1. A Revenue Management Journey Passenger to Cargo Immediate Applications – Minimal Technology Kathleen Mallery General Manager Pacific Revenue Management and YM Development March, 2009
4. Many Similar Business Attributes Seasonal Low Price Wins Share Low Price Alternatives Hub Complexities Laws of Supply & Demand Service Counts Constrained, Perishable Assets
5. Key Attributes in Both Problems Drive Revenue So why have we not used more passenger tools?
33. Thank you For further details, please contact: Kathleen Mallery email: kathleen.mallery@nwa.com
Editor's Notes
I’m Kathleen Mallery. I live in Minneapolis MN – and I’m loving the opportunity to cheat some days from the winter weather here in BKK. A thank you to Jamison Graff for the invitation. I work for the new Delta in Pacific Revenue Management and Yield Management Development. Prior to the merger I had 18 years with NWA, 10 in I.T. building YM and Pricing Tools, 5 in Pacific Pax YM, 2 in Asia Pax Pricing and RM analysis and the last year in Cargo – frankly re-learning things I thought were absolute truths. My group has responsibility for the pricing and space management of all the flights in and out of Asia. I also have responsibility for implementing new YM (and RM) tools for all entities.
During this presentation I’ll take you through some of my experiences in recently transitioning from the passenger world into a cargo world that didn’t have a lot of the underlying RM theory. I’ll tell you the steps I implemented to apply some RM discipline, the successes and setbacks, and a partial list of items I’d still like to tackle.
It’s easy to under-estimate the “other” side From my I.T. days, I thought I understood the business pretty well. Then I jumped to the business side – and realized what a small percentage I truly comprehended (plus the humbling experience of having to use tools you’ve developed – all vendors might consider that) While in pax RM we’d work occasionally with Cargo on new route viability. We’d always joke afterwards that they always had the same answer to the “how full will you be?” question – “Well you can weight out or you can cube out”. The true complexity of that simple statement was lost on those of us who have a set number of positions – one person, one seat
But we share many business attributes. Very seasonal – though many times opposite Follows supply and demand – US westbound prices are low with excess supply, when flights are withdrawn from a market prices might increase Passenger and Freight alike have to manage through hub complexities. Cargo maybe even more so with ability to split shipments – passengers pretty much need to follow a single route from A to B Plane takes off – revenue opportunity lost We both have to compete with low price alternatives – ocean and truck vs passenger’s options of car or Low-cost-carriers Passengers and forwarders both shop for lowest price – carrier consistently lowest in market will win share And service is key - Customers will pay for superior service at a fair price trade-off
Yet, at least at NW, little of the passenger knowledge was harvested into Cargo RM improvements. There seemed to be more weight on the belief the two businesses were too different.
After my 1 st day in Cargo – I learned of these complications. It was clear a straight steal of pax technology wasn’t going to work Pax forecasting highly sophisticated, but not right Sparse booking volume - much more like forecasting passenger groups – the church choir goes to Italy or they don’t, the school group that always travels to Washington at Spring Break decides to go over President’s Day weekend. We KNOW bookings are inaccurate – 300 kgs 1 cu meter Short or NO booking curve – and systems not set up to pass data and re-evaluate/re-forecast multiple times per day of departure – passenger always has a night to reforecast At NW in Asia, half the time the ‘bookings’ are on paper being ‘optimized’ and not entered until a few hours before tendering Optimization – cargo stations have typically ‘owned’ their flight – not thought much about how there’s a network interaction Pax systems have accurate OA price compares B2B model encourages “spot” pricing for ad hoc shipments Lowering rates doesn’t generate “new” demand – at best it shifts share – at worst it dilutes revenue . There is always a price that will encourage a passenger to take a trip they weren’t going to take – factories don’t make more electronics just because it’s cheaper to ship them
So no we’re back to the cube out vs weight out discussion Passenger weight doesn’t matter – I didn’t realize how easy that made the problem. For cargo: Some shipments weigh more than others Some shipments take more space than others Some shipments pay more per space unit than others So joining cargo meant I had to change a lot of what I knew as TRUE from passenger, yet still use what does transfer
Forecasting Put bookings into system Improve overbooking model – build cancel/no-show data Pay more attention to seasonality – better pre-cancels on freighters Optimizer People are shocked what an optimizer will tell you when they’ve focused on filling every spot as ‘optimizing’ Focus on network - build high-level solver opts Pricing Need rules so there are choices on service Mixture of customers allows for varied product mix and better options to optimize Market was tough on the pricing front – focused on YM first
So now I’ve gone thru all the things we already know we SHOULD do – but maybe still aren’t – how do I get results now even if I’m not even close to having the data for a forecaster and optimizer – OR the money to pay for someone to build it for me. I have eight basic elements to share – I’m sure we could all come up with more
No secret here – YM is not necessary unless planes are full (absent stronger pricing fences) First thing we did was pulled all the roadblocks to selling. RM offered ideas, Sales offered ideas – we implemented all we could. We needed to remove barriers to sell. Goal was to get to the point we were choking on the freight – then figure how to back off from there. Sales needs to focus on Sales – tried to pull back more reporting, back office work to RM Used Inside Sales team for outbound calls and also sales back-office work
Also known at NW as the “BRUTE FORCE” method Review loads constantly and all the aspects of load – not just weight, but capacity, positional load factor and density and whether Ops is cutting freight. You have to know if you’re happy or sad – we had all new cargo folks – were we happy or sad with 75 tons, had we ever done better, what freight did we take when it was better, are charges per position making up for lighter freight. We have a great daily report for all entities – combined now with DL flights. We spend a good portion of every morning reviewing and deciding if action is needed
Feed Sales details – call out winners and losers. I sent an email out to Sales and RM daily at first, now every 2-3 days. I specifically did not make it a ‘formula’ email where I repeated slightly different report stats. It was much more blog-like. Sometimes short, sometimes long. Sometimes funny pictures – anything to make sure people opened it every time it came. I’d call out names with questions to be answered – to make sure they were paying attention. Share the reports – they can’t think they are doing fine, if they can’t see how well other stations are doing. Call them – set up conference calls – ask them for the market intelligence – we can’t RM in a market vacuum Threaten – in a nice way of course – Beg and Plead – we need to pull of the next 2 week of loads of X to make budget – and I don’t want to explain a missed forecast Bribery – Works quite well – I think there’s something in being the mother of 3 boys that helps me with this job We set goals and put out lucrative prizes and cash for station teams that meet goal. Some are more casual ‘contests’ I add to the blog – let’s get 6 more 90 ton freighter days by end of month. Prize for highest 757 weight load in the month.
