Have you ever wondered what steps your cards travel through when sent in for grading? How long does it take to grade a card? How many cards does one grader process in a typical day?
I have always been curious about the exact process, and talked to a couple of present and former employees of PSA who were kind enough to answer some basic questions.
The takeaways for me are as follows: 1) The quality control measures are impressive and 2) The lack of any AI or advanced technology used in the grading process is a bit surprising.
While reading, enjoy a few random images of PSA-graded cards!
Grading Steps at PSA
The overall process of grading a card at PSA looks like this:
- Employee opens box and enters order into computer
- Front and back card images taken. PSA calls this “First View.”
- Researcher determines card details such as manufacturer and year.
- A grader grades the order for a first time.
- Label printing and encapsulation.
- Order is verified for QC / correct information on the label (PSA calls the label a “Flip”).
- A grader quickly grades the order a second time for quality control.
- Technician images cards in slabs. PSA provides these images to customer once the order is finished.
- Employee packs order and prepares for shipment back to consumer.
There’s nothing earth-shattering and we can assume the general process is similar at SGC and elsewhere.
Who Grades Your Cards?
I remember hearing years ago that the position of Grader was esteemed at PSA. You had to really work your way through the ranks to get there. I also assumed that during and post COVID, in the grading boom, perhaps grader qualifications had dropped off a bit. If so, it is only slightly. By all reports, it’s tough to become a full-time grader.
Becoming a grader is a lengthy months-long process. Prospective graders go through a long series of tests along the way. Each test is pass / fail and there are no retakes. Candidates who fail any portion of the training are excused from the process. Interested parties can reapply after 6 months.
On the Researcher side, an ultra-low .1% error rate is the goal. For you math-challenged folks, that’s one error allowed for every 1,000 card processed. And these errors are mostly caught in-house, meaning that the error rate back to the public from that side of the house is extremely low.
How Many Graders Does PSA Employ?
PSA reportedly employs about 100 graders, who often work long shifts while also being held to very high performance standards. My sources report shifts as long as 10 hours and ending as late as 1:30 am.
How Many Cards Does a Grader Grade Each Day?
The folks I talked to (in both Research and Grading) reported processing up to 40,000 cards per quarter. Let’s round down and call it 13,000 cards per month in a very busy month. That’s about 3,250 cards per week. Assuming a 5 day work week, that’s 650 cards per day. In a 10-hour work day, that’s an astounding 65 cards per hour.
This means that your cards likely receive about 1 minute of direct grading attention. Plus some additional quality control.
By the way: this math on number of graders checks out. Per Gemrate, PSA graded 1.2 million cards in December. 100 graders processing +/- 13,000 cards per month = 1.3 million.
What About Authenticity / Counterfeits?
I was curious: whose job is it to detect counterfeit cards? This seems to be sort of a shared responsibility between Research and Grading, with researchers flagging possible counterfeit cards. Ultimately it is the grading department that is responsible for determining authenticity.
The Role of AI
Given PSA’s purchase of AI upstart Genamint a while back, I assumed that AI, or at least some very advanced software, was assisting the humans at this point. Among other innovations, Genamint can “fingerprint” cards which could help identify anything from stolen cards to PSA crack / resubmits. But this seemingly is not the case. One source notes that the tech department at PSA is huge and they are definitely experimenting with AI’s possible role the process. But as of today, no AI is used at any step.
Check out this past Cardhound article for more info in AI and its possible role in the future of card grading.
What About Fraud?
Any collector has at some point heard wild speculation about cards being swapped out for lower-grade examples and the like. This always just struck me as sour grapes, or people simply not being very good at screening cards before submitting.
Both of my sources–including a former employee who left due to burnout–report that they never witnessed a single incident of anything resembling impropriety. Further, both note that every single workstation is under constant video surveillance.
Grading Myths
I asked my sources about a couple of pervasive myths. Myth #1 is that “certain clients” get “better grades” than you and me. This one has always struck me as a bit ridiculous, because with the sheer volume of business PSA generates, they have no incentive to give special treatment. To what end? The potential downside to such a scheme seems to vastly outweigh whatever sort of kickbacks are alleged.
One of my sources confirms that this theory is bunk. Graders do not have access to customer or order information at any point in the process. If an insider wanted to boost his pal’s grades, the conspiracy would be so vast, involving employees from several departments, all willing to risk their livelihood for the endeavor. Unlikely.
Myth #2 is pop control. As the theory goes, gem grades of certain cards, such as the 1968 Topps Ryan rookie or the 1980 Topps Henderson, are “controlled” to remain artificially low. While no one will confirm the pop control theory, it is true at all grading companies that “lead graders” and extra sets of eyes will pass over a high grade, high dollar iconic card before it gets the seal of approval. I can see how a sort of pop control might result from a more stringent process.
In Conclusion…
Do mistakes happen? Yes, and we see them often. “One in one thousand” is a very low in-house error rate, but when you are pumping out 1 million cards each month, that’s up to 1,000 errors per month. The errors returned to the consumer would be considerably lower than this. But while the percentage is tiny, the “visibility” of errors at PSA is higher due to their volume, which is about 6x the next leading competitor.
While personally I’m mostly an SGC collector / submitter, the research for this article bolstered my confidence in the overall card grading process.
Thanks for the article! Regarding the goal error rate, you said it is “One in one thousand” (aka 0.1%) but you also say it is “.001%”, which would be one error every 100,000 cards (a big difference!).
Can you clarify which is correct?
It’s corrected and it is .1% (or .001 but not both lol).
Matt, regarding the error rate, what is defined as an ‘error’ and how and by whom is that determination made?
The error rate cited in the article is for the research side. Presumably the rate for graders is similar, but it would for sure be much more objective on the research side. For example, incorrectly identifying a parallel / refractor on a modern card would be an objective error caught by quality control. That would be a tough job given howe many different variations there are. Grades are likely reviewed during the “second” grading, by more experienced graders. But as you know that gets much more subjective. The second grader would presumably also be reviewing for authenticity–and this seems to be something all grading companies do well, because you don’t see many counterfeits in slabs.