Customer scoring and developing scoring models

Customer scoring allows you to categorise and rank your existing customers and prospects based on how likely they are to respond favourably to your marketing efforts. By scoring your customers, you can focus your marketing activities and campaigns effectively and increase their ROI.

Remember, each of your customers is different. While some customers may make small purchases on a monthly basis, others may only purchase one large item a year. Similarly, some customers in your database will have been loyal for years, while others may have only made a single purchase several years ago.

Due to this, some customers and prospects in your database will be more receptive to your marketing activity than others. For this reason, to maximise the ROI of their marketing efforts, successful businesses use customer scoring models to evaluate the likelihood that an individual customer will make a purchase based on the data they have available. They’ll then use this data to target customers that have the highest potential to convert.

By scoring each of your customers against criteria that are specific to your business and its goals, you can improve your marketing performance. This is because the process will allow you to focus all of your marketing efforts on reaching customers with the highest conversion potential. By ignoring those with low conversion potential, you’ll also save time and money.

If your business uses direct marketing to contact prospective customers or buyers, then it’s likely that you’ll have a large database of information available. As a result, if you continually market to all of these prospects every time, then you’ll find that your marketing campaigns are incredibly expensive and very ineffective. This is because you’ll be targeting a large number of customers who have already decided against using your business. On top of this, you’ll also be providing a large number of people with irrelevant information that will make them less receptive to further communications.

However, if you have a large database and you apply effective customer scoring models to it, you will be able to easily identify which message or product is right for each customer in your database in seconds. Following this, you can ensure that you only send your customers targeted information that’s relevant to their needs.

What is customer scoring used for?

The idea of scoring your customers is based on two key principles:

• Each customer is different: Every customer in your database is different from the last. Looking at factors like average basket size, shopping frequency and customer satisfaction rates will show you this. As a result, rather than targeting every customer with every campaign, you want to focus your efforts on the customers that shop regularly, are satisfied with your company and have a high average basket size. Customer scoring is vital for this process and can provide you with key insights
• Priorities are important: By drilling into the data and finding the customers who are most likely to make a purchase, you’ll improve the success rate of your marketing campaigns and maximise your ROI

There are a number of different reasons why marketing managers use customer scoring and modelling. The exact reason why your business might use customer modelling will depend on the goals of your business. However, although customer scoring can be used in a number of different ways and for multiple purposes, it’s usually used to identify customers in your database who should be included in your marketing activity.

Overall, businesses usually use customer scoring to:

• Target marketing at people who are most likely to take up an offer or buy a product. This can either involve identifying prospects who are not yet customers or generating additional revenue from existing customers
• Predict how likely it is that a customer will have a high lifetime value (LTV). If these customers can be correctly identified through modelling and customer scoring, they can then be sent highly-tailored marketing communications and receive special incentives
• Predict customers who are most likely to lapse. When these customers are identified, they can be contacted with incentives before this happens

After the scoring process is complete and you’ve attributed scores to each of your customers, you’ll also be able to:

• Highlight key characteristics that are most prevalent in your customer base and use this information to source more customers with the same characteristics
• Identify areas where you’re overrepresented and underrepresented
• Compare groups of customers who have responded to different campaigns and use this information to guide your future communications

Due to this, it’s clear that there are a number of benefits to using customer scoring methods, such as:

• Increasing response rates from your marketing efforts by targeting customers more effectively
• Learning about the profiles of your best customers so you can effectively target more customers that meet this profile
• Identify markets with the most potential
• Discover your most responsive clients and target them with upselling opportunities
• Detect groups of customers who have similar characteristics
• Process millions of records in seconds
• Build customer scoring models and campaigns in one integrated environment

Developing customer scoring models

In order to score your customers, you first need to create accurate customer scoring models. By using effective modelling, you’ll be able to identify your ideal target market.  
Customer scoring models compare your ideal target group of customers to the rest of the customers and prospects in your database. By doing this, the model will be able to highlight the shared characteristics that differentiate an ideal customer to someone else in the database.

