Analytical Engine

Accurately estimating demand at a granular level is critical for operations management in the travel industry. Traditional estimation models have used historical data to predict the demand. Recently, some companies have started experimenting with a novel source of data to augment the traditional models – the search data from their own websites.
Scandinavian Airlines has been a pioneer in this space.  It is using the volume of search activity on its website for flights on a particular day to estimate demand for the specific routes and flights.  They use this information to guide their operational decisions e.g. which type of aircraft to schedule for the flight segment.  As expected, there was a lot of resistance within the company to make decisions based on these numbers but the web analytics team collaboratively worked with the constituents to gain confidence of the operations team and bring about the change
 Online is already a significant channel in the travel industry.  Web analytics tool vendors like Web Trends and Omniture have made significant inroads and power most of the major travel websites. This means that there is a lot of data readily available for analysis. Travel companies should follow Scandinavian Airline’s lead and start exploring this data to make better operational and marketing decisions.

It is very challenging to evaluate a new product launch and cannibalization without clear baseline metrics. In a previous post we talked about measuring brand performance and cannibalization and hinted that the company should use a set of KPIs to measure the success of the launch. Every company and industry will have different factors (KPIs) which are important for the product to meet before it is called successful. This post presents a framework which identifies the attributes of successful product launch against whom KPIs can be developed.

We like the PERFECT model developed by Michael Burkett of AMR Research, PERFECT model attributes of a PERFECT product launch are embodied in the acronym itself:

* Profitable: meets business financial goals.
* Exciting: creates demand in the marketplace.
* Reliable: performs to defined specifications.
* Focused: delivers the most important product in a portfolio.
* Efficient: makes optimal use of available resources.
* Customer-centric: clearly meets defined requirements of the market.
* Timely: delivers at the appropriate time to satisfy demand.

The PERFECT framework is the first step in developing the KPIs and establishing benchmarks during the product planning phase – which can be monitored at regular intervals after product launch to evaluate its performance.

It would be interesting to have a debate on how FMCG companies define the success of a newly launched brand. Well, for the starters, you need to have some targets. What is a right target to have? Something that makes you break even in the shortest possible time? How long should be the planning horizon for such investments? Is there an investment recovery curve that the brand should have? Should the brand have success comparable to some other success story of yesteryears? And most importantly, how do we study the extent of cannibalization caused by the new brand launch?

There are several such questions that product managers grapple with when they are trying to define and measure the success of a newly launched brand and cannibalization effects.  On a recent project done by Diamond, we started researching similar questions and realized that there isn’t much literature available on this topic. Using the limited literature available, from an analytics POV, again, we came up with one such possible approach -

1.       Define a set of KPIs (such as sales, % of sales in first x weeks, brand share in week 1, aggregated brand share in the first quarter since launch, etc.) to measure the health a new brand. Noticeably, brand launch success will not just have metrics that capture week on week growth, etc. that are standard fare for reporting the health of stable brands.

2.       While product seasonality can be factored in, it’s not possible to factor brand seasonality. For instance, while we know that weekends have higher footfalls in retail stores, or poultry product sale is higher during the evenings, etc. we cannot generalize these results for a newly launched brand.

3.       Identify a set of brand launches by the company historically. (Tricky bit: Getting the right data/date for all the brand launches you want to analyze, and the change in data systems over the years)

4.       Study the post-launch sales curves to identify the duration after which brands start imitating the market trend. Let’s call this duration X.

5.       Build an X-week or month model for the product portfolio pre-launch that would forecast company sales in a product category, assuming no new product launches.

6.       Derive the equation for the product portfolio for the X-weeks of data after the launch of a new product.  Keep the current/historical state of the market/portfolio as a key independent variable.

7.       Derive the equation for the new product for the X-weeks of data after its launch. Keep the current/historical state of the market/portfolio as a key independent variable.

8.       Stabilize the equations in 6 and 7 for a set of launches. Each launch adds to the information around how a company has historically managed its launches.

9.       Forecast the portfolio as well product sales for a new launch that happens on a certain date.

10.   Compare the model run results with the actual performances in the X-weeks to come to assess whether the brand launch can be called a success, and whether there has been significant cannibalization

11.   Caveats

a.       Here, fine-tuning the model would require several business driven assumptions around promotional activity, some identified seasonal variations such as 4^th July, product extensions, other launches, competitive activity, etc.

b.      The choice of model is also critical to the market validity of the forecast.

c.       Need to test out the model on several time windows is essential.

d.      It is still debatable whether an aggregated model is better than an aggregation of models, e.g. combining the results from models built for each brand in a certain category (such as energy drinks) vs. modeling for energy drinks as a whole

e.      Finally, validate the results against the KPIs. For instance, a launch classified as success based on model results should also show strong brand KPIs.

 

 

There are limitations to this approach as well as the big question around the best metrics to measure the health of a newly launched brand, as any marketer would be quick to point. But let’s see, how many holes can you drill through this wall?

ÂÂ

After porting my wireless number to one of the GSM cell-phone carriers, I called the customer care to activate my international roaming for my trip to India. It was a 40 minute long call, and the customer care agent explained that since I had been a customer for less than 90 days, he would need his manager’s approval to activate the service which would take up to 5 business days. When I landed at the Mumbai airport, the service had not been activated and 10 days later it is still the case.

The carrier not only lost potential revenue, incurred cost of a long care call but also caused customer dissatisfaction.

Is there any way for companies to avoid such one size fits all rules trap?

One potential way is to use predictive analytics to design the operational rules and automate front line decisions, to increase both revenue and satisfaction.

One would think that Hollywood is an unlikely application area for predictive analytics. Not quite, according to some recent research.

Research by Wharton professors shows how predictive modeling can be used in Hollywood, to better evaluate the scripts and predict financial success of the movie.

The core idea is to make the green-lighting process, where professional readers cull movie scripts for production, more objective and rules based, using a technique called natural language processing.

A great use case of analytics helping create an objective decision making framework in arguably one of the most subjective domains.

You can get the entire paper from here