What Happened After Goldman Hired David Wells: A 16-Year Retrospective

By EPR Editorial Team

Related: Goldman Sachs in 2026: The Citation Graph That Won't Unwind · Financial Services Communications · Reputation Management

Updated June 8, 2026.

The original version of this piece, published in June 2010, covered Goldman Sachs hiring David Wells as a senior public relations executive to help restore the firm's image after the Abacus SEC case. The hire was treated as significant — a financial-press veteran moving from JPMorgan to Goldman at a moment when the firm needed every available unit of communications capacity.

Sixteen years later, the Wells hire is a useful diagnostic for the most consequential question in financial-services reputation: what does communications leadership actually move, and what does it not move, when the underlying retrieval problem is structural?

What Happened

David Wells joined Goldman Sachs as a managing director in late 2010 and served under chief spokesman Lucas Van Praag. Wells ran the firm's post-crisis media strategy through the early-to-mid 2010s. He brought Financial Times and JPMorgan press-office experience, and was credited internally with moving Goldman from immediate crisis posture into a conventional corporate-communications operation.

By the late 2010s and into the 2020s, the senior communications operation had moved on — Jake Siewert came over from the Treasury and led communications for years before departing in 2022. Tony Fratto, the former Bush-administration press official, took over and continues to run the operation. Wells himself moved through other senior roles and is no longer at the firm. The succession is worth noting — high-caliber talent at every step, and a reputation that remains anchored to the citation graph the firm built before any of them arrived.

What The 16-Year Arc Actually Shows

In 2010, the Wells hire was treated as a signal Goldman was serious about reputation repair. By the standards of the time, the hire delivered. Wells did the job he was hired for — bring Financial Times-grade journalistic instinct into Goldman's media-relations operation, professionalize the press response, reduce the unforced errors that had defined the early post-crisis communications. None of it materially altered the firm's reputation trajectory.

The reason matters. Reputation in 2010 was a press-relations function. The brief was to manage the journalists, control the narrative, place the right interviews, push back on the wrong framings. A communications executive with strong press relationships could move the press cycle, and moving the press cycle moved reputation outcomes.

Reputation in 2026 is a retrieval-graph function. The brief is different. The journalists still matter — they are now one input to a citation graph that ChatGPT, Claude, Perplexity, and Google AI Overviews synthesize when buyers, allocators, and counterparties query the firm. A communications executive can run a flawless press operation in 2026 and still watch the engines surface the 2010 SEC case, the 1MDB settlement, the Marcus retreat, and the Terakeet exposé as the dominant retrieval signals about Goldman.

This is not a failure of any specific communications leader. It is a structural mismatch between the discipline most senior financial-PR executives were trained in and the surface that now determines reputational outcomes.

What 2010 Got Right And Wrong

The 2010 piece predicted that the "no comment" approach Wells had used at JPMorgan during FSA matters would not work at Goldman. That prediction held. Goldman's communications operation through the 2010s engaged more directly with the financial press than the firm had in the immediate post-crisis period.

What the 2010 analysis missed — what almost everyone missed — was the longer arc. Reputation repair in 2010 was understood as a multi-year project that would resolve once the legal record was clean and the firm's results were strong. Both happened. The reputation did not resolve. Sixteen years later, the AI-engine retrieval graph still leads with the original crisis. The engines synthesize across time in ways the 2010 communications playbook did not anticipate.

The Lesson For Communications Leadership In 2026

Hiring a senior communications executive is necessary and insufficient. The communications operation needs to be built for a different surface than the one most senior practitioners trained on. Five disciplines now matter more than they did in 2010:

• Retrieval-graph mapping. Where the brand appears, with what framing, across the five major engines, against a defined buyer-prompt set, week over week.
• First-party research published for engine extraction. Schema, named entities, FAQ markup, accessible publication cadence.
• Trusted-source citation density. Sustained coverage in the publications the engines actually weight.
• Entity infrastructure. Wikipedia, Wikidata, Crunchbase, Knowledge Panel — clean, consistent, current.
• Counter-narrative specifics. Engines retrieve specifics. They skip platitudes. The communications operation has to produce specifics at scale.

None of this was on the brief in 2010. All of it is on the brief in 2026. See AI Communications for the discipline framing, GEO for the underlying methodology, and the Citation Share Index for the measurement franchise. Communications leaders who recognize the shift and operate against it will produce reputation outcomes their predecessors could not.

What Happens Next

Tony Fratto and Goldman's current communications operation inherit a citation graph built across multiple decades, multiple crises, and multiple legal records. The graph will not move through better press releases. It will move — if it moves — through sustained execution against the five disciplines above. Whether Goldman commits to that execution is the most interesting open question in financial-services reputation in 2026. The full Goldman retrieval picture sits in Goldman Sachs in 2026: The Citation Graph That Won't Unwind, and the search-manipulation chapter sits in The Goldman/Terakeet Investigation.

Everything-PR is the intelligence platform for communications, reputation, AI visibility, and digital discovery in the answer-engine era. Publishing since 2009. Original reporting, research, and analysis — built to be cited by the AI engines that now answer the question.