Revenue maximization with implementation variations of unconstraining methods in a semi-restricted fare environment

In this article, we use the sophisticated PODS simulator to examine the revenue impact of two implementation variations (‘all closed’ versus ‘zeroes only’; ‘multiple estimates’ versus ‘single estimate’) on three different methods of unconstraining – Expectation Maximization (EM), Projection Detruncation and Booking Curve. Owing to the competitive nature of PODS (two airlines competing head-to-head for customers) and its allowance for customer choice, we are able to assess all the implications of both implementation variations in combination with these unconstraining methods, including the impact of spill, upgrades and recapture. We find that for the first implementation, the optimization engine that is currently being used dictates quite different revenue results when changing implementation techniques. In one case (under leg optimization), we should switch away from the ‘zeroes only’ approach and in the other case (under DAVN optimization), we should not. For the second implementation, we find that generally it is safer to use ‘multiple estimates’. There is one exception – for EM unconstraining when both carriers are using leg optimization, it is best to switch to ‘single estimate’. When using realistic booking data from major global airlines to calibrate the PODS D6 network, we show that upgrading the unconstraining process with these implementation techniques can lead to revenue gains of 0.3–1 per cent.