A look at the analytical methods applied to visit data in order to spot trends within said data.
Looking for Patterns Over Time
When plotting visit figures over time (typically week by week or month by month) for a sustainable trend to emerge, there needs to be movement in the right direction over a number of periods, and not just a ‘blip’ for one unusual week.
Clustering Visits by Location
By clustering visit data by geography, trends can be identified as local or widespread. For example, if there is a drop in footfall to a particular town but numbers remain stable at other locations, this identifies a site specific issue as opposed to a wider trend. Analysts can therefore avoid drawing wrong conclusions by only viewing part of the overall picture.
Flagging Anomalies Against a Baseline
As with any other data set, there is a normal range for all the data collected by a Data Analysis Company. This will have established a baseline against which individual readings can be compared to identify anomalies. This is particularly important for identifying errors in data recording, for example a low visit count on a quiet Wednesday that is below the established range for that day of the week. More detail on how statistical data sets are structured and published is set out in the published guidance.
Cross-Referencing Frequency Against Outcomes
Frequency on its own is generally of little value. However, by correlating frequency with outcome data, such as sales figures or even the compliance data for a set of shops, it is often possible to establish whether more frequent visits translate into more sales or other desirable outcomes.
Segmenting by Site Type
When tracking the data from different locations (i.e. retail parks, high streets, etc. with standalone sites) it is important to remember that each location type will have its own set of trends and that analysing the data from all locations as one will obscure the real trends.
Spotting trends is mostly a matter of asking the right questions of the data, and drawing conclusions from it afterwards.