Using data to drive acquisition
Marketers today face the difficult task of trying to capture consumers’ attention while remaining relevant, and delivering personalised content.
The challenge is intensifying as media space becomes more fragmented, and consumers increasingly expect to communicate with us at a time, and via a channel, that best suits them.
Our core objectives however remain the same. We want to maximise profitability by improving revenue and reducing costs or to put it simply - get the most bang for our buck.
How then do we find real insights, make smarter decisions, and demonstrate the real value of marketing?
In 2001 two US banks - Wachovia and First Union - were to undergo a merger. Wachovia’s CEO at the time, Ken Thompson, was trying to avoid the dip in customer satisfaction that normally follows mergers, and approved an increase in advertising spend subject to 3 conditions.
He wanted marketing to demonstrate:
- The relative return on each major component of retail-marketing spend.
- The rate of return on overall retail-marketing spend, and
- The optimal mix of retail-marketing spend.
Wachovia’s management accepted this challenge and embarked on a project that saw multiple departments and marketing agencies work together to generate data models. Data on customer behaviour, economic trends, and marketing spend, covering a period of two years were integrated. Data models predicting ROI based on 3 components of customer equity - acquisition, retention, and cross-sell - were generated. The results were measured across geo-economic segments of customers, across deposits, credits, and investments. A robust test and control system was devised to verify the models.
The consequence of such a vigorous data and analytics driven approach meant that not only did Wachovia clearly define the life time value of its customer segments, it was also able to maximise its customer equity by targeting the more desirable segments; offering them relevant products and maximising the ROI of its marketing campaigns. Furthermore, it helped explain the financial performance better, by alleviating the effects of economic downturns or upturns and allowed for tracking of a more financially powerful metric – i.e. customer equity as opposed to satisfaction.
While not quite on the same scale as Wachovia’s analytics project, Royal Bank of Canada conducted an in-depth view of it’s customer base and identified doctors and dentists, who have their own private practice, as one of the most desirable retail banking segments. Revenue per customer in this segmented is greater than 3.7 times that of the average customer. The bank used this information to try and capture more of this market, by targeting them ahead of its competitors. They devised a special offering for medical students and young medical professionals, that included assistance with student loans, loans for medical equipment for a new practice, and the initial mortgage for the young professionals’ first office. Within a year, RBC’s market share of this segment rose from 2% to 18%.
Where Do I Start?
The benefits of using data are clear, but where do we begin?
- Identify the problem.
Rather than looking for a needle in a haystack, successful data practitioners start with a question or a hypothesis, and then use data to prove or disprove it. Knowing what we are trying to achieve is a first step in helping us get there. To paraphrase Einstein “If I had an hour to solve a problem, I'd spend 55 minutes thinking about the problem and 5 minutes thinking about solutions.”
- Define the success criteria.
While we are not looking for a solution at this stage, it is always good to ask this question, in order to be able to:
a) Ensure that the exercise does not remain a theoretical one, and b) remove any misunderstanding between management and data practitioners.
- Estimate the potential value this could bring to the business.
Clearly outlining the impact that the analytics project will bring to a business will ensure better project prioritisation. Comparing expected and actual results is always a useful learning exercise in identifying other constraints that may influence the results (i.e. if implementation of predictive modelling is not generating the same success rate as what was signed off in testing, are there any outside factors that have changed)
- Identify data requirements and understand components of the problem you are trying to solve.
80% of a data scientist’s job is spent collecting, amalgamating, cleansing and transforming data. While not very interesting, having reliable and consistent data is vital in generating meaningful analysis. It is also useful to identify and scrutinize different metrics that play a role in a problem we are trying to solve over the observation period, in order to eliminate any data biases. For example if the problem we are working relates to optimising outbound telemarketing campaign in order to increase revenue and decrease costs, we may want to track attempted records, contacts, sales, average value of sale, length of the call….etc.
- Identify data gaps
The world of data is full of imperfections and sometimes additional information may need to be collected or purchased from external 3rd party sources in order to be able to generate a more reliable analysis. Identifying data gaps will enable a more realistic cost, time, and benefit assessment of the project.
- Test and roll out
- Before we roll out our pilot, we want to make sure that the results of our findings are reusable and can be applied to accurately forecast future outcomes.
- If you would like to know more about using data to understand your customers, or other ways to use data to meet your business objectives, contact Squiz at squiz.net.