We've all thought about it: if only catching customers could be like fishing. Set out bright and early, cast your net and return at dusk to collect the finest catch of the day. It's nice to dream. But winning or retaining customers must deliver greater return on investment, be effective and measurable like never before.
A scoring system aims to integrate your knowledge of your customers and prospects through a prioritisation mechanism.
Lead prioritisation allows businesses to earmark which opportunities are red hot, which to postpone and which to ignore altogether.
Simply put, lead prioritisation is the scoring of prospects and customers according to their potential value to your business and likelihood to respond. It's like understanding which fish would go for your bait.
What do you want to achieve?
It's all about setting achievable goals - do you want your customers and prospects to spend more or is it response rates that you are looking to boost? Establish the reasons for applying lead prioritisation - and be specific. What percentage growth in revenue do you want? What are your customers responding to? Why should they respond?
Identify what is a good lead
Based on your objectives, identify the characteristics that will constitute a good lead. When looking to raise profits, ask yourself: "Which customers have the best spending history?" or "Which customers spent the most during X campaign?" Questions like these will assist in identifying who is most likely to make a purchase. The same principle applies to identifying good leads when you are trying to get customers and prospects to respond.
Source and clean your data
Once the first two steps are established you will be armed and ready to march down to your data controller and demand the data you need. To make life easier for them, make sure you determine where the information required will come from - sourced internally from your own database, or externally from a third-party database? This should be done whether you are targeting customers, prospects or both. Data, such as demographic, behavioural and geodemographic, can all help to prioritise your leads. Get the data cleaned and decide whether or not you will benefit from building a single view of your customer.
You will then need to ascertain whether your data needs to be enhanced using a third-party data set to fill in any gaps.
Identify variables that distinguish good leads from bad ones
To identify these variables you will first need to look at the business objectives and use them in conjunction with statistics to best predict the value of a customer. For example, you may have two 30-year-old women in Bath, one is a high-spender at your store and the other is not - you will need to identify what distinguishes them. Perhaps the woman who is not a high-spender behaves so because she lives closer to your competitor's outlet. The scorecard will also need to be regularly refreshed as customer and prospect behaviour changes.
Build your scorecard
The scorecard is created from the demographic, behavioural and geodemographical data collected about customers and targets. The scorecard can be used to adopt strategies that will target the most relevant groups. In an effort to grow its business and acquire new policy holders, ACE Insurance engaged The Database Group to build a data model to identify future purchasers, detect attrition characteristics and maximise revenue per customer for its telemarketing campaigns. The scorecard model was built using predictive modelling techniques so that ACE was able to prioritise prospects by their likelihood to respond. The results speak for themselves. A much more effective and measurable solution.
Pilot test your campaigns
Like any good idea, it is always best to test it out on a small representative sample before diving in at the deep end. Measure the results of the pilot test and find out if the propensity model had performed as expected. Are your high-spending customers falling into the top end of the scorecard? Are redundant customers sneaking higher up the scorecard?
Any anomalies will have to be refined by going back through the steps above. Once you are happy with the way the model is scoring your customers and prospects, you are ready to roll out en masse.
But, remember to ensure that it's right beforehand or you could waste a lot of money, offend or ultimately lose valued customers - and who can afford to do that?
B2B customer/prospect scoring
The same steps can be applied to score B2B customers and prospects. Sourcing accurate data can be hard work and enhancing available data can be complicated as there are fewer third party datasets on the market. Times are changing and many data bureaux are harnessing the power of the B2B customer, and a wealth of new products is available to help direct marketers clean and analyse their B2B data. According to Royal Mail, 69 per cent of all B2B mailings contain at least one error, which is enough for the recipient to assign it to the bin. The important thing to remember is that quality is all-important when considering B2B data and that there are different behavioural variables which can be relied upon.
Ronel Schoeman is business and analytical consultant at The Database Group
CASE STUDY: ACE INSURANCE
Founded in 1985, ACE Insurance provides global insurance, reinsurance and financial products. Wanting to increase revenue from its family insurance plan product and improve the return on investment of its telemarketing campaigns, ACE teamed up with The Database Group (DbG) to develop a data-driven strategy.
Hitting the target
DbG gained an understanding of why ACE Insurance's customers, from a sponsor prospect pool that had already been telemarketed, had made purchases by evaluating customer buying patterns (RFV - see Jargon Buster, p28) and length of relationship.
This data was used to create an intelligent data model to predict which customers from a similar prospect pool would be most likely to buy insurance policies from ACE Insurance. The model scored the prospects out of 100 according to their propensity to buy, with 100 being the best score.
Two sample groups of prospects were targeted - a control group of records with random scores and a test group with scores of 40-100.
The results were impressive, with the telemarketers achieving an uplift of 50 per cent from the test group compared to the control group. The percentage of policy cancellations was also reduced in the test group.
In real terms the conversion rate for the test group improved by at least 25 per cent, and the campaign as a whole improved by 86 per cent.
JARGON BUSTER
Attrition characteristics
The factors that determine why customers lapse or defect.
Predictive modelling
Statistical analysis using past customer behaviour to predict future behaviour.
RFV
Recency, frequency, value. A method for segmenting or rating your customers.
Scorecard
Defines the metrics used to assign an individual a measure of value to the business.
Single view
The consolidation of disparate data sources into one database.
Sponsor prospect pool
A pool of data supplied to one company from another, as part of a partnership agreement.
Third party data set
Lifestyle/geodemographic/demographic/market data used to add value to your database.