At the end of the modelling process, each customer and prospect in the database is assigned a score. The higher the score the customer receives, the more similar they are to those in the target group and the more likely they are to respond positively to your marketing activities.

In order to create your model, you need to make sure that you have as much data available as possible. While scoring your customers, you’re effectively ranking them in accordance with their level of interest in your business, their buying behaviour and how likely it is that they’ll engage with your offers. 

However, the criteria you use for scoring customers will vary depending on the objectives of your business. For example, if you work in ecommerce, you’re likely to rank customers based on how likely they are to make a purchase. By contrast, if you work in credit or banking, you may rank customers based on how ‘risky’ they’re perceived to be for credit.

To build a strong scoring system, you should include all types of data, such as:

• Socio-demographic data: age, sex, marital status, and profession
• Psychological data: interests and opinions
• Behavioural data: purchase history, date of last purchase, purchase frequency, website behaviour, number of customer service complaints, email response rate, etc.

You can source all of this data from various areas. For example, behavioural data can usually be sourced from your CRM and socio-demographic information is often provided by the customer when they make a purchase. The more information you have about your customers, the more data you can generate and the more accurate your scoring system will be. If you’re short on data, then you can run customer surveys to enhance your data pool.

Once you’ve gathered all the relevant information about your customers and decided on the objectives of your business, you need to build a formula or a set of rules that can be used to calculate scores for your customers and identify the ideal target group.

There are a wide range of customer scoring techniques you can use for this and the right method will depend on your goals. If you use a piece of data analysis software like Apteco FastStats®, you’ll find scoring methods are included with the software. Alternatively, if you’re working manually, you can try techniques like RFM scoring (recency, frequency and monetary value). This will determine customer purchase probability based on three criteria: the date of their last purchase, their purchase frequency and the monetary value of the purchases they’ve made.

You need to make sure that the customer scoring techniques you use accurately evaluate the key characteristics of your customers. By providing your chosen model with this information, it can then predict likely prospects by seeing if they share these key characteristics.

If your model is applied correctly, it will give high scores to customers in the target group. To check the model has worked correctly and is analysing your customers accurately, you should run the model on some control data before you apply it to the data you’d like it to interpret.

Once your model has been built, tested and applied to your database, it will provide all customers in the database with a numerical score. At this point, your discretion will be required in determining what the cut-off point for a positive score is and how many customers you should target.

If you’re unsure how to create a customer scoring model or want to create a process that’s quick and replicable, then take a look at Apteco FastStats®. As well as processing millions of customer records in a matter of seconds, Apteco FastStats® also produces a profile report that highlights the characteristics that are statistically the most prevalent within your existing customer base.

By streamlining the process of creating and applying scoring models, you’ll find it easy to select customers for your upcoming marketing campaigns. Plus, because you won’t need to use any external applications to create your models, you’ll find it simple and quick to create accurate models that are backed by data.

As soon as your customer analysis has been created, Apteco FastStats® can score, rank and segment every record in your database. Plus, with Apteco FastStats®, you can take advantage of several modelling techniques such as:

Profiling

Using a patented predictive weight of evidence (PWE) method that combines information theory and Bayesian probability, this technique scores individual customers with a minimal amount of user input.

Decision tree models (including CHAID)

This method produces a set of ranked rules that identify distinct segments of your database that contain your best customers. If you have an external database, you’ll find that decision trees are particularly useful.

Clustering

Cluster analysis identifies groups of customers and prospects with similar characteristics. The clustering method uses the K-means technique to allocate each record to the nearest cluster centre, enabling you to better visualise and segment your database.

As an added bonus, you can test the model using the model report tool. By inputting business financials, such as costs and revenue, this tool enables you to identify the point at which your model produces the maximum return on investment (ROI) or keeps you within budget.

With the help of Apteco FastStats®, you can effectively analyse even the largest data sets to gain insight and improve campaign effectiveness. If you’re interested in using Apteco FastStats® to improve the performance of your next campaign or would like further information about any of the products in our range, then get in touch with us today to book a demo.