Define Measure Acheive. Repeat


Archive for September, 2009

Service Desk Measurement - Be careful what you wish for!

September 29th, 2009

Mini-series on measurements that can come back to bite you…..

The trouble with performance metrics is that they can actually encourage inefficiency, de-motivate resources and result in misinformed management decisions if not thoughtfully designed and carefully monitored!

Take for example an organization that has a published target for their Service Desk’s ability to resolve inquiries at the first point of contact (FPOC rate).

This measure is expected to positively impact two important facets of the service desk – customer satisfaction and cost.

Customer Satisfaction - The premise behind this measure is that customers’ satisfaction is positively impacted by quick resolutions.  After all, who doesn’t like to have their questions/issues resolved quickly and without being transferred or bounced around an organization?

Cost  - Generally organizations have tiered levels of support with resource costs increasing at each level. It makes sense that the more incidents/inquiries that can be resolved at FPOC, without handoffs to more expensive tiers of technical support, the more money that can be saved.

Also, organizations often use this measure to compare themselves to other service desk organizations and to communicate their service desk’s value proposition.

While on the surface this seems simple, practical and a no-brainer there are potential “gotchas”. Consider the following simple scenario.

An organization’s service desk proudly promotes an FPOC rate of 80% or better consistently month-over- month.  Remarkably the Desk has been able to maintain this for 6 straight months! Agent’s performance reports show that individually they are meeting or exceeding their assigned FPOC targets.  Comparisons to other service desks FPOC are favourable and management assumes this is a very positive measure – right?

Maybe not, a High FPOC rate may actually be signalling a lot of repetitive incidents.  Agents at the desk can get very skilled and efficient at resolving the same issues over and over again and the hidden cost can be easily overlooked.

Points to Ponder

While initially customers will be pleased with a timely restoration of the service, their satisfaction will drop quickly if they keep having the same problems over and over again.

Resolving the incident quickly with lower cost resources is, on the surface, efficient; however, if that incident is happening repeatedly what is the total cost of the repetitive incident?

From a value perspective, which desk would you choose?  One that has an 80% FPOC but is continuously resolving the same type of incidents or one that has a 50% FPOC but continuously reviews incidents trends and permanently removes repetitive incidents from their environment?

Simple Tips

1. Don’t look at your FPOC in isolation.  Always look for correlation to satisfaction trends and hidden costs.  

2. Seek out and destroy repetitive incidents. Analyze the types of incidents that are being resolved FPOC and identify repetitive incidents for permanent resolution.

3. Use your incident data to “expose the cost” of repetitive incidents and be sure to report the cost savings/avoidance achieved by removing the incidents right along beside your new, possibly lower FPOC number.

4. Your service desk agents are a great source of information on your top repetitive incidents – tap into their knowledge & experience.  Reward them for identifying repetitive incidents and other improvement opportunities.

One Final Thought with a Bit of a New Twist on an Old Measure

Many desks set high FPOC rates but then do not give the agents the tools or permissions they need to achieve it!

If you have a lower than desired FPOC resolution rate, you may want to consider measuring “designed for FPOC”.  You may be surprised to find that your service desk is actually not able to achieve an acceptable FPOC because of the way your support delivery is designed!

Too often the service desk is not provided with the necessary information (troubleshooting scripts, configuration information, etc) or the necessary permissions to actually resolve incidents at FPOC.

This very simple measure, “FPOC by Design”, reflects an organization that has designed support specifically for each product/service, and is monitoring how well its service desk is performing against achievable targets.

Low FPOC on a product that has been designed for FPOC provides an important area to analyze agent performance and training opportunities.  Conversely, high incident areas with low FPOC for products/services NOT designed for FPOC make be a great place to review the support model and look for opportunities to empower your service desk further.

Measurement is definitely an important part of continuously improving your Service Desk and driving up value and customer satisfaction.  Relentless review of relevant, insightful metrics will keep you at the forefront. Next time let’s look at Mean Time to Restore Service (MTRS)………

Have you come across any measures that have done you more harm than good?  Do you have any really GOOD or really BAD measurement stories to share? I’d love to hear them……

  • Share/Bookmark

Maria Ritchie Uncategorized , , , , , ,

If at first you don’t succeed…improving incident assignment accuracy.

September 17th, 2009

One of the biggest incident management challenges for IT organizations today is ensuring that when an incident needs to be assigned to another support team, that this functional escalation is performed accurately. Assignments may be necessary if the help desk is unable to resolve at first point of contact, or if a Tier 2 technician detects a fault or degradation that they are unable to resolve. It is important to be able to measure assignment accuracy as it is one area where process improvements can really pay off, both in terms of improved resolution times and overall customer satisfaction.

