Category: "Outsourced Customer Support"

Acronis and Independent Hardware Restoration

I would like to believe my horrendous experience chronicled here is just likely an exception and my sample size is just too small to draw far-reaching consequences. But being in contact with Acronis Customer Support 5 times and getting the same experience every time, it is becoming statistically significant and perhaps others learn from it.

Early last week I called Acronis to inquire about a hardware-independent restore need I have (my old PC is a Dell Precision 670 w/ a SATA RAID0 and I would like to load that system volume image onto my new PC, which is a Lenovo Thinkstation D20, w/ SAS RAID0 RAID as system volume and of course this PC has a different chipset and hardware components).

I use ShadowProtect from StorageCraft as well, but that failed to do the restore in this case as the target volume was a GB smaller than the imaged original partition. I called Acronis and asked whether the specific task I had could be solved with their newest and bestest software, Acronis True Image Home 2010 ( with its trumpeted Plus Pack (

I had a chat session with one of their people and I was reassured it was "absolutely" possible. I was skeptical so I specifically asked to consider my relatively involved hardware requirements. Based on that recommendation and went ahead and purchased the upgrade and the PP.

I created a boot CD and since my old machine ran ShadowProtect (and installations of the 2 imaging rivals cannot exist on the same PC at the same time), I did not have Acronis image of my old PC. I created one from the boot CD's restore environment and then I went ahead and attempted restoring that image onto the new PC.

That's when I ran into problems and contacted support I had the worst of experience. The agents were incompetent and unhelpful. Overall I had 5 chat sessions (including the initial). Some of it can be found here:




These 2 samples accurately capture the way all sessions went down. I received a feedback invite, which I filled out at first with all of the questions answered with the worst available (1) answers and then when I hit "submit" it produced an error (imagine my frustration that on top of struggling much already, trying to find a channel to relieve my frustration, I lose many paragraph of writing.

In any case, I repeated the survey, this time making sure I don't pick all bad choices (in case the system filters those and that's why an error message was produced) and it went through. Here are the screen shots of the survey ( and the text as not all of that is visible in the screen shots ( Please match the numbered text in the doc file with the free text fields of the survey.

Basically, I blasted them for (a) not being interested in solving my problems; (b) running a customer support system that is dysfunctional because of the quality of the agents.

I am not in this business, but I know providing software support is immensely costly and difficult. The price of the software does not cover an hour of wage, if you want to employ a competent and capable agent. You could instead hire somebody in an outsourced facility for 1/10 or 1/15 of that rate, which some of these companies opt for. However, you end up with a completely incompetent and useless bunch of imbeciles.

For example, in my case, I spent hours and hours chatting over several days and my problem was not solved. Maybe it is an aberration. But the problem I was having, even if not widespread, not unique either.

I believe (and saw samples of it elsewhere), there are 3 main ways of improving efficacy and overall customer care/technical support experience, without necessarily increasing the costs to the point of hurting the business model.

1. A tiered model of customer support is almost a must. The 1st line pawns should be encouraged to escalate a problem that they can obviously not solve. When it comes to escalating, I am afraid the current system disincentivizes 1st tier agents. They are pushed to do as many calls as possible and document that it was "resolved" in some manner. They wish that by luck they get all no-brainer easy calls. They dread complicated issues. When they get a problem that cannot be easily solved, they are faced with a dilemma. If they are ethical, they escalate it, maximizing the chance the problem actually does get resolved. But if they escalate too often, they increase costs, because higher tier agents need to be involved and higher tier=higher cost.

Therefore they feel a pressure to call problems "resolved" to avoid escalating. A good compromise is to give the customer dummy tasks that take time. That way they get rid of a complicated problem (so that they can work with easy ones to improve their stats and let some other loser get the complicated problem when the customer calls back finding at the end of the dummy task that their problem is unresolved). The dummy task option works perfectly for complicated problems they realize they can't resolve, by moving on from it quickly while not hurting their stats. The problem is that it can only go so far as there is only a limited number of "dummy tasks" that customers (who may not be complete dummies themselves) may not eat any sh*t the agent tries to sell (i.e. questions if an obviously unfeasible task is given to them).

