Transcription Platform Rev Slashes Minimum Pay for Workers

CEO Jason Chicola
CEO Jason Chicola
Photo: Rev

Rev, one of biggest names in transcription—and one of the cheapest services of its kind—opted to alter its pay structure with little warning for thousands of contractors on its platform, some of whom are furious at what they expect will be smaller paychecks from here on out.

Advertisement
Illustration for article titled Transcription Platform Rev Slashes Minimum Pay for Workers
Screenshot: Rev

Launched in 2010, Rev made a name for itself by charging customers who wanted transcriptions of interviews, videos, podcasts, or whatever else the bargain-basement price of $1 per minute of audio. That’s attracted some notable clients, including heavyweight podcast This American Life, according to the company. (Some teams at Gizmodo and its sister websites have also used Rev for transcriptions.) According to one whistleblower, a little less than half of that buck went to the contractor, while about 50 to 55 cents on the dollar lined Rev’s pockets.

Advertisement

But in an effort to “more fairly compensate Revvers for the effort spent on files,” Rev announced on an internal message board on Wednesday that its job pricing model would change—with a new minimum of 30 cents per minute (cpm) going into effect last Friday.

“There was an internal forum post made two days prior, but not everybody checks the forums,” one Revver who wished to remain anonymous for fear of retaliation, told Gizmodo. “A lot of people found out when they logged on on Friday. People are still showing up in the forums asking what’s going on!”

We’ve reached out to Rev for comment and will update when we hear back.

Illustration for article titled Transcription Platform Rev Slashes Minimum Pay for Workers
Screenshot: Rev
Advertisement

In an update reviewed by Gizmodo, marked as being appended last Wednesday, Rev sought to downplay the impact of the pay algorithm change. “30 cpm will be a starting price for a very small number of jobs. On the other hand, some jobs will now start at 80 cpm,” the company wrote, according to screenshots of the Rev’s internal message board. “The goal is NOT to take pay away from Revvers but to pay more fairly for the level of effort/skill required.”

According to the Rev worker who spoke to Gizmodo, “People have seen cpm start higher on a few jobs, but for the most part, it feels like the floor has shifted downward, and rates have dropped with it.”

Advertisement

Regardless of what goal Rev had in mind, the effect among some transcribers is to see this as a way to short-change workers. They’re right to be skeptical: After all, the gig economy has been the genesis of some clever accounting—whether its platforms shifting their pay algorithms or subsidizing wages with workers’ own tips—that rarely favors contractors.

Update 11/11/19 8:43pm ET: A spokesperson from Rev provided the following response:

Rev freelancers have full control over what jobs they decide to work on. They can accept or reject any projects without consequences as well as set preferences indicating what jobs they choose to accept, with full control over all parameters including file types, audio length, and pay. We informed our community of these changes on our central online communications hub in advance of them going into effect on Friday, 11/8. We welcome feedback on these changes and will continue to make adjustments in order to ensure our freelancers receive fair pay.

Advertisement

If you have a tip, please reach out via email, Twitter, Keybase, or leak us documents on SecureDrop.

Senior reporter. Tech + labor /// bgmwrites@gmail.com Keybase: keybase.io/bryangm Securedrop: http://gmg7jl25ony5g7ws.onion/

Share This Story

Get our newsletter

DISCUSSION

I am a “Revver+” contractor for Rev, meaning I have top level metrics for my work and maintain a certain minimum of minutes. The “perk” for being a Revver+ agent is that I get to see the newest transcription files 1.5 hours before the regular “Revvers” and “Rookies.” I’m also a grader, which means I get paid to review clips of other people’s work and provide grades which affect whether they’ll reach Revver+ status or, in some cases, have their metrics bottom out so badly they get terminated. What we contract workers provide through Rev is a usable transcription that should require extremely minimal, but preferably no, editing with proper spelling, capitalization, formatting, and punctuation, which is something that AI can’t perfectly do.

Speech-to-text is still a very flawed system because human beings are very flawed speakers. Most of the files available for transcription on Rev are typed straight from audio or video into an editor which turns the results into a transcript document. Rev also has “Line” files, which use speech-to-text and have human transcribers edit to align them with writing conventions, edit some pretty insane “mishears,” and properly sort out who was speaking. These “Line” files are often a hot mess and can sometimes take just as long as transcribing from scratch does. Even the good ones aren’t that much of a time-saver.

