In 2019, Casey Newton’s reporting for The Verge offered a close look at Facebook content moderation in America through Cognizant contractors working at a Phoenix, Arizona site. The workers reviewed posts flagged by users, applied Facebook’s rules, and described a job shaped by difficult judgment calls, close performance monitoring, and exposure to disturbing material. Their accounts reveal one workplace at one point in time—not a universal account of moderation or evidence of current conditions.
Who were the moderators in the Phoenix account?
The reviewers in Newton’s February 25, 2019 feature were employees of Cognizant, a contractor Facebook relied on for content review. The story focused on the company’s Phoenix site, where moderators used Facebook’s Single Review Tool to work through a queue of posts reported by users. Facebook supplied the platform rules; reviewers decided how those rules applied to individual cases.
The scale figures in the feature are historical: it reported 1,000 moderators at the Phoenix site and 15,000 content reviewers worldwide. They describe the publication-era workforce, not present-day staffing.
What did reviewing flagged posts involve?
Moderators assessed reported posts against Facebook’s Community Standards and internal guidance. Some decisions were relatively straightforward, but others depended on context, cultural interpretation, or what a post meant in its particular setting. Reviewers could disagree, and guidance changed or received clarifications over time.
Do these 3 things before closing this tab:
1Scan for outdated or missing drivers - takes under a minute2Repair Windows errors before they cause bigger problems3Fix the driver behind crashes, sound loss and screen glitchesPerformance was assessed through audits of decisions. Newton reported that Facebook set a 95 percent accuracy target for audited contractor decisions; the Phoenix site was around 92 percent at the time and had not sustained the target. The gap helps explain the pressure of the work, but it does not mean every individual decision was wrong by a fixed rate: the figures describe the site’s reported performance against an audit target.
Was every post graphic or disturbing?
No. The feature presents a contrast between the possibility of encountering severe material and an employee’s account that much of the queue was comparatively mild. A policy manager identified by the pseudonym Brad said that most material he saw consisted of rants or reports made because users disliked a post, rather than because it violated a rule. His testimony describes his experience; it does not establish the mix of content for every moderator, shift, or site.
At the same time, the work could include content that workers found disturbing. The existence of milder items in the queue did not remove the possibility of serious exposure, nor did it make difficult decisions or their effects identical for all workers.
What did workers say about stress, management, and support?
Workers interviewed by Newton described distress associated with the job and tight performance management. Their accounts included concerns about the support available to them, including what happened after they left the work. These are reported experiences from interviewees, not a representative survey or a finding that all moderators received the same treatment.
Windows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallOutdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchThe feature also includes Cognizant’s perspective. Bob Duncan, then identified as the company’s vice president of operations for business process services, said workers were shown examples of content they might encounter so they could consider whether the job suited them. That account of preparation and worker choice sits alongside former and current workers’ descriptions of the job and support. The perspectives are not interchangeable: a stated policy or intention does not by itself establish how every worker experienced access to help.
What did the job pay, and how did the article compare it?
Newton reported annual pay of $28,800 for a Cognizant moderator in Arizona in 2019. This is a dated, location-specific figure, not a current wage estimate.
Rank #4
The feature compared it with $240,000 in salary, bonuses, and stock as the average total compensation for a Facebook employee. That comparison was not like-for-like: one figure was moderator annual pay, while the other combined salary and other forms of compensation. Neither number should be treated as a current rate.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.What happened after the 2019 reporting?
In May 2020, NPR reported that Facebook had agreed to a $52 million settlement for content moderators who developed PTSD. The amount is a settlement figure, not an individual award and not a measure of how common PTSD was among moderators. It provides later legal context for the workplace issues but does not show that every moderator experienced trauma or shared the same outcome.
A 2020 congressional hearing record also references the Verge feature and the settlement in a discussion of workplace safety. The available accounts do not establish present-day U.S. moderation conditions or the current availability of support for former moderators.
What the reporting can—and cannot—show
- It can show: how interviewed workers and company representatives described one Cognizant workplace in Phoenix in 2019, including the review process, performance expectations, and differing accounts of preparation and support.
- It cannot establish: the prevalence of mental-health conditions among moderators, conditions at every moderation contractor or site, or current U.S. workplace practices.
The account is most useful when read as reported testimony about a particular workplace, not as a statistical portrait of every person who moderates Facebook content.
Quick Recap
Sources
- Casey Newton, The Verge, February 25, 2019
- NPR, May 2020, reporting on the $52 million settlement
- 2020 congressional hearing record
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