So what can they do when all obvious dummy tasks are exhausted? Again, they are faced with a dilemma. They could either escalate at this point, if they are minimally ethical, or they could lie. They could suggest another dummy task that they know is completely useless AND EVEN THE CUSTOMER WOULD KNOW that it is completely useless HAD THEY BEEN HONESTLY INFORMED about the specifics of the recommended task.

But that's the point. The agent has to lie about the task, otherwise the customer would just hang on further. In our example, the agent first (after clearly realizing he could not solve the problem) suggested we discontinue the chat session and he would email the solution. Right. You know when he would have emailed. Of course he might not have realized that I saved the chat, but then again, the chances this would have come back to hurt him in any way or form are almost nil and he knows that.

Of course I refused the email offer (see the 2nd chat link above), knowing it was just a cheap trick. Then he came up with the ISO file, which was clever on his part, I should say. Given we discussed earlier (see chat file) that I burned a boot CD from an ISO I created from the Home Image 2010 software AFTER I installed PP (which is how it is supposed to be done), he sends me a link to an ISO file that contains a boot CD for version 2009! Of course that is not obvious from the name of the ISO and I only found out this when I burned the ISO and booted my PC from it.

Obviously, had he told me that the ISO he was sending was an EARLIER version of the software, which lacks the universal hardware support of the newer version, I would have told him that we should just forget that step as there was no conceivable chance it would have worked after the later version failed. In any case, I tried, and, as expected, it did not even go as far in the process as the 2010 version.

Sometimes older versions can work in certain situations better than the latest version, because some features in the new version are no longer available. If he tried to explain to me that the reason he wanted me to try the Home Image 2009 was related to some similar assumption, I would probably have gone with that, very skeptically, and asking further assurances and information. Clearly that is not how he did it. He pretty much got me on this one and achieved what he wanted, which is to end the painfully long and hopeless chat session and move on. On his end, he probably documented everything was perfectly resolved as I received an email from Acronis to that effect.

2. I digress, but let's get back to the "perfect" customer support system. The second element (additionally to a multi-tiered model, which is a must) is proper followup. Rather than sending an email next day and a survey, etc, a supervisor should check after EVERY call or chat session as to what happened, how the session ended, what was resolved what not, and why. Presumably the supervisor would be a higher paid, literate high tier agent, who could rather quickly figure out what happened. Tier 1 agents performance would be then rated based on the supervisor's documentation. If the problem was not resolved as it should have been, the supervisor could either reconnect to the first Tier 1 agent, connect to another one, or help solve the problem himself/herself, depending on the situation. If a task is given, the supervisor could check if that was appropriate. E.g. in this case, when (after seeing the 3 days of chats, especially the last one) a supervisor saw that the Tier 1 agent linked me to a clearly hopeless earlier software version, he could have then connected back to the Tier 1 agent and questioned him for why this was done and would have offered a more appropriate solution. If such a system existed, I am pretty sure my problem would not have required me to contact them 5 times and would not have still been left without a resolution.

3. Finally, in terms of incentives, at least for customers, it would be better to incentivize the Tier 1 support staff based on percentage of problems actually resolved as opposed to how many calls they could take in a given time. I sense some companies have this right. The agents are rather aggressive about the problem and pursue some form of followup as opposed to finding ways of moving on (and documenting the problem as "resolved"). I know there is more into it than what I can grasp based on the information I have. There may be some strategic decisions about how this should be handled and given the relative rarity of highly complex issues (and thus, because of their rarity, their low financial impact), it is more financially viable to have a large work force of Tier 1 agents, who are exceedingly low paid (and thus are unintelligent) and incentivize them by the number of calls taken as that system will satisfy 90% of customers (and exorbitantly piss of 10%, but if they are pissed, it probably doesn't matter how much they are pissed; the fact they are 10% limits their impact one way or another anyway).