Step-by-Step Guide: How to check staff levels for answering telephones

1. Choose a suitable period and prepare

It is important to make sure that the data collected is typical and represents the ‘normal’ level of demand. The points below suggest some of the things to look for when choosing a period to measure call length and demand.

Choose a period of ‘normal’ demand

This is simply a matter of avoiding obviously unrepresentative periods such as holiday periods when many of your patients may be away (or if you are in a holiday period when there may be unusual visitor demand). It is also worth avoiding any week that is close to or adjacent to a bank holiday because this too may affect demand patterns.

Capture everyone

You need to minimise the number of calls that cannot get through to the practice. So you should make sure that all of your normal receptionists are in. You might consider extra cover for particularly busy times in the week if you suspect that these are occasions on which patients have trouble in contacting the surgery.

Prepare for the data collection

If calls are to be counted by receptionists, then it will be necessary to be sure that they understand why you are looking at the process as well as exactly what they need to do to help. The work that you are doing is to help them to manage the difficult periods and make sure that they are in a position to carry out their job effectively. So they will welcome your interest. At the very least, it is courteous to explain to them what you are doing, but most practices will find that they are keen to get involved and help.

2. Count the existing demand and measure call length

In this step, we describe two alternative ways in which the key information can be collected. You may wish to vary the approach to collect some additional information at the same time. For example, you might wish to categorise the different types of call received, to understand what the proportion of calls are about. This may be useful when looking at the further options described in the GP Access Guide. If you are doing this for the first time then try and keep it simple – receptionists will usually have at least part of their day that is so busy that additional tasks, especially those that do not address immediate patient needs, are sidelined. It is important that no calls are missed in the count and that the estimate of call length is reasonably accurate, so avoid asking for so much more that inaccuracies result.
Unless you have a phone system that includes a good and well understood telephone reporting system, we recommend that the counting of calls received in each hour of the day is recorded by the receptionists. Use a simple ‘five bar gate’ approach illustrated in the example below.

• Decide which calls are to be excluded (for example, direct dial calls that the receptionists do not answer) but count all calls that are answered by reception staff.
• Devise a simple tick sheet for counting the calls in each hour like that shown below.
• Tip – it is most important to make sure that every call is counted in the right time period, so keep any additional information that you want to collect very simple.
• Make sure that those filling in the sheet understand exactly what they have to do.
• Run a trial for a morning to make sure everyone is clear what to record – and make sure that you include any part-time staff.

Click on image to enlarge.

 

If you have adopted this approach, then you will also need to get a reasonable estimate of the average call length. It is not necessary to record this for all calls during the week.

In practice, we have found that asking someone else to sit and record the length of 100 calls received during representative times during the week gives a sufficiently accurate figure for Applying Erlang’s formula. We strongly recommend that you do not ask the reception staff to collect the start and end time of each call as they are receiving them. This takes time to record and adversely affects the performance that you are trying to measure.

Again, a simple form such as that shown below is easy to devise for the observer to record findings. You may choose either to list the start time and length of call (if a stop-watch is being used) or start and finish time (if a clock is to be used). Most people find it easiest to use a digital device rather than an analogue clock or watch to reduce the chance of errors in recording the various times.

Click on image to enlarge.

 

3. Calculate the required number for each hour

• Typically, most practices have greater number of incoming calls on a Monday (or after a bank holiday) and a reasonably consistent level of demand on the other days. But it may be higher following a half-day closure or on days when the practice is open for extended hours.
• Your first task is to examine the pattern of demand you’ve measured and decide which days are sufficiently similar so that they can be treated as following the same pattern, and which need to be considered separately. It may be useful to prepare a graph of the number of calls in each hour for each day to make this easier.

For simplicity, in the following example we have assumed that the demand was identified as higher on a Monday, but that on Tuesday to Friday it was sufficiently similar for the demand level to be averaged.

Click on image to enlarge.

The above example is for a practice where the average incoming call length to reception staff was found to be 84 seconds. Using the look-up table (included in step 4) the next higher call length of 90 seconds was chosen and a service level of 90% of calls to be answered in 30 seconds was selected. The first and second columns show the times of day and the number of calls received (for Monday and then for the average of the other days). The third and fourth columns come from the look-up table, whilst the final column is the number of reception staff that the practice finally decided on. There are slight variations in the third column. Because telephone demand varies significantly, the practice decided to go for a higher number of receptionists than the look-up table suggested at 12.00 till 13.00 on Monday, for the first three hours, and for 16.00 to 17.00 on other days. This judgement comes from comparing the second (number of incoming calls) and fourth column (maximum capacity) as well as by any ‘local knowledge’ such as how the week in which the count was made compared with the norm.

Important: It is very important that if service levels are to be met with the predicted call volumes, then reception staff must have answering the telephone as their primary responsibility.

Although even at the busy times in the example above, usage only just exceeds 30%, and at times is below 10%, any task given to the reception staff to do ‘in between calls’ must be one that doesn’t compromise answering new calls. If staff are drawn away on to tasks that they cannot drop immediately, service levels will suffer.

4. Validate after implementation

The analysis was carried out based on a ‘snapshot’ of calls over one week. There are a number of reasons why it may be necessary to adjust the staffing levels:

• Demand has increased or decreased since the snapshot (perhaps because of seasonality, perhaps for many other reasons).
• The snapshot has been found not to be entirely representative of the norm.
• There appear to be times when the receptionists are not at all busy – or when patients seem to have difficulty in getting through to the practice.
• There are pressures from staff or the practice to modify the rotas, for example to release time for other necessary tasks.
• Since the time of the initial work the length of calls has changed (perhaps because of training of receptionists) to ensure more consistent, welcoming and effective answering of calls.

Two alternative approaches may be chosen to check that the staffing levels are appropriate.

Recalculate using Erlang. It is not necessary to repeat the full analysis that is described above. Based on the initial analysis, it will be apparent which times of the week might be closest to the tipping point, so these are the ones to look at. A simple exercise to repeat the count and re-measure the average call length focused on these borderline cases is likely to indicate if there is scope for adjusting the staffing levels.

Examine the record. For providers with telephone reporting systems, examining the reports for time to answer the phone reveals whether the targeted service level is being met. For those that do not have such systems it may be possible to ask your telephone service provider if they can set up call logging for a period for a modest cost. That will allow you to understand the numbers of calls, length, response time, and percentage of cases abandoned by hour of the day.

Look-up table for Erlang capacity

This table defines the maximum capacity of a given number of agents in calls per hour, given a service level and an average call length as shown. Should your values be outside this range, this tool will calculate your own capacity for a given number of receptionists.

In the example below, the service time (average length of call) was set to 90 seconds, the number of agents to 3 and the service level set at 90% of cases answered in 30 seconds. The engine has calculated the maximum capacity (52.63 calls per hour, rounded down to 52 in the table) and the average waiting time of 9.37 seconds.

Click on image to enlarge.

In the example within the text the call length was 90 seconds and the desired service level was 90% of calls answered within 30 seconds. This gives a maximum capacity of 5, 25 and 52 calls per hour for 1, 2 and 3 dedicated agents. The table is also used to calculate the number of agents required by looking for the number of agents that will give a capacity that is larger than the expected number of calls – so 28 calls in an hour requires three agents, given the same service level and average length of call.