As previewed in the last FiO, Analyse this!, today we kick off our four-part Figure It Out series showcasing the potential of Advanced Business Analytics. The theme will be next summer’s much awaited London 2012 Olympics and each week we will also feature a new cartoon focusing on a specific analytic technique.
For the first week we’ve decided to take a look at Segmentation and how it can provide deep insight for organisations. We’ll examine the benefits and more specifically, how the Olympic Committee may have used it to help with the mammoth task of selling tickets to the public.
But first, with thanks to our Mistry Cartoonist, a brief cartoon to show “what the OR team can do for you”:
Olympics Ticketing Ballot
In what has been the biggest ticketing exercise ever undertaken in the UK, more than 1.8 million ticket applications were received during the 4-week window last month, totalling more than 20 million requests for the 6.6 million tickets that were available for 2012 Olympic events. In an event with such huge demand for a limited supply of tickets, many are inevitably left disappointed and there has been much debate about whether the ballot method used by the London Organising Committee of the Olympic Games (LOCOG) was the fairest. While many are frustrated, it is difficult to think of an alternative way to sell such a high volume of tickets priced at different levels – the Straight Statistics blog suggests some alternatives but without using a complicated system which would confuse the public, or certainly crashing their website using the first-come-first-serve method (as anyone that’s ever applied for Glastonbury tickets can attest to), it’s hard to think of a better method than LOCOG’s.
After celebrating or bemoaning our luck in securing tickets, the process itself is worth examining for the sheer analytics potential that presents itself from collecting some basic data from over 2 million people. Last year, LOCOG offered us the chance to sign up for Olympic ticketing alerts and news updates, in return for a few basic details about ourselves and our preferences for what events we would like to see. Collecting the same few pieces of information from such a large audience presents an organisation with vast possibilities for gaining insight into their customers, something that many do not take full advantage of. LOCOG conducted analysis into the information they collected and published some high level statistics in 2010 which revealed insights such as the large proportion of those interested in the modern pentathlon and track cycling being in the 45-55 age range. By simply asking for each person’s sex, age, region and the sports they wanted to see most, they could see which events were most in demand and whether there were correlations between user characteristics and the sport they were interested in. Statistical packages, such as SPSS, help to provide this insight, so organisations can build a picture of the drivers of demand. There are many techniques employed in gathering reliable data from the public – incentives such as prize entry draws always work well, as well as some clever psychological techniques such as the use of ‘autofill’ in online surveys, which relies on human behaviour to illicit the correct information. For example, prepopulating ‘female’ in the sex field of a survey has been shown to increase data accuracy due to our desire to correct information that’s wrong about us, thus increasing the likelihood of collecting accurate results than if the field had been left blank.
Once data is collected, segmentation can be used to create a detailed picture of demand. Segmentation analysis is a way of dividing up a market to identify trends in it, by dividing customers into ‘segments’ based on their common wants and needs or common characteristics. Using the Olympic ticketing example, customers could have been grouped in a number of different ways, depending on LOCOG’s goal. For example, grouping customers according to their predicted buying power in groupings, like ‘resellers’, ‘prestige event buyers’ or ‘budget buyers’, would help to determine which are the most profitable customers. Grouping customers with similar demographic characteristics can help determine the best channel to advertise to them (e.g. using Facebook to advertise to under-30s). The OR team uses a variety of advanced techniques when carrying out customer segmentation, such as factor analysis, cluster analysis and multiple regression, particularly when the customer base is high volume/high complexity (like the Olympics example).
There are many ways that organisations can realise the benefits of customer analytics and segmentation:
- Marketing and advertising – LOCOG targeted the U16 segment of their customer base by offering ‘Pay Your Age’ tickets to encourage families to buy tickets by offering discounts on children’s tickets. LOCOG could also have targeted campaigns for particular sports where demand appears to be very low if they thought there was a risk of having to offer discounts to attract buyers. Knowing some basic facts about who is coming to an event will also dictate the type of advertising sold at the event and its price – for example, the popularity of, say, beach volleyball to a particular demographic will attract advertising demand from products and services that appeal to that demographic.
- Pricing strategy – segmentation can be used to modify prices according to demand or according to a particular demographic. LOCOG has not said specifically that they used demand to inform their pricing strategy but they did say that they used market research, so they may have utilised demand data they gathered to increase prices for popular events and decrease prices for less popular ones, to encourage ticket sales. Their pricing strategy features the inclusion of a London Travelcard in the price of all event tickets and the absorption of last year’s VAT increase in ticket prices. These decisions could well have been based on insight gained from segmentation about customer requirements, for example, the proportion of buyers travelling from outside London may have influenced the Travelcard decision. The OR team has experience in consulting with a number of large clients on their pricing strategies and building models to support pricing decisions. We built a pricing decision support model for an Irish energy company which enabled them to understand profitability over 10 years and the impact of applying competitive discounts to different tariffs and behavioural discounts to specific customers. We also built a yield model for a mainland European airport which allowed the client to vary the discounted prices offered to the targeted customers and then calculate the attractiveness of each product, the upgrade movement of customers and the impact on capacity and revenue.
- Publicising demand: Although Olympic organisers made initial demand statistics public last year, they decided not to show predicted demand to customers during the ticket application process, making it difficult for customers to decide which events to apply for, if the customer’s buying strategy was to try and get tickets to any event, regardless of the sport. Without this information, customers employed a number of different strategies when buying tickets, depending on their buying power and preferred events. According to a poll of Guardian readers, the best strategy was to apply for a large number of high priced tickets – a strategy which favoured those with higher buying power which has led to some controversy.
Could the ballot algorithm have incorporated segmentation to ensure that each customer segment received an appropriate amount of tickets? Perhaps lesson the allocation from the ‘rich’ segment and increase the ‘budget buyers’. Whatever the solution, I know I was left in the ‘unlucky’ segment…