Most of us do not use our own speech-to-text software because of the reasons above - there are so many idiosyncrasies that vary from speaker to speaker, plus mumbling and accents and “crosstalk” and just really crummy audio quality. (You have no idea how many files sound recorded from across the room or through two tin cans and a tight string!) We have some disabled agents who have put a lot of time into using such software, but they still have to do so much editing that the payout is not good.

Finally, a large number of submitted files contain industry jargon, abbreviations, company names, and proper names that we are required as part of the job to research thoroughly in order to accurately transcribe them. Revvers can (and should) take hits to their metrics for not properly researching a name or term. Speech-to-text software doesn’t transcribe these things, nor does it do the research for us. I have personally spent up to 10 minutes following clues to hunt down the name or word, only to have to do it again for another one a paragraph or so later. When we’re getting paid by audio minute and not actual working minute, too many of those can mean a big hit to the paycheck!

Rev’s “official” response is misleading spin. The “central hub” they refer to is a forum that not all agents visit regularly because while it has work-related information, it also contains some of the same workplace drama and internet fighting that any forum might have. In the past, Rev has made its major announcements via email and many of us find it suspect that they chose not to email their contractors about such a major change as this.

Also missing from this article, because there’s no way the writer could know, is that the Rev representative sent to communicate with us on the Forum has said that part of the “incentive” to us to take some previously overlooked or rejected files it to lower the pay for the “easy” files. They claim that the best transcribers are wrong to pick up the “easy” files, despite the “freedom without consequences” the Rev statement claims, and therefore they decided to drop the pay for the “easy” files so that they will be less appealing work and people wanting to earn more per minute will pick up these other overlooked files.

Besides being punishing and manipulative, their reasoning fails for the simple fact that these other files that Rev has called “middle of the road” take more time to complete and therefore in practice the pay per minute ends up being just as low as taking those “easy” files.

And truthfully, Rev’s assessment of what is an easy file versus what is a difficult file, supposedly figured by algorithms, is already flawed. A file may have clear audio quality, but have 25 speakers and/or accents and/or need an inordinate amount of research and/or have a lot of speakers talking over one another (crosstalk) and/or have one guy calling into the meeting from a tunnel on Alexander Graham Bell’s original telephone who gets to talk on and off for more of the file than the Revver realized during their hour-long “unclaim” window. (We have one hour from claiming the file in which we can preview and release the project if it’s not to our liking or too difficult, etc., without hurting the “Commitment Ratio” part of our metrics.) Sometimes, we end up giving away an hour of our time in previewing an audio file and typing some of it out.

Rev’s algorithm has no way to gauge the audio content for the amount of work that will go into completing the file, meaning that some of these 30 cent per audio minute files that they have deemed “easy” and worth less will require major research time on top of the proofreading that our best transcribers do to make sure that our (Rev’s) customers get a professional grade transcription that doesn’t require the customer to go back and redo the work they paid us to do for them. It takes a lot of time and effort to keep one’s metrics up, especially when even the best typists and grammar aficionados can fall victim to typos as easily as the next guy.

Unfortunately for the customers providing good, clear audio, Rev has forgotten that they are equally worthy of a quick turnaround time and an excellent transcription. As a very biased Revver+ having to transcribe these files, I would say that such customers are even more worthy of having their good submissions rewarded with a good outcome, especially when some customers are notorious for submitting bad audio or speeding up their audio before submission so they don’t have to pay as much for their transcription. However, because many of the experienced Revvers are now talking about refusing to pick up a file priced below the old minimum of 45 cents per minute, when many already had a personal rule of not working for below 50 cpm, that may no longer be the case.

It seems that Rev has forgotten to value both its labor force and its best clients. Like many of my coworkers, I find that especially sad. We appreciate these customers, mark them as favorites, and come to really like and admire them from afar. I personally am rooting for so many people out there whom I will never meet!

But sadder yet is the reality that a labor force being held to reasonably and understandably high standards is not being paid in an equivalent manner. From the testimonies on the forum, more Revvers transcribe because they have to than for pocket money. From disabilities to major illnesses to job losses to single parents, Rev runs on the (sore) backs (and necks and wrists and fingers) of people living PayPal to PayPal. And from the response Rev has given us on the forums and you at Gizmodo, it doesn’t seem to care.