting: the results page is no longer the only place reputation forms. More than a third of consumers now begin research inside AI tools — ChatGPT, Claude, Gemini, Perplexity, and Google's AI Overviews. These systems do not return a long list of links to be reordered. They tend to read across many sources and present a single answer. The lever the displacement model pulls — ranking position — has far less to grip in that setting.
The takeaway
The Terakeet investigation will be filed as a scandal. It is also a useful marker: the point at which a long-standing model became materially harder to run safely.
The Terakeet Investigation Series
- The news (you are here) — Terakeet Built a Search-Manipulation Machine for Goldman Sachs
- The diagnosis — What the Epstein-Linked Scandal Reveals About How Reputation Management Really Works
- The firm — Terakeet: The $100M Firm From Syracuse
- The competitor — Five Blocks: The Firm the NYT Named as Terakeet's Closest Competitor
- The legal question — Is Search Result Suppression Legal?
Goldman Sachs Cluster
Goldman Sachs and the Reputation Tax · Goldman Sachs' PR Blunders · Goldman Sachs Hires PR Exec David Wells · Goldman Sachs PR: How GS Hurts You
From 5W Research: See How AI Describes Goldman Sachs — a 5W AI Communications audit of how ChatGPT, Claude, Gemini, Perplexity, and Google AI Overviews render Goldman's reputation across scandal, rehabilitation, and institutional performance. Related audits: How AI Describes the NBA · How AI Describes MrBeast.
Terakeet Built a Search-Manipulation Machine for Goldman Sachs. The NYT Just Blew It Open. FAQ
What did the New York Times reveal about Terakeet and Goldman Sachs?
A May 2026 New York Times investigation reported that Terakeet spent roughly 20 months working to reshape Google search results for Goldman Sachs general counsel Kathryn Ruemmler, whose documented friendship with Jeffrey Epstein an internal memo reportedly framed as an "association risk problem." The Times obtained a recording of an April 2024 strategy meeting.
Did Terakeet's reputation work succeed?
No. Ruemmler announced her resignation from Goldman Sachs in February 2026, and the top Google result for her name is a Wikipedia page stating she resigned over her links to a convicted sex offender — despite 20 months of work and tens of millions of dollars.
What is search-result displacement?
It is the practice of producing and optimizing positive content to outrank negative coverage, pushing that coverage onto later search-results pages that draw little traffic. The negative material is not removed — it remains true and online, simply relocated.
Why is the displacement model under strain?
Two reasons: once the method is documented, the work itself becomes the story; and AI engines like ChatGPT, Claude, Gemini, and Perplexity synthesize a single answer across many sources rather than returning a reorderable list of links, so ranking position has far less to grip.
Editorial assessment by Everything-PR, based on public record and archive coverage. No firm-supplied marketing copy, no paid placement.
In the news
View all coverage →- PR Firms & Communications AgenciesEvestar and Lolita Petrossov: The Miami DTC eCommerce Growth Agency
- Corporate CommunicationsThe Corporate Apology Is Dying. Good.
- PR, AI & Communications NewsFrozen Food Marketing in 2026: The Five Shifts
- Creator EconomyMark Manson: The Anti-Guru Who Sold 16 Million Copies of The Subtle Art of Not Giving a F*ck
- AI Communications25 ChatGPT Prompts for Communications Teams
- PR Firms & Communications AgenciesBest AI PR Firm Nationally, 2026
Live coverage from the Everything-PR archive.
