Errors in forecasting execution: what went wrong?
The forecast was right, but what did we do wrong? Maybe this is the question that executives need to be asking, argues Tom Bacon
‘The variance was a whopping 8%. Actual revenue came in 8% below the forecast. Clearly, we need a new forecast model,’ says a senior airline executive.
But do we really? Was the forecast wrong or was the shortfall an indication of a missed opportunity? Did the forecast miss …or did our execution fail?
Often a forecast error is attributed to a unique, unforeseen event or external factor. It could be a competitor’s initiative, weakening economic conditions or even a change in customer behaviour. Many times the variance explanation appears to be outside our control. In these cases, we need to consider modifying the model so that we don’t miss out like that again.
But it isn’t always true that the model was at fault. Another reason for a shortfall is that the forecast was right but that execution failed. Potentially, the forecast reflected what should have happened. Of course, most organisations would rather identify an external factor, and modify the model, than own up to an execution error.
A hotel industry perspective
At a recent Eyefortravel conference in Atlanta, participants discussed a merchandising tool for front desk agents at hotels. The tool provides agents with a prioritised set of upsell opportunities based on each customer’s individual characteristics and tendencies. The tool leverages the growing importance of ancillary revenue to many hotels. Given the proliferation of ancillary services, the model identifies those services that the agent should focus on. Obviously, this depends upon the individual guest; the tool presents that particular set of ancillaries most likely to be attractive to each individual. Should the agent try to sell an upgraded room or a spa treatment or a breakfast? The tool can be used by agents to manage upsell conversations during speedy check-in.
However, when the discussion moved to forecast accuracy – are the upsell priorities presented to the agents truly optimal, we decided this wasn’t the right question. How successful have hotels been in using this tool to drive more revenue? There is, of course, a wide variation in agents’ success with this tool. Potentially, however, the forecast isn’t ‘wrong’ even if the upsell doesn’t occur. The upsell opportunities that are presented based on the model represent analytically based ‘potential.’ More often, however, it’s agent execution that ultimately drives upsell; different agents certainly have varying success meeting the Potential identified by the tool. Despite this variance, arguably, the model is highly “accurate.” Execution-related factors include:
Training: Are the agents across a large hotel network properly trained in the tool?
Recruiting: Turning agents into sales folks is a huge change at many locations; the whole process of recruiting and hiring agents may need adjusting for the new skills required.
Incentives: Upsells can take more time while speedy check-in is also an objective for agents. Incentives may need adjustment – and agents may trade-off objectives differently without explicit policies or procedures.
Of course, another possible hurdle relates to agent interaction with the model. Is the model fully built into the agent’s check-in process; does it truly support the agents? Would agents be able to upsell more successfully if the upsell recommendations were presented in a different format? Would upsell increase if the model displayed more, or fewer, options? Again, the ‘potential’ could be valid but execution may limit actual performance. Agent usability is obviously a high priority for hotels using the tool and should be continually tested and perfected.
Often, firms need an accurate forecast model that accounts for challenges in implementation; such forecasts account for execution issues including, for example, training and recruiting. Financial cash-flow forecasts need to accurately account for execution challenges across a firm.
In merchandising, on the other hand, it is often valuable to identify the ‘potential’, identifying opportunities for sale or upsell. Often, especially when depending on individual sales efforts, performance needs to be measured against what is possible rather than what is truly expected. In this way, execution gaps can be actioned – with new training or recruiting or system modifications. There may be reasons for separately modelling execution hurdles – to determine the impact of improved training, for example – but a forecast that accurately quantifies ‘potential’ remains a useful objective for all merchandising and sales activity.
Tom Bacon has been in the airline business for 25 years and is now an industry consultant in revenue optimisation. He leads audit teams for airline commercial activities including RM, scheduling and fleet planning. Email Tom or visit his website