One key element of a successful incident support model is the emphasis on the role of the help desk in functional and hierarchical escalations. There is often a tendency for Tier 2 support teams to want to assign incidents to their functional peer groups directly. This is not to be confused with assignment of an incident to individuals within their team, which is anticipated in some situations.

An example of inaccurate assignment would be when a server hosting team determines that the incident was assigned incorrectly to their group, but on inspection of the incident realize that the network team will likely be able to restore service, and they assign the incident directly to networking. This can often lead to ping pong assignments, where Tier 2 groups pass the hot potato until the incident finally reaches the right support team. And, if your customers receive notification on assignments, a useful and common mechanism to keep them in the loop on resolution progress, they will likely become frustrated and form a negative opinion about the effectiveness of the IT organization.

Why does this happen?

The core reason is that Tier 2 support teams are often composed of specialized, focused resources who have knowledge relative to their functional area. They are typically not equipped with knowledge that would ensure assignment accuracy, and as a result cannot be held accountable for inaccurate assignments. However, there is a perception that sending an incident back to Tier 1 is a step backwards. What Tier 2 may not realize is that if incidents are not returned to the desk, the following direct and indirect impacts can occur:

1) The desk may not be aware it is assigning inaccurately, and long term effects can lead to a more challenging task of changing this learned behaviour when this issue is tackled.

2) The incident may not be correctly re-categorized or classified on reassignment. This can affect reporting and may make it appear that one Tier 2 group is resolving incidents outside of their area of expertise. This makes incident trending for problem management identification challenging to say the least!

3) Escalations due to delayed resolutions may not be initiated unless the incidents are being tracked in real-time. While some enabling technologies perform this activity, they often depend on accurate incident categorization which may not be the case for mis-assigned incidents.

4) Service level treatments may be incorrectly applied, as the initial assignment from Tier 1 may result in the incident being monitored against the wrong service level than what a re-categorized incident should be.

It is not uncommon for the cluster in an IT organization to resist the model of reassignment back to Tier 1. That is expected, but this challenge can be overcome. Aside from training and communicating the value of a Tier 1 reassignment strategy, you may want to employ a top 10 mis-assignment strategy. Record the number of assignments and compare these two an average assignment count that you feel is representative of an accurately assigned incident. If your enabling technology assigns an incident to the help desk on initial save, or if your organization sends incidents back to Tier 1 so agents can confirm service restoration with the customer, you will need to factor this into your calculation. You should also factor into your analysis the situation where Tier 2 resources detect an outage prior to customers feeling a service impact. Then, report on the frequency of Tier 2 to Tier 2 assignments, sorted as a top 10 list by support group, where incidents were reassigned more times than your acceptable threshold. Report on this internally to all the support groups on a monthly basis until your top 10 list represents less than 5% of the total incidents assigned. This will provide both visibility and a motivation for support group queue managers to monitor and address inaccurate assignment activities. If the problem persists, you may choose to report on the specific resources, by name, in this ranking. This may encourage those who are resistant to change to avoid having their name in lights.

After introducing this first step, you should begin to see an increase in assignments back to the desk. This can be troubling at first but this is expected. Ensure that you have established a quality review procedure to address these re-assignments. This can be accomplished through additional training or the updating/introduction of a knowledge base, intended to ensure resources are leveraging assignment diagnostics. This may also result in the help desk requesting more detailed diagnostic information from Tier 2 groups or service owners as means to enhance assignment accuracy. Include a help desk assignment accuracy statistic after you have implemented the statistics at Tier 2. This will encourage the desk to reach out after Tier 2 to Tier 2 assignments begin to drop. The rationale for this staged approach is that both measures are interdependent. Improving assignment accuracy require a feedback mechanism to the desk (re-assign to Tier 1), and Tier 1 accuracy improvements require enhanced information from Tier 2 (diagnostic logic).

If this top 10 approach is communicated and sold with a focus on improvement of IT for your customer, this should not be viewed as a negative. Instead, this may foster some competitive spirit between your IT groups. Other metrics can be added over time. You would be surprised how effective sharing statistics can be on changing behaviour.

Assignment accuracy is just one outcome of a successful incident support model. I will provide some additional benefits to support modeling in future posts…

  • Share/Bookmark

Michael Oas Uncategorized , , , , , , ,