Predicting Your Helpline Call Answer Rate

One role of helpline managers is to manage their workers so that they can answer the most calls possible within the available resources. Even helplines that run 24-hours and have 100% coverage can’t answer 100% of the calls that come in if they have more callers calling in than workers available.

Using a system like Chronicall can give you real-time information on the calls that you answer and don’t and prepare more detailed results (for instance, noting where calls are not answered because the worker is already on a call.)

Given a series of values that are related to each other, regression allows us to predict values where we either don’t have the data or where we want to know the “average” of a piece of data.

For this task, we assume all you have is the data about how many hours your helpline is covered (either in hours or percentages) and the percentage of calls that you answer.

While we can use the regression formulas by-hand, Excel provides simple techniques for deducing the formula. The first step (for the purpose of this article) was to do the calculations by hand to demonstrate. You can see the regression article for full details on how to do this.

Regression By Hand

b = (12*20574 – 263*937) / 12*5817 – 263^2
b = 0.71969

a = 937 / 12 – 0.71969 * (263/12)
a = 62.3101

So our final equation is:

Y’ = a + bX
Y’ = 62.3101 + (0.71969)X

Using Excel

We can use Excel to simplify this calculation. Starting with an Excel spreadsheet containing our X and Y values:

Next, we use Excel’s LINEST function. This requires you to select TWO cells at once. The first required value (called an “argument” in Excel) is the known Y values. In this case, it is C2 through C13. The next value is the known X values (B2 through B13.)

The third argument is whether to set b to zero, or to calculate it normally. Since we’re using the equation Y’ = a + bX and not the equation Y = mx + b, we’ll set it to TRUE. The final argument asks whether we want additional statistical information included, so we set this to FALSE.

So our final equation is:

=LINEST(C2:C13;B2:B13;TRUE;FALSE)

After we’re done typing this, instead of hitting enter like normal, we hit Ctrl-Shift-Enter. This is very important! If we neglect to do this, Excel will only give us part of the information we need. If we’ve done this correctly, Excel will put brackets around the formula, like this:

And you’ll notice that both cells you selected are filled in. The first cell holds the b value and the second cell holds the a value. Putting them into the formula, we have:

Y’ = 62.31024 + (0.719685)X

So, if we want to calculate what our answer percentage will be if we have 21 hours of coverage:

Y = 62.31024 + (0.719685)21 = 77.42

This falls right in line with our expected values, and this technique can be used with any other data where you need to predict values in a linear fashion.