The Problem with (Actual) Volume Data

One of the common mistakes I encounter while working with Planning and Analysis Departments is with their frequent use (and abuse), of actual, system-reported volume data. Yes, actual volume is required to fuel trend forecasting models and to create labor standard reporting, but analysts should take care not to perpetuate negative business performance by using actual data from time periods when guest service fell short or when targeted revenues were not realized. In many cases I observe analysts indiscriminately feeding actual data to their models from time periods of good business performance as well as bad, the result being inaccurate volume forecasting and reduced operational buy-in of their labor standards.

Take for example the case of a large Hotel Front Desk Operation using actual hourly check-in volumes to drive Guest Service Agent staffing targets – and then using actual volume counts reported during an hour when Agent understaffing resulted in line queue timings which significantly exceeded established service targets. Without adjustment the actual data reported during this hour would potentially drive labor standards which would perpetuate the same inadequate staffing levels that yielded the poor guest service performance in the first place! To make matters worse, if understaffing at a Hotel Front Desk resulted in long wait times in one hour then the opportunity existed to “push” a portion of the actual reported volume to the next hour. A labor model driven by the raw volume occurring each hour would then not only erroneously report the staffing levels as being correct in the first hour, it would also incorrectly report labor requirements during the second hour as well.

The answer, of course, is to refine the information being fed to your labor and forecast models by first analyzing the actual data reported by systems against multiple guest service and financial feedback channels (via direct observation, Supervisor reporting, financial reporting, and available guest feedback mechanisms). Make the appropriate adjustments to the actual volume being fed to your models during time periods when understaffing had a direct negative influence on the amount of reported volumes, I would also suggest that you re-label your volume indicator’s name to include the term ‘Adjusted’ so as not to confuse your report users.

Refining actual data prior to feeding forecasting and labor standard models affords an analyst with the opportunity to include Operations, Finance, and Marketing in the process. Not only will your models potentially yield more accuracy, you may also see the buy-in of your results go through the roof as well!