Read in awe as a former Quality Assurance Specialist divulges the deepest, darkest secrets of outsourced technical support centers. Learn what happens to “rogue” call centers who refuse to give terrible customer service, why the tech support guy stops listening to you after you say certain keywords, and so much more.
The only thing worse than being a level 1 tech support rep is being a quality assurance specialist who listens to level 1 tech support reps ALL DAY LONG.
Here are my ten confessions from when I worked at HyperQuality, a quality assurance contractor for Accenture and AT&T DSL outsourced tech support call centers: Sykes, Teletech, Teleperformance, Telvista, and a few other smaller ones I can’t recall.
1. “This call may be monitored for quality assurance purposes” applies to about 0.01% of all tech support calls.
This is considered to be “good sampling.” Calls are recorded and then given to QA agencies for playback and scoring later. The idea is to select a group of call centers for monitoring and review at least two calls for each agent. In reality, call center management is allowed to select the recordings to send out for QA. Guess which ones they sent? That’s right: 20 calls from their best agent over and over again. Good sampling, indeed.
2. Agents say stupid things because they’re required to.
That excessively long branding at the beginning and end of the call, the annoying use of your formal name, the pointless verification of useless information? All are requirements forced on the agent by brain dead internal QA policies. Firms like Accenture and HyperQuality who are hired to improve these policies and procedures to improve customer service ratings, but are actually paid to reduce customer service costs. Instead of improving procedures and training so that agents know what they’re doing, QA spins its wheels trying to figure out the nicest sounding script for why an agent needs to transfer you to yet another wrong department.
3. Outsourced call centers and QA agencies are in a race for supremacy.
AT&T executives know they can’t trust their own internal QA, so they outsource. QA wants tech support agents to be outstanding for the executives. Outsourced call centers interpret QA agencies as threats (rightly so), and will do any amount of lying and number fudging to make QA look incompetent instead of fixing their broken support team. Brain dead AT&T junior executives who are on the hook for hiring these defective call centers will then question the QA agency’s methodology and recommend that the QA experiment be scrapped in favor of returning to an internal QA process, which they know has no power to change anything whatsoever. Corporate politics, clusterfuck is thy name.
4. Offshore (non-U.S.) agents are trained to clue in on keywords, not comprehension.
Offshore agents have highly varying proficiencies in English. They can speak it better than they can hear it. In order to help agents overcome this limitation, they’re encouraged to pick up on keywords and formulate support solutions around them. The problem is that agents frequently stop listening after they find the first keyword that they understand, even if there’s five more minutes of talking after the keyword.
For example, if you say, “Whenever I sign on, my modem starts making noise as if I were dial-up, even though my DSL is okay.” An experienced tech would say, “hmm, this sounds like the browser is set to always dial a connection. Let’s open up Internet Properties.” However, a keyword-focused tech would hear “sign on” and try to verify your account information so he can reset your password. Another keyword-focused tech would hear “dial-up” and try to transfer you to the dial up department. Google is not their friend, apparently.
5. Onshore agents get to be as racist as they wanna.
One challenge of the QA team I worked on was a Midwestern U.S. call center that was outright rude to any customer who sounded Latino, Puerto Rican, Chinese, or even Finnish. This call center would use the customers lack of mastery of English against them, repeatedly saying in very loud drawls, “ma’am, I — I can’t understand what you’re saying. You need to find someone who speaks English and have them call in for you.” (Lie! There are plenty of translation departments within AT&T.) We red-flagged calls on a weekly basis for six months, citing overt racist behavior. At the end of six months, the call center’s contract was finally terminated. When we inquired why, the official reason was “Quality Assurance issues.” The un-official reason was “cost control.” The day after the onshore center was terminated, a new offshore center was hired for 1/10th the cost.
6. When you call in and get a different answer every time, blame the internal documentation.
Agents are required to research answers against the internal doc system, which is essentially a poor implemented wiki. But it functions oh so much more poorly. If only AT&T had taken their $150,000 investment in a proprietary documentation system and invested it in a free wiki package. There are several documents for any particular issue. Documents should be labeled for use by specific departments and teams, but frequently aren’t. What’s worse is that QA can’t score down an agent for reading the obviously wrong document if the document is poorly categorized, or has been available for less than a day. After all, you have to give agents a reasonable amount of time to catch up on the litany of “OMG NEW EMERGENCY FOLLOW THIS CRISIS PROCEDURE” bulletins and documents that are published every day and then never maintained or retired once the crisis has been abated.
When an agent tells you you have to rebuild your modem firmware because your speed is a little slow, he’s probably reading from a 4 year old technical bulletin. He’s probably been working off that bulletin since the day he started.
7. Agents consistently failing a particular attribute? Obviously it means the bar was set too high.
When scoring agents, we look at forty different attributes, ranging from objective standards such as using the member’s name three times during the call, to softer skills such as how positive and upbeat the agent sounds. Male agents in India always fail the upbeat attribute. Many regions in India are highly homophobic, and sounding American-style upbeat and positive apparently means you’re gay over there. Sounding monotone and disinterested is the height of masculinity. Agents would rather fail this heavily weighted attribute than make such a transgression, so they fail the call.
Call center managers, instead of helping their agents adopt a more internationally-minded open attitude, will campaign to have the attribute dropped altogether. Their reasoning? “Oh, he sounded very upbeat and positive *to me*.” Yeah, because you’re a great representation of the customer, pal. The sad part is this reasoning will win, call scores will rise dramatically, and outsourced call centers will receive “great improvement!” trophies and bonuses.
8. Quality Assurance could use some quality assurance themselves.
Nothing says “the blind leading the blind” like outsourcing your quality assurance staff to the lowest bidder in India. Which is precisely what HyperQuality does. Here’s how it goes:
Me: “Hey Rashim, don’t send out that powerpoint. It has some old data in it, and I need to update it before tomorrow’s meeting.”
Rashim: “Okay, I won’t.”
Client: “why does this powerpoint have such old data on it?”
Rashim: “Wha? I sent out the powerpoint like you said to.”
Ugh. Where did Rashim work previous to HyperQuality? You guessed it: call center for AT&T DSL Tech Support.
This happened constantly across the board. I don’t want to sound racist on this point, though. We did have a couple of incredibly smart staff that came from India. They were a pleasure to work with and we paid them well. I’m sure the rest of the staff could have come up to speed with more training.
9. Smart call centers would rather go rogue and receive a failing grade than assimilate into the clusterfuck.
The true and singular pleasure of this job was monitoring a rogue call center from the Southwest. They used their own documentation, had their own scripts, and hired incredibly smart and capable techies. We “red flag” failed each and every one of their calls because they did not follow the prescribed metrics that we scored against. This team had some of the best I’ve ever seen from both an infrastructure and service standpoint. They shrugged off action plan meetings, ignored saber rattling execs, and just kept delivering amazing service. Their average QA call score was 23%. For whatever reason, this contract could not be terminated, so they were slowly starved of calls until they could not support themselves and had to lay off staff.
10. The QA metrics themselves are the battleground for a power struggle between different AT&T factions.
So some AT&T junior exec gets it into his head that he’s going to single handedly rescue all of AT&T’s support issues by transforming the entire customer service paradigm into a baseball metaphor, and insisting that each call result in a “home run.” Said exec then proceeds to start injecting this philosophy into as many agent documents as possible, overriding all of the work that QA and other management teams have done. None of these overriding documents say what has to actually be done to achieve this home run, only that achieving the home run is key to the success of the team.
The fatal flaw in this plan is revealed when multiple off-shore call center managers respond with, “vait, vhat is dis base-boll? Ve do not know American shports games.”