Statistical Process Control in Sales Management

11/3/2015

Statistical Process Control in Sales Management

By Timo Sivula
http://www.finngrowth.com

More often than not account and sales managers are artists more than scientists. A good phone sales person can perform a hundred times better than the average sales person. Personal features such as voice, empathy and persuasion skills make all the difference in sales, be it then over the phone or in a business to business sales meeting. Measuring personal features is difficult to do and it is easy to conclude that Statistical Process Control SPC has no place in sales. Especially in a market with strong seasonal variations, SPC may not seem like the most appropriate tool to apply.

It is easy to focus on the personality of the sales person and forget the most important part of sales: hard work. You may have the best skilled sales executive on the planet but if he does not make a single call, does not send a single offer and does not meet any customers, you will not sell anything. Despite all the magic in good sales, the act of selling is a process. And as any other process, the sales process can be measured and improved. I would dare to say that a hard working mediocre sales person beats the gifted lazy one. Hence, with a systematic approach to sales results can and will be improved with Statistical Process Control SPC.

Applying the Ishikawa diagram and XY analysis to find key inputs to sales process

Typical outputs, the Y's of a sales process are sales volume and sales margin. To improve these the inputs, the X's need to be identified. A good way to find key inputs is to use the Ishikawa, or fish bone diagram, listing all possible reasons to less than satisfactory sales performance. Typical inputs could be: number of cold calls per sales person per week, number of offers sent to customers, number of calls made to customer after offer has been sent, number of customer meetings per sales person per week, etc. The X's vary depending on the company and the industry. A properly executed Ishikawa analysis will provide for enough coverage of the inputs affecting sales performance and the key inputs can be identified. Alternatively an interrelationship diagram can be used to identify those X's that influence the Y and the other X's the most.

Measuring sales using control charts

The key inputs found by the XY-analysis can easily be analysed by Control Charts. However, if the markets show strong seasonal variations, these need to be taken into account and normalised. Typical examples of natural seasonal variations are the market changes for agricultural equipment in four season environments. A good way to normalise such fluctuations is to use the seasonalized sales budget as a divisor to achieve a number that is relative to budget and market state. For example: if delivery of harvest equipment happens from June to September, the actual sales actions need to happen early enough, for example during the first quarter of the year. More cold calls and customer meetings are planned in Q1 than in Q3. The measured number of sales calls is divided by the budgeted number of sales calls. The results is a relative percentage that is comparable to all other months of the year, provided the budget is properly done.

For sake of simplicity in the following example, lets assume the product to be sold is a household product that is sold door to door. It is fairly obvious one key input to the sales process is the number of customer visits per sales person per day. A commonly accepted standard sample size for control charts is 30 samples. Assuming a 5 day working week we need 6 weeks to gather necessary data for the sales persons. As every process, the door to door sales process will exhibit variation and a mean. Plotting the results on a control chart we will quickly identify mean, Upper Control Limit UCL and Lower Control limit LCL and whether the process is in control or not.

Analysing measurements, common and special cause variation

By definition all processes have common cause variation and special cause variation. Common cause variation is very difficult to eliminate, whereas special cause variation is the key focus area for improvement work. Any sample outside the control limits, or other defined rules, is caused by special cause variation. Let's take the example of our door to door sales person. He has an average of 6 customer visits a day and UCL at 8 and LCL at 4 visits per day. In the analysis we find one day with 0 meetings and one with 9 meetings. Looking at these two closer we may find that he was able to make 9 visits one day because he was working in a tightly populated area and the 0 was caused by him being ill that day. Both are fully valid special causes. Medical leave may or may not be worth its separate control chart to make sure it stays in control. The fact that the density of houses affects the number of customer visits may be used as an input to the next phase where the sales process is to be improved. We do not worry for the variation that happens within the control limits at this point.

Using Statistical Process Control analysis results to improve sales process outputs, the Y's

The results from the control charts are interpreted and clues for improvement ideas are sought for. The key suspects for improvement areas are the out of control samples. If there are a lot of absences due to medical reasons this can be difficult to change quickly. A program to improve the health and job satisfaction of the sales personnel should probably be started. A faster way to improve Y is to plan the sales work into areas with denser population to enable the sales persons to spend less time moving from customer to customer and more time meeting customers.

In this simplified example the key improvements to the sales process are a) the setting up of a job satisfaction and health program for sales persons and b) improving the planning of the sales route for the door to door sales person. Separate Kaizen projects are set up for each of these improvement areas. A good project brief defines the start and end for the Kaizen projects, which also defines when the next measurement is to be done.

Confirming improvement results using control charts

Once the Kaizen projects targeted at improving the inputs, the X's of the sales process have concluded, a new sample of the inputs is taken. In our example 6 weeks of customer meetings per sales person is measured again. If both Kaizen projects, the one focusing on job satisfaction and the other on sales route planning, succeed the mean of the new sample should show a clear improvement and the special causes identified in the previous measurement should be absent. Provided no new special causes have appeared during the improvement projects, the process should also show no out of control samples and thereby by definition be in control. Now is the time to have a look at the Y, the output. If the selection of inputs to improve was done correctly and the Kaizen projects have been successful, an improvement of the sales volume and margin should be visible. The statistical significance of the difference is confirmed with a statistical test suitable to the sample type used, using e.g. Minitab or SigmaXL.

Sustaining the results achieved using Statistical Process Control

Once the improvement of the output has been confirmed, it is time to create an Out of Control Action Plan OCAP. The Upper and Lower Control Limits of the Control Charts are mathematically calculated to indicate process stability, they do not reflect process specification limits. Hence, the Sales Management sets specification limits for the inputs, i.e. how many sick leaves can be tolerated and what are the Upper and Lower Specification Limit for them. In a similar manner specification limits are set for the number of customer visits per day per sales rep. The sales assistant is given the responsibility to weekly update a range chart measuring these two inputs. The charts are reviewed in the weekly sales meeting and the sales director presents them in his monthly report to the company steering group. Actions are agreed on how to react when specification limits are exceeded. The Sales Director makes sure those actions are executed without delay if an out of control situation is detected. Over time the new level of sick leave and daily customer meetings will become the "new normal".