Establishing a data collection plan
Vital element of improvement program - process involves:
1 - Selecting type of data
2 - Assess capture point
3 - Define sample size
4 - Understand definitions
5 - Define Capture method
6 - Collect Data
7 - Analyze
Establishing a data collection process should be seen as a fundamental step at the start of any improvement activity. A data collection process ensures that a project can efficiently and accurately collate data enabling the improvement team to measure and establish a baseline of current performance whilst quantifying (and proving) later improvements. Without accurate and timely information, improvement projects can flounder, change agents may guess at fixes and resultant solutions be inappropriate.
In the context of an improvement plan, data collection should be seen as a process with quantifiable steps. Whilst there is no hard and fast rule best practice would indicate that it is advisable to initiate a data collection plan. Such a plan includes a number of commonly used steps within a data collection process, including the following.
1. Select the type of data that will be used
There are two common types of data available to improvement projects:
* Continuous data such as delivery schedule adherence or cost.
* Discrete data such as errors recorded.
Typically discrete data cannot be broken down or subdivided (i.e. you can’t have a fraction of an error). Think through which metrics best represent the picture your are establishing – try to use continuous data as this is likely to be more readily available and more suited to be tracked over time.
Consider how the data will be segmented, for example, where looking at time periods will the data be segmented into days, weeks, months or years.
2 – Assess where data will be captured
It’s likely that data can be captured at various stages of a process - typically there is input and output data available, and the project will need to assess which is most appropriate. Commonly where the performance of a process is being analyzed it’s best to record output measures (the cause factors however may arise from input). Assessing the input diagrammatically can help here as can determining the relationships between input and output data (i.e. does the input data cause the output)