Revenue management is a very complex algorithm that takes into account a lot of variables, resulting in a really hard task to reverse engineer and predict how much our competitors will change prices.
When considering OTAs, they could also change their markups as often as they want and we can’t predict their moves.
Doing competition analysis and market benchmarks there’s always the question of how many hours or days of past history we should take into account to make conclusions. If we take too much time maybe the prices have changed and the insights are no longer relevant. But take too little time and there might not be enough data in order to have statistical value to make data-driven decisions.
Experiment
That is why we decided to take a few routes and monitor the prices hourly to see how many times they changed during a given day, always searching the same exact parameters. Data was collected for a whole business week, meaning from Monday to Friday. Results are from Kayak to consider both airlines and OTAs prices for the test purpose. The dates searched were always for next year, to minimize the impact of covid-19 uncertainty.
Results
Here is a list of the four routes we used and how many price changes were detected during the full week.
- NYC — LAX: 2 changes
- NYC — LON: 5 changes
- NYC — SAO: 1 change
- NYC — TYO: 1 change
It seems that for most routes price changes an average of once per week, which means that for a market benchmark considering past seven days data is a good place to start.
Worst case scenario prices appear to change at most once daily, so in order to do some more specific dynamic pricing strategy you should consider looking at the past 24 hours and you will be making decisions based on recent data with a very low probability it changed during the last few hours.
Selecting the right time frame for your competition analysis is very important because it’s obvious that the more data the better, but since there’s always a cost associated with collecting the data, you should be wise and look at as much past time as possible keeping an eye on price changes and expired results.
