With record numbers of new routes set to launch in 2015, how do you get your forecasting right? In Part 1 of a new two-part series Tom Bacon takes a look

Anna.aero (or Airline Network News and Analysis) tracks new airline routes launched worldwide. It predicts that 2015 will see a record number of new routes – even more than the 2,550 new routes launched by airlines worldwide in 2014. 

Most revenue management approaches– which are fundamentally based on historic trends – are not designed for totally new routes. They must be actively managed to produce satisfactory revenue results. How do you RM a route that has no history?

The textbook approach for new routes is to select a ‘proxy’ route, an existing route with history that is likely to mirror the new route in key attributes (full fare percentage, business/leisure mix, booking curve, day of week patterns, seasonality). 

For example, Qatar Airlines will begin service from its hub in Doha to Amsterdam this year – Qatar already serves a number of European destinations including Copenhagen. What existing destination is expected to behave in a similar way to Amsterdam? The RM managers can point their RM system to such a ‘proxy’ route as a way to initialise booking patterns and fare allocations for Amsterdam.

Any such ‘initialising’ will be overridden as the system compiles actual performance. As the route gains its own history, its own peculiarities will slowly drive the RM system to establish more accurate forecasts and thus more optimum, market-specific allocations over time.

Slowly….probably too slowly! Each new route requires substantial intervention to help speed progression toward the ‘optimum’. There are a few tools to aid the analyst:

1.  Historic weighting: RM systems typically use a form of exponential smoothing on history to determine trends. The analyst can adjust the weight of three-year-old history versus a one-year-old history, for example. For new routes, a smaller coefficient for smoothing is recommended, allowing the system to more quickly adapt to new information.

2.  Forecast variances: The proxy route comes with both a demand forecast for each bucket and an uncertainty calculation (historic variance or standard deviation).  Obviously, the forecast uncertainty associated with a new route is much higher than for an existing route. To calculate proper allocations, the RM system algorithm would ideally acknowledge the greater uncertainty initially. An optimised result in a market of greater uncertainty – given everything else the same -- would generally allocate fewer seats to higher fare buckets, for example, in favour of taking ‘a bird in the hand’.

3.  Use of business rules and flags: Even in the best scenario, a new route has differences versus any chosen proxy route. Each key attribute needs a business rule or a flag that will highlight differences in performance versus the proxy route, for example:

• Are we tracking way behind in booked load factor 30 days out?
• Can we obtain ten load factor points of high fare traffic close in?
• Is the day-of-week pattern really consistent?
• Do no-show results match the proxy route performance?

As the analyst begins to feel more comfortable with the actual performance of the new route based on such flags, he must intervene to speed RM system adjustment. He can help the system move toward the market-specific allocations necessary to optimise revenue results for the new route.

In conclusion then, the record number of new routes expected to be launched by airlines in 2015 will challenge revenue managers across the industry. Use of proxy routes will help initialise RM system allocations but RM analysts must intervene to help their systems converge more quickly onto more appropriate route-specific parameters.

Tom Bacon is a 25-year airline veteran and industry consultant in revenue optimisation. 

Questions? Contact Tom at tom.bacon@yahoo.com or visit his website http://makeairlineprofitssoar.wordpress.com/