Statistics for measuring people's performance

[Antonio Valle_1]

Hello!

Althought the subject is a little "hard", I would like to comment with you what do you think about what statistical values are good and will measure a HelpDesk performance with justice.

For example, we have a "First Contact" time, wich is the time between the registration of a call by the operator and the call to the customer by the specialist.

If our objective is to provide a 5 minutes first contact time, and we want to pay money to our specialists if they get this objective, wich statistic value is good for this measure?

If we use the mean, an outlayer can break the statistic very easyly...

Antonio

[Cheryl Griffin]

Just a few thoughts:
First Contact is a good measure as to the responsiveness to a request, but if the response does not add value, what good is it?

In addition to first contact, there is "time to resolve". Again, if the resolution was quick but is not of value, what good is it?

Then there is "quality of response" which is highly subjective. Maybe the answer is right, but it's not what was expected. How can you measure or who should measure quality?

[Antonio Valle_1]

Well... I was talking about statistics, but your thought about other metric is interesting.

We use mainly 3 metrics: time to first contact, time to escalation and time to resolution.

We only can measure quality in terms of change: a good quality change is that change done within time scheduled, within the budget assigned (hours or money), with the lowest impact to the customers AND a change that causes no aditional calls or problems.

A metric for the "quality of the response" is *very* difficult to find, and I think that quality needs to be evaluated from the customer point of view, and not from the technical one. Using this approach, a good customer satisfaction survey and tracking the evolution.

Antonio

[Mike McKinlay]

Statistics like this should be used only to identify whether someone is performing at expected levels. Tuning compensation to it creates pressure to beat the statistics, which may not be the best way to serve your customers or employ your staff.

For example, one HelpDesk operator received 10 calls in an hour. He cuts a ticket as quickly as possible and gets out to answer the next call. As a result, the time from the inbound call to the second level call is reduced for all ten calls.

A second operator also receives 10 calls in an hour. She spends a few minutes in troubleshooting on 8 of the calls and as a result does not escalate five of those issues to second level. Her average time escalate an inbound call to second level is higher because she spent more time with the customer, who probably received better service. Yet the simple metric of time to escalation doesn't adequately measure her performance, which not only helped the customer right away, but left the second level staff out of the process, meaning they could work on other problems.

The first operator, having found the key to success, exploits the statistics for his own benefit. The second operator, concerned more with helping the customer, ends up hurting her performance review by not measuring up to the bar.

In a call center taking orders for Yanni's latest albumn, base statistics may provide a good floor for performance measurement. In a less cut-and-dried environment like a general Help Desk, where almost any question can get asked and expect an answer, statistics cannot measure the difficulty of the problem, the effort expended to resolve it, or the quality of the answer. You need more subjective information for that part of the evaluation.

[James R. Ferguson]

Hello all!

A few of my thoughts...

As a scientific, technical person, I value Cheryl's "quality of response" above all else; and yes, that's highly subjective. On any help desk survey this would be the area to which I would look to measure the desk's return on investment. As a user of a service, I'm willing to be somewhat patient if I know I can expect and receive value, a priori.

One of the metrics my organization's help desk tracks is "call abandon rate". It's "good" if a Help Desk answers a new call before the new caller tiers of waiting, even if that means "abandoning" the original caller to put that original call into a service queue. This translates as "good" for Help Desk statistics -- answered calls are high; the abandon rate is low; and you can advertise that everyone gets instant gratification!

Conversely, I might argue that a higher "abandon rate" is actually "good". After all, it suggests that perhaps more staff is needed; or that more education of the users of the service is required; or that users are receiving a high quality of response to their questions in the first place. This is essentialy the theme that Mike discusses above.

What's that old adage? Figures never lie, but liers always figure. You really have to analyze your data, and you really have to be careful when considering what your measurements mean. It sure seems like a good, old, subjective questionnaire is a great way to measure just how effective a service you're providing.

...JRF...

[Joseph T. Wyckoff]

There is a very good measure of customer satisfaction...that is re-enrollment rates...the only problem with this measurement mechanism is that once you have lost the customer, you probably won't get them back.

However, a pay as you go service model does inherently capture this data...customers only call when they think they will get excellent, relevant service.

[Rita C Workman]

I am now so glad I don't have to review the Help Desk staff performance here.
I guess I would offer this....Have you tried doing what others do. I mean keep your internal statistics (in whatever fashion you feel is most fair and relevant). But to truly evaluate Help Desk performance, is to me, tied to the word Help. Did they really Help?
So....I would suggest some sort of follow up survey back with the customer. Doesn't have to be every call..just random... Make enough of these to check and you will find out who was really helped...and who really did 'HELP' and not just 'play the system'.
Does this take time?...Yes...Is it a bother?...Yes....
But how much time and effort is being spent trying to figure out a system that can't be gotten around. And if your looking for customer satisfaction .. then ask the customer if they were satisfied.
Just a thought,

[fang hai]

Maybe I am late. But I just join the fforum & interest in this post.

I think there will be 4 stages in managing a customer call, I call it 4C -- Connect, Complete, Confirm & Call Back.

