SLA reporting

We have an SLA for a 5 day turnaround of post say 85%. My question is how does everyone count their SLA, is it based on the day that it comes in or the day it was completed, our % is a monthly and a weekly target so our problem comes with the cross over of the month or the week.

For example, if you have 100 items come in on the 31st May and 50 get processed on the 31st May in SLA and then 50 get processed on the 10th June, would you say the May figure was out of SLA (when the post came in) or the June figure (when it was completed)

So how would we calculate a monthly KPI percentage then?

We do look at it daily, weekly and monthly, and each item of post gets a calculated turnaround time, then we say how many items completed in the month was in KPI as a %

I think what it comes down to is one manager wants to say of all the post received in May x% of it was answered in KPI, whereas the way we are currently calculating it, the statment reads more 'of all the post completed in May x% of it was answered in KPI'.

Hi Kerri,

the above statements are all sound. Your manager is quite right, in that measuring the % in is better that the % of completed.

This should provide a clearer picture. In its simplest terms, your current system has a weakness in that if you have missed the SLA on an item of post, simply ignoring it rather than replying late would give a SL of 100%. I'm not saying you would do this, but there is an element where backlogs could become involved that would result in an overstatement of service achieved.

If you migrate to the new measure make sure it is known an expected drop in % will occur, as it is a different measure, and not a reflection of a decrease in the teams performance.

I wouldn't say the new KPI should be the only thing used though. To try and measure a whole day or months performance with a % is limiting. Of course there is a need to express performance in a snapshot, but I find 3 or 4 SLA's to paint a clearer picture, without being too long for the bigwigs to understand. Particularly I would compliment it by looking at reporting on AHT as well (this will give an indication on what performance was, and will be particularly illuminating during times when your % drops. You could have 20% and have replied on average in 6 days, or you could have 20% and replied on average in 15 days.

Hope this makes sense.

Hey Kerri.

Have you considered simply offsetting your month slightly. If you have a target of 5 days, you could measure may's performance based on volumes received, between 26 April - 25th May. Whilst exceeding your SLA in the first few days of the "month" will result in some work done in the last few days of April being marked toward May this would remove the issue of "out of SLA" replies negatively effecting a different month. If you consider that anything received after the 26th May doesn't need to be done till the start of June, this starts to make a bit more sense.

If you receive mail on the 25th May, you have those final five days of the month still reflecting what the agents are doing in the month of May, rather than when the post was received. Offsetting time periods can be a little confusing at first, but in some instances I believe it is the best way, and I haven't met any big problems.

With the idea of doing a frequency count, it isn't without value. What if you turned it instead to a cumulative count, expressed as a %, to say 16% were done within 1 day, 34% were done within 2 days, 78% in 3 days and so on, with 6+ being 100%. This should allow a reasonably clear picture of how up to date things are, and moves away from the issue of a single % counter being a sort of all or nothing scenario (as mentioned in my previous post).

It is also an interesting exercise, even as a one off, for setting a TAT target. We suggested 3 days, but if you can see you only typically achieve 40% in 3 days, but 86% in more days, then 4 days is more viable with your current set up.

The only issue I would mention with this approach is it isn't great for those looking to draw a clear line between good and bad service. if you failed to reply to any letters in 5 days (and therefore would traditionally get 0% service, which would look terrible) but managed to do all of them in 6 days, is that such a bad thing? It tends to blur the lines a little. In a slightly different context I use a measure like this with phone SL. If a contact centre managed to answer every call in 21 seconds (whilst there target was 20) this would be very good service, even though the 0% suggests otherwise.

This approach is a little more realistic, as there is no clear point where the turn around time becomes unacceptable, but it does take quite an open minded manager, and certainly isn't for everyone.

I hope this gives you a bit more to think about. Really I would suggest playing around with some existing data to see how each ones looks, and get a feel for what you might work with your team.