Hi Nina! We use average speed of answer (ASA) as our service level metric for our live chat channel. Besides measuring customer satisfaction through the use of surveys, we measure ASA on a daily basis to make sure we minimize abandons. Why? Because chat is a very efficient channel for us! Our live chat reps are configured to handle a minimum of three chats simultaneously (some can handle as many as five at a time!). This make chat a very productive way of handling customer support volumes for our site!

Hi, yes there are many metrics you can look at: -Service level -Concurrency (# of chats handled at the same time) -AHT I would say that those are the three most important metrics with regard to chat. Depending on the scheme that your company uses to pay or to be paid (I dont know if your are a Vendor or Outsourcer), concurrency can help you improve your efficiencies and productivity by telling you how many chats your team handling simultaneously. Beware that the increased number of chats handled at the same time has a direct impact on your qualitiy and satisfaction metrics. Hope this help you.

Remember that each one of the 100 agents will need two days off per week -- more if you are doing any creative shifts such as four 10-hour days or three 12-hour days. This will reduce your available staff each day. Also take into consideration your shrinkage due to call outs, vacation and traiining, not to mention breaks and lunches. All of these factors must be taken into consideration when creating a staffing plan. In your example, the first group is working an eight-hour shift from 7am-3pm, and the second group is working a nine-hour shift from 3pm to midnight. I don't know too many agents that will let this discrepency in the number of hours worked go by without voicing their objections, especially if they are on the second shift. Use of a good workforce management program will help you determine the exact number of agents needed to reach your goals. If you cannot or do not have access to this type of program you can use the Erlang C method to get as close as possible to the ideal number of agents per interval.