Practical e-CRM Part 2 - Customer profiling
Topics
Profiling as part of permission marketing. 1
Using profiling information for targeting. 3
Introduction
In the previous article on e-CRM we looked at how web site services can support customers through the buying process. Here we continue to look at e-CRM. The topic is learning more about customers through profiling them online in order to deliver more targeted offers.
Using profiling to learn more about customers characteristics, needs and behaviour is central to e-CRM. Through collecting and analysing such information, marketing campaigns and products can be customised to appeal better to groups of customers, or the individual. In this article the aspects of profiling we focus on are profiling visitors to web and then making use of the information collected in e-mail marketing campaigns. Of course, collection of e-mail addresses and profiling of customers is equally productive in the real world and a plan should also exist to maximise offline collection of e-mail addresses.
It will be no surprise to direct marketers that through tailoring e-mail campaigns to individual preferences, the effectiveness of e-marketing can be increased significantly. DMNews (2002) reported on an analysis of more than 90 million messages delivered in the first quarter of 2002 by the Data and Analytics Group of US e-mail agency Yesmail. It was found customer response rates almost double when three to six data elements are leveraged to personalize e-mail communications and triple when seven or more data elements are used. On average, a message personalised using personalised characteristics such as name, interests, gender, age, purchase history and message frequency preference delivered a response rate of 14.8% compared to 4.7% for a one size fits all bulk e-mail campaign.
Looking at efforts by organisations to collect customer information online suggests that many companies have recognised the need for profiling and think, we need a form to collect customer information. The planning ends there. The detailed design of the online form must be thought through. To be effective, the fields used to prompt the customers for their details must be carefully selected according to how they will be used for targeting. Collection of customer data is not a one-off activity, so the initial data capture activity needs to be looked at in relation to all permission marketing activities and the legal constraints of data protection law.
We start the article by putting profiling for e-CRM in the context of permission marketing and the latest legal constraints. We then move onto examine how the fields on the form should be selected to fit in with data collection, and finally how we can make use of the data collected.
Profiling as part of permission marketing
Permission marketing is a term popularised by Seth Godin, formerly a VP of marketing at Yahoo! in his classic 1999 book Permission Marketing (Godin, 1999). He argues that there is a need for permission marketing since there are an ever-increasing number of marketing communications bombarding consumers, which he calls interruption marketing. He notes that while research shows that we used to receive 500 marketing messages a day, with the advent of the web and digital TV this has now increased to over 3,000 a day! From the marketing organisations viewpoint, this leads to a dilution in the effectiveness of the messages how can the communications of any one company stand out? From the customers viewpoint, time is seemingly in ever-shorter supply, customers are losing patience and expect reward for their attention, time and information.
Permission e-mail marketing is synonymous with opt-in. Opt-in means that a prospect or customer proactively agrees to receive future marketing communications, including e-mail. By checking a box on a web form they indicate that they agree to receive further communications from a company. For many companies this is preferable to opt-out, the situation where a customer has to proactively ask not to receive further information, by unchecking a box.
To see what opt-in and opt-out mean in practice visit these sites:
http://www.ibm.com, Choose My Account. This is classic opt-in. To receive e-mails and marketing communications in future the site visitor has to proactively select this option.
http://www.siebel.com, Choose Sign Up. This is opt-out the visitor has to de-select the option if they do not want to receive information in future.
http://www.dell.co.uk/winanotebook. Opt-in is implicit here. If you dont opt-in to receive the newsletter, then you cannot enter the prize draw. Opt-out is used for the form of communication required.
http://www.microstrategy.com. Choose, subscribe. This is also implicit opt-in the incentive is not available unless you provide your e-mail address there is no opt-in/opt-out checkbox. Further details are required after submit is pressed.
Deciding on whether to use a traditional opt-in, opt-out or implicit opt-in is not an easy call. It is also an important call because it will affect the size of your list, response rates to future e-mail campaigns and how your brand is perceived. Opt-out is often used since it will result in the largest list. However, some e-mail recipients will receive e-mails which they see as unsolicited since they completed the form without proactively agreeing to receive information. This will result in lower response rates, although the absolute volume of responses may be higher. Opt-in is the pure permission marketing approach and will result in a more receptive audience although the list will be smaller. Given this, the implicit opt-in approach is used increasingly since this is a compromise which results in more subscribers than pure opt-in, but the recipients are more likely to be disposed towards communications.
It seems that permission marketing is sometimes seen as ending with opt-in. However, there are further elements of permission marketing to consider. Incentives can be used for continued dialogue - customer needs are researched as further data is collected throughout the customer lifecycle. The aim is to increase the level of permission by collecting more detailed information and responding accordingly. Customers can also decline receipt of information, effectively ending the relationship. This is opt-out or, in an e-mail context, unsubscribe.
Finally, we should note that this approach is not entirely new. Ross (2001) says the large number of direct marketers who start the selling process with an ad offering information and take it from there with a graduated programme of data collection a
nd follow-ups will wonder how it is that have been practicing permission marketing for all this time without knowing. Ross also notes that in order to achieve permission, it is necessary to gain the attention of prospects, and this is of course, only possible using.interruption marketing.
Forms of opt-in
EMMA, the UKs e-mail marketing association (www.emmacharter.org) uses these definitions related to opt-in in its charter:
Opt-in means an approach in which a user who desires to be added to a list must request, actively, such as by checking a box on a web page, to be added to the list. Any personal data collected without a positive action from the individual will be deemed to have been collected on an opt out basis. Without a positive action from the individual, they will be deemed to have opted out of having their personal data being used for direct marketing purposes
Double opt-in means an approach in which after a subscription has been received and entered, a confirmation E-mail reply containing (i) the source of the subscription request and (ii) instructions to confirm subscription to the list.
Notified opt-in means an approach in which after a subscription has been received and entered, a confirmation E-mail reply containing (i) the source of the subscription request, (ii) instructions for how to unsubscribe from the list, and (iii) instructions for how to report that the subscription request was in error is sent to the user.
Opt-out means that a user who is about to be added to a list, at the point of personal data collection can actively indicate that permission is not granted.
Notified opt-in is the most common approach an e-mail autoresponse is received after subscription. It is not necessary to confirm subscription, but there is an option to opt-out, if the subscription has been in error. Double opt-in is arguably truest to permission marketing, but it has a nuisance factor if you have opted-in once, why do you need to be immediately asked again.
Legal constraints
Nowhere is the maxim consult a lawyer more relevant than for e-CRM and e-mail marketing since new legislation and case law means that it is practically impossible for a marketer to keep up to date with the details of legislation. However, to stay abreast of the latest developments there are some excellent newsletters to assist:
Marketing Law (http://www.marketinglaw.co.uk) Monthly e-mail update
Berwin Leigthon Paisner (http://www.berwinleightonpaisner.com) More frequent Techmedia eBrief e-newsletter.
The law on data protection has been updated recently as a result of the EC Communications Directive which has been in force in the UK since August 21st 2002.
The main tenets of the new legislation are that:
1. Commercial e-mail must be identified as commercial, with senders name. This implies the need for a direct subject line, clearly explaining the offer.
2. Unsolicited commercial e-mail must also be identified as such. If you purchase rented lists, you should use a statement of origination in the e-mail to explain to the recipient how their details were sourced.
3. Full company details needed in the e-mail as for other business communications, with a named contact point.
4. Country of origin principle applies. This means that as long as a company adheres to the law in their own country, they can use similar approaches in other EC countries.
Details of the legislation are at: http://www.legislation.hmso.gov.uk/si/si2002/20022013.htm with a more accessible interpretation at: http://www.dti.gov.uk/cii/ecommerce/europeanpolicy/ecommerce_directive.shtml
When the legislation was under discussion, it was expected that the new laws might require all e-mails to be opt-in. However, the legislation does not cover this level of detail. However, further modifications to the law are planned for next year.
Profiling strategy
Selecting which fields to use to profile site visitors is a tricky balancing act. We want sufficient information to understand the characteristics of our online visitors and target them effectively in future communications, but we do not want to put them off completing the form or giving inaccurate information.
According to Peppers and Rogers (see www.1to1.com), we want to be able to Identify customers in order to Differentiate, Interact and Customise.
In the early days of the web, many companies asked far too many questions and this obviously didnt work. Marketers then swung to the other extreme, and collected e-mail addresses only. Today we are seeing more sites, where companies have thought carefully about the information they really need and have restricted their questions to essential areas. So as well as Name, e-mail address and organisation, other essential fields include job title or role for B2B, Age, Sex and Postcode (to link to lifestyle databases) for B2C. Remember that you do not have to ask for all of the profile information up-front. Web analytics supplier WebTrends (www.webtrends.com) limits questions on initial subscription, but then runs campaigns that involve collect further information to better understand the customer.
Here are some examples of companies that have minimized the number of profiling fields:
http://www.microstrategy.com. Choose Subscribe. This company is interested in the nature of the relationship with the site visitor are they new to the company, or are they an existing customer or partner.
http://www.3mhealthcare.co.uk. Choose Sign up. A Job title field is used to identify the business role of the person. Their areas of interest are also asked for tailoring content in future.
http://www.nescafe.co.uk. Choose Events and Promotions. Postcode and age are the key profiling fields.
http://www.marketing-insights.co.uk/spreadsheet.htm. This form has just four fields, with position and company name sufficient to determine whether a follow-up phone call is worthwhile. Note also the use of the phrase Your e-mail address will not be supplied to other companies to reassure companies and increase response rate.
Further guidelines on collecting data to maximize completion and accuracy, taken from Chaffey (2002) are:
1. Explain the reason why information is being collected. A customer will more readily give up personal information and spend time filling in a form if they know the reason why it is being collected. Explain how it will benefit them by the proposition of the communications such as a newsletter.
2. Explain use to which data will be put clearly. This is required by law in a privacy statement, but if the data will not be shared with third-parties this should be highlighted since it may encourage opt-in.
3. Make it easy. Use the features of HTML to ease data entry. For example use drop-down lists where appropriate. These improve data quality by giving consistent data in the database. However, don't you hate those lists of 137 countries with Afghanistan at the top. Surely prospects can enter their own country?
4. Indicate mandatory fields. Extra information can be collected from customers if they have time by marking essential fields in a suitable way (perhaps through an asterisk or highlighting in bold).
5. Validate. Checks should be performed after the form is filled in to check that the user has filled in all mandatory fields. Fields should also be checked for validity has the customer entered a valid e-mail address with the @ symbol, is the postcode or zip code valid? The user should be clearly prompted with what information is wrong and why. Such validation can be performed using script languages such as Javascript.
6. Provide opt-in. Check-boxes should be made available that the user can select if they do not want to receive further information through e-mail or communications through other media.
7. Provide prompt confirmation. After a user has filled in a form, a company should respond to acknowledge confirmation of receipt as soon as possible and describe what the follow-up actions will be. For example, if a customer has ordered a product, the confirmation note should thank the customer for shopping with the company and state clearly when they can expect to receive the product by courier.
8. Analyse attrition. Use web site visitor statistics package to determine the attrition rate for the page. If one hundred people load the form, how many actually complete the form? Experiment with the design and copy on the form to improve completion.
Using profiling information for targeting
As mentioned at the start of the example, higher responses can be achieved, the more the house list is broken down according to the fields we have collected. Different creative, offers and timing can be used for each of the different selects from the customer database.
Lets run through ten of the common options for targeting, highlighting their relevance to e-mail marketing:
1. Demographics (age, sex, geography). These are generally the preserve of B2C markets. With the relatively low cost of creating and dispatching e-mail creative a different style and tone of creative can be developed according to age and sex. A Whisky brand developed different creative style for younger and older drinkers and also developed creative for female partners to encourage gift purchases. When opening new branches Boots Dentalcare used postcodes to target potential customers within a 20 minute drive-time of the branches. This resulted in a greater response than either radio or print advertising.
2. Lifestyle or psychographic. The providers of lifestyle classifications such as MOSAIC or ACORN provide services to integrate this data with e-mail lists in order to target according to lifestyle.
3. B2B (Company or individuals). For business-to-business organisations, the primary segmentation is usually according to company characteristics such as sector or size, proxied by number of employees or turnover. Dell Computer uses a classification based on employee size as its primary segmentation. Different web site and e-mail content is targeted at each of these segments. The individuals that make up the decision making unit can also be targeted separately.
4. Product (categories of purchase). This refers not to particular products, but general categories. For example an IT supplier might distinguish between customers who focused most on hardware or software categories.
5. Purchase history. This evaluates the characteristics of purchase through time using evaluation schemes such as Recency, Frequency and Monetary value analysis. E-mail offers great potential for targeting according to history of purchase and category of interest. E-mail services supplier E-mail Vision have developed a rules-based personalisation approach for retailers which allows them to target by product and other characteristics in the following manner:
By spend (over last n months):
Customers who have spent 300+ get message A
Customers who have spent up to 300 get message B
By product:
If profile says they buy books on-line get offer C
If profile says they buy CDs on-line get offer D
By Interest:
If they like Golf give them message E
If they like Tennis give them message F
If they like neither give them message G
The rules are devised such that the e-mail is created by combining these messages to form a unified e-mail.
6. Customer value. The simplest form of customer value calculation looks at the value to customer over a fixed period. For example we could assess total spend across the last 3, 6 or 12 months and customers could be placed in categories of spend such as the example above and given different offers. The future value of customers can also be calculated using lifetime value calculations. These include assumptions about the rate of churn, average spend per category and referrals to other members. Some companies such as Deutsche Bank use complex modelling of profitability for targeting purposes (see Nitsche (2002) for a summary. Their approach is iterative enabling the targeting capability to be refined through time.
7. Customer loyalty. A typical definition of loyalty is suggested by Sargeant and West (2001) who describe it as the desire of the customer to continue to do business with a given supplier over time. Customer loyalty is linked in with purchase history and customer value. Customer loyalty can be measured simply by length of time a customer has been with the company, but more meaningfully by the frequency and value of purchase using RFM or Lifetime Value Analysis.
8. E-mail preference. This can simply refer to the format required for e-mails whether text or graphical (HTML). Many companies are now also establishing a communication channel preference whether phone, e-mail or post. E-mail preferences may also state the type of content or offers a list member has expressed an interest in receiving. http://www.dell.co.uk/winanotebook.
9. Time on list (in months). This is arguably a more critical tool for targeting in e-mail marketing in comparison to traditional direct marketing. With opt-in for e-mail communications we would expect that the susceptibility to response is highest immediately following opt-in. Response rates tend to be highest when the most recent list members are targeted. Note that it is quite easy to forget to collect this information on initial profiling since the field is not explicitly used on the form.
10. Responsiveness to e-mail campaigns. This measure combines a number of targeting methods we have talked about such as purchase history, loyalty and time-on-list as part of the customer lifecycle.
Targeting can be refined during campaigns or in follow-up campaigns according to the recipients response to the initial e-mail. This is a response behaviour-based segmentation. We can again leverage the technology to automate this response process. The categories of response to each e-mail that we can identify are:
A. Dont open.
If the recipient does not open the message this suggests the list member is not responding to the Subject Line. This can be revised to have greater impact or refer to a different offer. The recipient may not be responding because of a lack of time, so another option is to send it out on a different day of the week, or at a different time. It is only possible to tell whether an e-mail is opened for an HTML e-mail.
B. Open, dont click.
If the recipient does open the e-mail, but doesnt click on it, this shows they are not responding to the creative or the offer. They have indicated a good level of susceptibility, however, so there could be a follow-up with different creative and / or offer.
C.Click, but dont act.
In this case the recipient has clicked through on one of the hyperlinks in the e-mail to the microsite or landing page. They have a very high propensity to act, but something has stopped them acting. A specific e-mail perhaps acknowledging their interest, but offering an improved incentive as an additional carrot to persuade them to act might be effective here.
D. Do act.
If the recipient does click through and follows the instructions on the landing page, then the individual should be flagged in the database as having a high susceptibility. A follow-up with a complementary offer should be planned.
References
Chaffey, D. (2002) Total E-mail Marketing. Butterworth Heinemann. Oxford, UK.
DMNews (2002) E-mail Marketing Weekly [E-MailMarketingWeekly@dmnews.com] 17th July 2002 www.dmnews.com.
Godin, S. (1999) Permission Marketing. New York: Simon and Schuster.
Nitsche, M. (2002) Developing a truly customer-centric CRM system: Part Two Analysis and Campaign management. Interactive Marketing. Vol 3 (4) April/June 2002, 350-366.
Ross, V. (2001) Review of Permission marketing. Interactive Marketing. Jan-March 2001, 292-4.
Note: This article is one of a series originally written between 2000 and 2003 for The Chartered Institute of Marketing What's New in Marketing Newsletter. Many of the concepts remain valid, but some more recent concepts such as Web 2.0 and social networks aren't referenced. More recent editions of my books contain more up-to-date examples and latest thinking.
For the latest on these concepts see my DaveChaffey.com blog site for E-CRM articles.