Connect is to measure the time between customer call in to the first technician ansew the call, or we could call ir response time

Complete is the problem resolution time includind escalation

Confirm is to confirm with the customer the problem is really fix or not after the technician close the case. It could be measure by case re-open rate

For call back, it just like a short random survey to see what we can improve & if there any other isuues we should follow-up. It would be better if the survey is done by third-party.

[Devbinder Singh Marway]

I agree with cheryl , ( if theres no value to the call , what use is it?) and I believe the value should be long term resolution and not a five minute resolution.

Measuring statistics against individuals makes the work environment unpleasant. The main focus should be the TEAM , .i.e. IF someone is praised ,the TEAM should be praised. If you have a good TEAM, knowledge is passed around, there is no real reliance on other people because each person in the team are capable of doing each others task, so call response times , resolutions etc.. would be much quicker .
As we all know, the customer has the last say , and we here have service delivery managers who liase with the customer regarding the business , and end result is a report is produced showing , server uptime (100%)
User avaliablity ( 100%), Performance of the server (100%) , Batch Job success (100%)
This report is produced every month, and any problems are pointed out at this stage ..

Just a few thoughts

[Willen Teoh]

Hi guys

I am tasked to do a Helpdesk Productivity Model, does any one has a sample templates on how you guys measure the people's performance etc? Many thanks.

[Ravi Sodhi]

Hello! Just a few quick thoughts. In another envirionment although it was usefule to have start and close times. We developed a long list of types of problem communicated by customer vs problem found. We also looked at type of problems by individual. We were able to focus on several problem areas, develop communication collateral and educate customers via FAQs, to help them help themselves - which reduced incoming calls in those focus areas. By looking at problem type, time to clear by individual - allowed us to implement mini training, brown bag sessions to build skills and capabilities in the Helpdesk. It also allowed us to recgnize individuals who clear sets of problems rapidly, this helped their self esteem at being singled out for positive recognition, it also helped establish them as role models within their group to be sought out as mentors or when one encountered problems in a specific area.

The danger of focusing only on time the call came in vs time completed, is that some Helpdesk folks may avoid certain types of problems due to difficulty or handling one large problem made them look bad against a team member who did three problems in same time. Such singular focus can lead to troubled behavior where service reps may hang up on customers due to problem difficulty, thus when customer calls in again they land another rep. You also want to avoid behavior where rep encourages to close out this problem for now and encourages customer to reopen ticket the next day. Such practices make the rep's numbers look better!

Another approach we had used was to capture time to close but add another category around what service did we provide to the customer - mentoring,troubleshooting, faxed document, create custom info that was emailed etc. This allowed us to move groups towards a services orientaton in technical environments. So on a montly, quarterly and annual basis w would proactive communicate to clients and management, the vlaue we provided opr created. ie. we handled 345 service units last month, highest area of call in interest was "printers". Such data allows us to work with problem source to help redesign, improve product, its instructions etc and provides marketing feedback to product or service developers from ease of use, suggestions etc. It also allows us to share with management, team call volume is increasing at 34% monthly, please send more headcount etc.

It is all about running IT as a Business!

[paul courry]

Talk about nailing jello to a wall.................

Everyone around here is expert at managing their statistics and avoiding tackling any problem that might damage them. The favorite technique is to transfer the call to the person making the call in the first place.

I personally think that there is no meaningful method of evaluating performance except by the old fashioned method of getting familar with the person and their work then evaluating them.

BTW, I'd love to hear from HP management their experiences in managing the HPRC and what they have learned.

[Uffe Gavnholt_1]

Hi,

I couldn't help my self thinking about a seminar system as the ITRC is using...

Why not let the customer evaluate the quality of his solution, and measure your employees on the avrg customer rating?

cheers

Uffe

[Neil Shapiro]

Some very interesting metrics mentioned here! I particularly like the idea of whether or not the user was actually helped being a consideration. That, after all, the raison d'etre for a Help desk! One member proposes a survey to users to garner their opinion and that is certainly something that could be pretty easily implemented. However, surveys lead to various problems what with delays in response (or even the fact that most surveys garner well under 10% response rate) or responses that are not well thought out. It occurs to me reading this that most Help desks do track one metric which in many cases could well measure just how helpful that "First Contact" might have been -- the number of repeat calls to solve the problem. Whatever points or measurements are awarded for speedy first contacts and other positives, perhaps a negative value could be assigned to either each repeat call on the same problem -- or to some threshold of repeat calls which would invoke the negative measurement.

[Uffe Gavnholt_1]

It's true an issue will be getting the actual responses from the customers.

I would say a way around it is first of all to let the customer know the consequences of his response. 2nd the engineer will be very interested in making the customer provide his feedback (in case of a happy customer). 3rd it will slip from time to time, but no response must be treated as an avrg, meaning neither good or bad for the engineer.

I havn't seen anyone using a system like this, If anyone has experience with this kind of measure, I would appreciate some feedback.

Have a good day

Uffe

[Terja]

Having being 2nd and 3rd level support for 2 years behind a helpdesk. a better measure would be "how many calls can the help desk answer quickly". If the customers problem is solved in the initial contact, the customer is more satisfied, rather than having to think thru the "oh its just a receptionist" attitude.
Frontline should be allowed the tools and the time to solve the calls. ie. make the max calld duration 2* the time it takes to record the call. probably 5 min per call would be ideal. If at this time the call cannot be actioned pass it. customer will then understand.