Recency, Frequency Value scoring

We are not all the same and neither are our customers! So it should come as no great surprise that customer segmentation is key to ensuring you market successfully to your customers; right message at the right time.

At the heart of direct marketing is data and using this valuable resource to ensure you send appropriate propositions to the right customers segment.

Many organisations think that they need to buy in external lifestyle or business demographics in order to better understand their customers. Whilst there may be benefits from these data resources, the value of in-house data should not be overlooked.

The most common areas around which to segment customers are:

  • Geography: Region, City or Postcode
  • Product: Category or even specific product
  • Response: (which campaigns have they responded to)
  • Source: (how their details were acquired)
  • Gender

These do not provide much insight into what makes a customer different when compared with any other customer in for instance a similar geography. But when combined with a measure of customer performance they become more revealing.

The classic method for measuring customer performance is to score each customer based on:

  • How recent was the last order placed by this customer – Recency 
  • How many times has a customer ordered on average within a predetermined period – Frequency 
  • How much they spend or what margin they contribute when they order –Value 

How to score customers

For each customer run enquiries on your database to identify:

  • Date of first and last order
  • Number of orders
  • Spend value (this can be gross margin, order value, whatever appropriate measure of value of customer expenditure)

From these calculate:

  • RECENCY: Time since last order is current date minus date of last order (Days) 
  • FREQUENCY: Duration of custom: current date minus date of first order (Days) divided by Number of orders 
  •  VALUE: In some cases you may wish to calculate a total value (i.e. sum the value measures) in others an average is adequate, calculated by dividing the total value measure by the number of orders 

These give you absolute values for each customer. The next stage is to categorise these into segments by determining relevant groups of values and then apply a corresponding score to each customer.

Taking each element in turn, the objective is to achieve categories that have broadly speaking similar numbers of customers in each.

The number of categories will vary depending on your business. For example a mail order clothing company:

RECENCY:

  • 1 to 30 days – score 5
  • 31 to 60 days – score 4
  • 61 to 90 days – score 3
  • 91 to 180 days – score 2
  • 181 to 365 days – score 1
  • Greater than 365 days – score 0

FREQUENCY:

  • At least once per month – score 5
  • At least once in three months – score 4
  • At least once in six months – score 3
  • At least once in greater than six months – score 1
  • Once time only purchase – score 0

VALUE: Clearly this depends on the business in question. If Gross margin is not readily identifiable, then the sales value could be used, but consideration should be given to the distribution of margin across the products your organisation sells.

RFV0

Assuming similar ranges of scores are applied for Value then your best customers would be those with scores of 555 (highest worth) down to 000 (lowest worth).

Remember this is not a discrete value of say five hundred and fifty five (highest worth), but an identifier composed of 5 elements: 5-5-5.

This therefore provides an identifier for each customer which can be used as selection criteria for campaigns e.g. You want to communicate with all reasonably recent high spenders who are not the most frequent customers with the objective of increasing the number of purchases they make in a period; so you might select customers with an RFV quotient of 414, 425, 414 and 525.

An on-going scoring process enables you to track purchasing and the effect of campaigns on customer behaviours. So in our example, how many has our campaign converted into 525s and 555s?
RFV1

This comparison will also provide direction for future campaigns to address churn, unprofitable customers and up-selling.

Other attributes can be score in a similar way: how long the customer has been a customer (longevity), number of complaints, cancellations or returns, referrals to friends and relations, abandoned baskets, quotations not taken up, etc., all adding to the behavioural profile of the customers.

Marketing systems can accommodate automated workflows for communications, reporting or alerts when customers significantly change score, to drive relationship marketing strategies.

For more guidance on RFV modelling or help with building your own models contact Michael Collins at mc@dmcounsel.co.uk.

©Michael Collins 2007-2018

 

 

Advertisements

Scoring your customers

Companies must continue communicating with their customers through a consistent strategy that is driven by customer insight. Rating current customers is not solely based on how much they spend but has as its basis the classic RFM (or RFV in the UK) model that creates a score based on recency, frequency and monetary value. My recommenation is not to stop there but to add more elements such as Returns, Complaints, number of enquiries, length of time they have been a customer and create a contact plan that reacts to the dynamics of these scores.

Often we will identify cohorts of customers that represent bad business. The solution can either be to review the business process for how you do business with them, e.g. relegating them to an ‘exclusive’ on-line relationship where the cost of managing them is reduced rather than taking up the time of a salesman or telephone agent; alternatively the bullet might have to be bitten and you resign the account.

However, building segmentations or communities of valuable, profitable customers by profile and comparing their behaviour will also drive the communication; but don’t just use it to determine when to make contact. Customer insight should also drive the ‘next best proposition’ for each customer so that the sales person can be proactive in establishing opportunity. What you know about your customers can also be used to drive new customer acquisition by comparing prospects’ profiles with your customers and determining the best proposition.

Customers rate companies with whom they deal by the quality of the communications and this means relevance, personalisation and timeliness.

The extended 3-letter acronym to drive retail direct sales

Analytics software is useful, but some concept of what should be analysed is also of value. Retailers will often create a RFV model, combining Recency (of last purchase), Frequency (regularity of purchase) and Value (either total spend or average order value, for example) to create a score for each customer.

We have found that adding additional elements expands the score from the 3-letter acronym to a multi-letter acronym such as RFVRCA where you add on the instances of Returns, Complaints and Abandoned on-line baskets committed by the customer.

