Customer Service - A Real Life Example: Beyond Blinking Lights and Acronyms

Sitemeterlogo I use Sitemeter to collect statistics on visitors to my blog. It is an excellent product that provides me information on how many visits and pageviews I've had, geographic location (don't worry, it is just to the city level - it doesn't give me your home address), length of visit, how they were referred to me and what search terms they used if it was the result of using a search engine among other items. I also use Statcounter, IceRocket and Google Analytics. They all give similar information although each has its own way of presenting and summarizing the data. I do use all 4 products but use Sitemeter the most. I do a quick scan of Sitemeter every day to get a feel of how people are finding me and what articles are of most interest.

Because I use Sitemeter extensively and really like it, it has been disappointing to recently suffer 2 problems with their service. Unfortunately, this type of situation is not unusual with IT service providers. However it is important because as I will show later in this post it does have customer (revenue generation) impact. I bring this up not to complain (well OK maybe a little) but to use this as an example of how these situations could have been mitigated and customer service improved with very little effort and little or no cost. Perhaps we can all learn from this.

OK, so by now you may be saying what happened and what, Mr. IT Hotshot, should they have done about it?

In January, I noticed an unusual decline in the number of visitors as reported to Sitemeter. In checking with the other reporting systems I found that there was actually no decline at all, for some reason Sitemeter wasn't registering my visitors. Somehow I found an email address (it was for the webmaster not support) and emailed them to see what the issue was. They replied that "problem you are experiencing is due to a handful of sites on your particular server exceeding normal traffic levels, which unfortunately affects all those assigned to this specific server". Their answer was to either set up a new ID on yet another server or just accept the reporting delay until they could resolve the issue (no time estimate given). I chose to go with the new ID.

The second issue occurred this weekend. On Sunday I found that according to Sitemeter I only had 1 visitor which the other reporting systems showed to be significantly understated. This didn't change at all during Sunday and on Monday morning it was showing zero visitors which was again not correct. At this point I again emailed the Sitemeter webmaster describing the problem and hoping I wouldn't have to establish yet another ID on another server. Although they did not respond to my email I subsequently found a posting on their weblog that stated:

s30 - Delays

March 26th, 2007

FYI - s30 started acting up sometime over the weekend and the result is lag and delays in reporting. We’re looking into this and will update everyone when we have it resolved.

The Sitemeter Team

UPDATE 12:00PST

During the weekend when s30 was doing its scheduled re-boot the Tracker Software failed to reset. The results were logs files piling up and not being processed into the database. We’ve restarted Tracker and s30 is quickly chugging through the backlog. All account should be current later this evening.

As they stated the server is now working and is catching up.

I am not paying Sitemeter anything for their service. Although it is free to me that doesn't mean someone isn't paying for it. It is the advertisers that pay for the service and when it is not working no one is clicking on the advertising either so it isn't just a service issue to just me. In addition I'm sure there were paying subscribers on that same server who were also impacted so there is some expectation of a certain level of service.

Coincidentally, I've been discussing the use of these reporting statistics with a friend of mine, Kent Blumberg, who has suggested that I may want to use the premium service, i.e. pay, to get more detailed information. However, reflecting back on these customer service issues why should I reward poor service by paying for something when they haven't shown that they can provide good service in return. In this case poor service has resulted in the loss or at best the delay of new revenue.

Realistically, the financial impact of this particular incident is very insignificant. However, you may not be so lucky. If your system goes down and it 24+hours before you notice it will the financial impact be insignificant to you? If it isn't you might want to think about what to do about it.

So what are the take-aways from this that we can all apply to our own operations?

Apply some simple monitoring - There are a number of automated monitoring tools available to see if your system is performing as it should and depending upon your situation you may want to consider these. However there are also some simple things your can do. Set up some simple test accounts and have your data center people use them to verify the system is working as it should. They can do this on a periodic basis and also after events such as a re-boot to verify proper operation. Sitemeter data center personnel could have found the problem in the same manner I did. You don't need expensive software, just use the system as a customer would and take a look at the results. You can also do things like monitoring logs and queues. If you see an continuinal increase it may indicate a problem. I don't want to get to far off topic but six-sigma/statistical process control tools can also be monitoring performance.
Give people a way to report problems - As I mentioned I had to look hard for a way to contact Sitemeter and never did find a support contact, just the webmaster contact. No doubt "No Support" is one of the provision of the "free" level but at least consider a way to report issues.
Monitor the problem reporting system - I'm sure my email to the webmaster sat in their inbox until they came into work on Monday and got around to reading it and forwarded it to the operations folks. This assumes that they found out about this through my reporting it but with no evidence to the contrary what else can I assume? If they had a trouble reporting process and if someone in the data center had been looking at it the problem the problem could have been fixed much earlier.
Respond to problem reports - No one likes to be kept in the dark. At least acknowledge problem reports. They could have thanked me for reporting the issue and directed me to their website for updates and progress. In the absence of any communication people will assume the worst which is probably not what you want them to do.
(Perhaps the most important item) View your services from the customer's perspective - If you were the customer rather than the provider would you be happy with the way things are?

Although I'm not happy with Sitemeter's handling of this situation, in the big scheme of things this is not a major incident by any means. However, it does hopefully remind us of how small problems can have big impacts. How well are you set up for this kind of situation? Let me hear how you handle these kinds of problems.

p. s. I'll send the Sitemeter webmaster an email with a link to this post.

If this topic was of interest, you might also like these: