IT Policies

Lies, damned lies, and statistics

Do you rely on statistical reports and help desk tickets to review your department, or do you look into the ongoing story of support to gauge your sucess?

Call tickets are raised so that, after the event, it is possible to see where all the effort of the support team has been used. So why does the biggest failure get the same number of events logged to the system as a forgotten password?

-------------------------------------------------------------------------------------------------------------------

Everytime somebody calls the help desk, we raise a ticket. At the end of the month the managers sit down and count how many tickets have been raised and how long we spent fixing them.

It has been the practice of the teams that I have worked with to log multiple calls caused by a single failure on one ticket, because to raise an individual ticket every time a user rang to ask why the network is down would take more time than we have to use.

This seems a sensible thing to do, but at the end of the month, unless the managers responsible look into each occurrence, an event that may have caused the whole department to work furiously all afternoon would carry the same weight as a forgotten password or a faulty mouse.

One of the problems we face is the increasing incidence of line managers being remote from the operation. All the evidence they have to work with is based on the reports they take off the logging system, but the spreadsheet does not show how the logs affected the system. Reporting total downtime, number of password resets, and total number of tickets raised may give only a partial picture of what has occurred on the system. It might take a written report from the help desk team leader to give a narrative to the events of the week.

On a ship, the captain keeps a log of the day-to-day running of the vessel, noting everything from the number of sick and injured to the distance traveled, the amount of fuel used, and the number of other vessels sighted. In the event of an inquiry, it helps to recall any information that might at the time seem trivial but turn out to be helpful later.

I advocate keeping a help desk log that records such information as who is at work, how many calls were taken, and what, if any, unusual events took place. This log can form part of the monthly reports and put some flesh on the bones of the normal reports that are generated purely from statistics. Whenever you are assembling reports from stats and making decisions from them, you should remember the famous line said to originate from either Benjamin Disraeli or Mark Twain:

“There are three kinds of lies: lies, damned lies, and statistics.”

7 comments
craftslady2008
craftslady2008

that's why I work by myself, screw the ticket system it don't work

reisen55
reisen55

Outsourcing firms love to dictate and conform to endless SLA metrics. I love, too, metrics - the new buzz word. I used to use inches, but now we are into metrics. Soon we shall be measured in molicules, then atoms, and quarks. At any rate, metrics and measurements usually measure only the obvious: server downtime, call response within 4 hours, from call to fix, etc. BUT what about those hard ones, such as email outages that effect everybody, lost business and revenue. What about work submitted to Bangalore that has to be redone and re-created by the stateside people because it is sooooo badly botched over there. Nobody measures that cost. What about shoddy workmanship for IT solutions, time on helpless desk phone calls waiting while India studies a problem script that has no relation to the real problem? Oh, those are too hard to measure. But employing India is certainly cheaper on that expense thing called salary and health benefits are non-existant. BS in a bigway. We had a good metric in my IT group when WE were the on-the-spot support. It was easy too: RSN or real soon now. Call us, we respond and we fix. No metrics really. Just good success ratios.

prcrimm
prcrimm

Smart move. Documentation, Documentation, Documentation. Life blood of an org.

Gate keeper
Gate keeper

any statistical report peppered generously with the following phrases can be twisted to fit any position. -'false positives' -'margin of error' -'correlation does not equal causation' -'the sample size was not large enough for an accurate assessment '

cmanjunath
cmanjunath

All calls need to be logged if you are following a standard such as ITIL. If there's an outage and you get a lot of calls for the same issue, you need to log as many child cases and relate them to the master outage ticket. So when your manager runs the report at the end of the month, he would know how many Outage tickets were logged and how many calls you got on each of these. Again, this may not be possible unless you are using a good Incident Management tool such as Remedy.

NickNielsen
NickNielsen

A help desk log is an excellent idea, but have you considered weighting the calls? For example, our set-up uses severity codes 1 through 4, with 1 the highest. Essentially, a severity 1 outage stops operation, severity 2 severely impairs operation, severity 3 affects operation, and severity 4 has minimal effects. Set the weights by the number of users affected. In most cases, server or network outages affect all users, printer outages affect many users, and PC problems (usually) affect only a single user. Thus, a server/network outage is severity 1, loss of a network printer is severity 2, and most other calls (bad mouse, can't log in) are severity 3. Most help desk software allows you to assign a default severity code for each piece of equipment in the inventory; the help desk can then adjust this up or down as required. For example, failure of a triply redundant power supply might not carry as much weight as a drive failure in the server RAID array. During reviews, you can now group calls by weight and see why it was so important for the entire department to jump through themselves last Thursday afternoon... :)

ProperName
ProperName

Great idea here Nick. Can I borrow it? Thanks. I currently write every thing down on paper, using the logs to write my end-of-week report. Currently I am my own boss...., but soon I hope to be hiring a couple of full-timers and any tips for the help-desk are greatly appreciated.