This score then becomes a comparitor between customers, a selection criterion for campaign or proposition and changes in the individual’s score over time become a valid measure of success of the relationship.

Applying demographic, lifestyle and psychographic profiling provides an even more granular segmentation and the ability to apply the concept to prospects, with the benefit of making relevant offers to convert them to customers.

Such models can be created without investment in analytical software and can be applied in rules-based workflow and business process, with the dynamics regularly reviewed and the model(s) enhanced.

Data Fusion delivers the “Why” Chromosome In Customer Management

When considering the uses of a database for marketing, the issues go far beyond the realms of purely direct marketing. There is a role for the database in supporting most elements of the marketing mix. Advertising can benefit from the profiling of the database to determine the tone of voice of copy, the image of the characters in TV commercials or press ad photography and the media selected to carry the ad. Similarly, market research can be segmented and directed to profiled individuals so that more robust samples can be constructed. The database can assist in product development, the management and motivation of a sales force or channel of distribution and in sales promotion.

However, as we have seen in this column in previous issues of Review the integration of data into data warehouses or data marts for marketing analysis means that the elements of the marketing mix can contribute as much to the database as the database benefits those elements. Knowing who has bought what product, when and in response to which campaign, through what communication channel and with which payment method is very useful. We can understand the “Who”, “What”, “When” and “How” that were behind the customer action. What if we were able to add the “Why” factor – the psychographical element that establishes true differential between customers. Why does a customer remain loyal? Why does the customer value the relationship with the company? Why does the customer maintain a dialogue with a preferred supplier?

The same techniques that have brought about the expansion of data warehousing have facilitated the use of data fusion, the process by which marketing research data can be introduced into the marketing database and matched to customers and prospects so that the complete view can be established.

Whether the marketing research is bespoke to the marketer or is syndicated or publicly available, the ability to match it in to the customers either by direct personal identifier or by segment or cluster, behavioural type or demographics can permit the extrapolation of the research factors right across the marketing database.

This means that response by a representative panel of customers or prospects to questions regarding key purchase considerations, performance indicators or customer expectation levels can be levied right across the database to help derive more intense customer profiles and establish the key differentiators between clusters previously considered similar.

Knowing a customer’s expectations in terms of quality of product and service, preferred channels and extent of communication and preferred promotional activity can enable marketers to tune services and propositions to meet perceived customer needs. Knowing how the company matches up to these expectations provides an important new dimension to customer relationship management which, if acted upon, will result in customers that feel valued and are likely spend more and be increasingly loyal.

Whilst scoring techniques such as compiling recency, frequency and value quotients give an important basis on which to determine segmentation, it does not necessarily demonstrate loyalty. Customers will find their own level of relationship with the company and cannot be forced to assume a closer one.

There is nothing to say that a customer at the left end of the spectrum is any more loyal than the customer at the right end. The fact is they may be both equally loyal but view the relationship differently, both equally satisfied. The customer who expects everything, however, may end up being less profitable than the other extreme, since the level of communication may will be greater for perhaps an equal (or even lesser) amount of sales activity. Hence the need to understand the customer attitudes and aspirations to establish the performance indicators that mean the company can gauge if they are getting it right.

The actual data fusion process requires an understanding of the data that will contribute to the analysis universe. This will normally start with the company’s house database of customers and prospects. The data will need appraisal and a high level audit to ensure that it is fit-for-purpose and to highlight any issues concerning data quality or integrity that may affect the creation of the survey panels and eventually the outcome of the analysis.

An initial round of data analysis to establish customer segmentation and profiles for the research would then need to be undertaken. Often this will entail constructing purchase value or product profiles and the introduction of external qualification data such as demographics or lifestyle information, to help establish the segments for research. Then it is just a matter of running extracts of the data that meet the profile requirements for the research panels; remember, the research will normally be far more meaningful if applied to segmented panels rather than solely right across the database. These extracts can then be used to create representative samples, output as:

–         Call lists for telephone research

–         Mailing lists for postal research

–         E-mail lists for electronic questionnaires

In each case the segment reference is maintained against the customer record and the response process designed to make the loading of the responses as easy as possible.

Having carried out the research, the raw response data can then be matched back to the database, with response outcome extrapolated throughout. Once complete, the database can then be analysed using the analysis tools I have discussed in this column previously, to follow train-of –thought exploration and data mining discovery techniques to identify the latent trends, opportunities and threats.

 This technique can address some of the key business imperatives uppermost in the minds of marketers:

 –        The need to know more about their customers

–        How to identify and protect against churn

–        Direction for business growth

–        Areas where customer service needs improvement

–        Pinpointing the opportunities for cross sell and up sell

 The company can be in a stronger position to increase sales to current customers, improving customer retention and loyalty and to find new customers that match the best customers already on file. Similarly, by profiling prospective customers, the research findings can be applied from the very inception of the relationship, thereby increasing the effectiveness of the prospect conversion process. In both cases, the company can see increase in turnover, a reduction in prospecting costs and maximisation of the efficiency of the marketing budget, meaning more to bottom line profit.

The combination of research and database analysis will also provide the basis for development of new products and identification of new markets and can be a key factor behind maintaining and increase market share. It will provide a competitive edge that not only helps prevent market share erosion but also assist in gaining rapid share of new markets.

 Last, but by no means least, the ability to carry out the data fusion process and regularly refresh it provides a powerful measurement tool to assess the effectiveness of marketing initiatives.