For passenger flights - We had a semi-accurate capacity forecaster – Sales didn’t have any confidence in it – this stifles their ability to sell RM team had taken to just saying it was always high and put a hard-coded fudge-factor in to drop the capacity to avoid overbooking, service failures Historical densities and bag volumes taken into formula to calculate a “volume weight” Use minimum because our pax planes tended to cube out before weight out Overbooking Technically needs to be calculated off of true demand forecast/cancels and no-shows. You can still overbook. Study pre-departure capacity vs take-off cargo weight - by equipment/market; determine a confidence level and use seasonal averages. Study flights that ‘book’ full, but have remaining capacity at take-off – use that average to allow for conservative overbooking. We just started with a percentage and backed it off if we hit oversold situations – much cheaper to find a way to work with customer to move freight around than perhaps turn down a shipment when you would have had space
An automated forecast is a lot of work – time and money – and you really need to understand what behavior you’re forecasting – booking, tendering, shipment – depending on a carrier’s business you might choose different measures. Better to investigate ‘on the cheap’ first. Work up a spreadsheet that makes sense for the network (this is just an old freighter example). Track freight estimates vs allocations of what that station ‘should’ sell. Send out to everyone so if one origin is short, others can try to find freight to fill the hole
In one slide I won’t explain the LP concepts – I assume all of you are familiar anyway The trick is having an analyst skilled in developing the actual excel model Model chooses right Origin & Destination mix over a hub – optimize by dow or total week. Tradition said filling every spot generated highest revenue. We were actually turning down higher RPP to fill a lower RPP leg hole! But if O&D shipment values are different – some open holes can be revenue positive Also used to evaluate extra freighter stops during fuel crisis as well as broader schedule changes. Very small test network could use Excel Solver. Inexpensive 3 rd Party add-ins available for larger problems
Optimizer outputs “hurdle” rates for each evaluated leg – hurdle rates give a minimum value of contribution that shipment needs to contribute to buy-on to the leg or legs. Re-running the optimizer as the forecast changes is crucial – you don’t want a high hurdle if all the freight has canceled, and you don’t want a low hurdle if the flight is down to it’s last position Best option would be for booking system to accept/reject booking based on it beating a “hurdle” – depending on your system and technical resources – this can be a big project. But Hurdle rates from the base optimizer and tracking of last available positions can be put on shared file access – managed by inside sales, call center, and/or RM flight controllers. Sales then has constant ability to see a rising ‘floor rate’ for hot flights or a dropping floor. Goal is for shipments to beat the hurdle – plus include a few other ‘extenuating” factors of the booking/customer like year-round value, alternate market support, etc. – On the passenger side we’d call these “push” requests….they fare they want isn’t available on that flight date routing, but can YM take the booking anyway.
Communicate Constantly with Sales – I’ve said this already – do it again – and while you’re at it revisit all the elements again.
Not only does a clear, open dialogue between Sales and RM help improve the business – it’s a much more fun business environment to work in – no one likes being the “dreaded YM department”
In a year when demand was up 5-6% NW freighters hit 30% YoY gains in unit revenue. These results were clearly a mixture of FSC, removing unprofitable routes and base RM performance. We’ve estimated the split to be about even between the 3, maybe a weighted a bit heavier to the route removal One thing I can’t overstate is the value of upper management who knew and supported RM – We had the control to try new things, mix things up and we came back with the results
Didn’t have time for the marathon – dire business situations mean quick results needed now, we can’t slow down. Need to find ways to take breathers. Pull back a bit on the constant push – Remember to recognize the wins no matter how small. That forecasting of when a plane will cube out vs weight out is a real problem. Best we could do was give sales data to recognize when two lighter, lower revenue shipments are worth more than one heavier shipment that wastes a position because weight limit is hit We need to avoid garbage in/garbage out to the optimizer – especially when using for routing decisions – our human “arrows” don’t shoot that straight. Need to use sensitivities instead of considering the answers absolutes.
These were items on our list at NW before the merger – merger alignment is keeping us quite busy but we still have a number of RM discipline items to work on and are working on projects in parallel with the merger.
The going forward list has a mix of YM and more traditional Pricing elements Getting back to contracts requires flexibility Tracking and rewarding annual/global value of a customer and of a unique route Extend overbooking into wasted allocation space Freighter ‘packing’ is still very manual – there are simple things that can be done to make this easier Spot pricing is perhaps too random depending on experience of Sales – RM needs to give more guidance Figure out that position vs weight problem without risking excessive spoilage
Manage like and borrow from passenger where it makes sense. Leverage unique elements of Cargo business everywhere else to create new advantages Open communication and fast response are possibly more relevant to revenue growth than expensive systems You can’t stop working on RM just because it’s hard to accomplish Element #1 – Fill the planes in this freight environment. You have to be ready with the RM tools – even homegrown – for when the economy turns around.