12 - Maximizing Customer Information: Personalization
The author refers to mercantilism, before the emergence of mass-markets, where most people bought goods from shop-owners who knew their customers by name, knew their preferences, knew their sizes, knew an awful lot about their regular customers and bent over backward to provide them with a high level of service. The personalization of the Web is, in that way, more of a return to older practices than blazing new trails.
Unfortunately, the first wave of "personalization" services on the Web were multi-million-dollar database systems that touted one-to-one marketing, but ultimately did only a few parlor tricks that didn't result in a significant increase in revenues.
Segmentation
There are various ways to segment your customers on the Web:
- Technology - You should be able to tell whether a customer is using a PC or a Mac, which browser they are using.
- Interest - You should be able to tell, from the pages they visit and the products they buy, what categories of merchandise are of interest to a specific user.
- Demographics - You can determine the customer's geographic location fairly easily. Learning things like age and household income may be more difficult.
- Recency and Frequency - Knowing how long it has been since the customer last visited, and how often they visit and purchase
- By industry and job function - This example comes from a job-seeking site, where the information is disclosed.
- By attitudes and lifestyles - The author cites VALS, and I've seen ESRI data, that groups people into certain character types that have distinct behaviors and attitudes.
- By surfing mode - You should be able to examine past behavior to determine whether a given person will browse the store or hunts for the item they want, whether they are prone to compare alternatives before making a purchase, whether they are a person who tends to complain a lot after the sale, etc.
The relevance of this data may be:
- Clearly relevant (if you're selling software, knowing a person uses a Macintosh computer tells you not to show them PC software)
- Obliquely relevant (A person who owns a Macintosh is willing to pay a premium for a more upscale product)
- Statistically relevant (While there is no logical connection, your statistics show that 85% of people who buy a given product are Mac users)
Models of Behavior
The author looks at a couple of categorization schemes for customer behavior online.
Nielsen/Net Ratings defines seven kinds of visits - a single shopper may have several visits of different kinds over time.
- Quickie - A visitor whose total visit is less than a minute, in which they view a small number of pages (3-5) to gather information, then go away
- Fact-Seeking - An average visit of nine minutes, but they still have rapid page views (around 15 seconds per page) of a larger number of pages. They may buy, but more often are just gathering information.
- Single mission - Average ten minutes per visit, but view fewer pages than fact seekers. They are gathering detailed information on a small number of options.
- Repeat Research - This individual has made a previous visit, generally of one of the kinds above, and is spending more time gathering information on a smaller number of products.
- Loiterers - These individuals spend about half an hour on a site, browsing various pages (product and other information), but seldom make a purchase
- Deep Research - These individuals spend around 40 minutes on a site, gathering in-depth information. They are generally on-mission rather than browsing leisurely, and have a higher probability of making a purchase than any other group.
- Surfers - These individuals spend around an hour on a site and have long page-views, but their intermittent activity suggests that they may be comparison-shopping with other sites.
Swinyard and Smith, professors at BYU, categorized Web site visitors by their shopping behavior.
- Shoppers (11.1%) are individuals who enjoy buying online and do so frequently. They have a tendency to frequent certain retailers and will tend to make repeated purchases at a favored site.
- Explorers (8.9%) are also frequent buyers, but they tend to comparison shop rather than staying with one supplier. They also tend to be opinion leaders online, and will advocate retailers they like (and rail against those whom they don't)
- Learners (9.6%) are individuals who are cautious about buying online due to their inexperience. They are not fearful of the medium (are not reluctant to give a credit card online), but need an easy-to-use experience that provides guidance and coaxing through the shopping process.
- Business Users (12.4%) are highly computer-literate, but are shopping for business purchases. They have already decided what they want, and will not be distracted or cajoled.
- Fearful Browsers (10.7%) are people on the cusp of buying online. They do a lot of window shopping, but primarily use the internet for gathering product information and will buy offline. IT is suggested that addressing their fears about credit card security, shipping charges, ease of return, and other "dangers" may help them overcome their shyness.
- Shopping Avoiders (15.6%) are a poor target for online retailers. Like fearful browsers, they use the internet to gather information but do not buy online. However, their reasons are generally things a retailer cannot overcome (do not want to wait for shipment, want to see merchandise in person)
- Technology Muddles (19.6%) have computer literacy hurdles. They don't spend a lot of time online, and generally need a lot of handholding and support. Given that they generally have a low income level As well, they are not an attractive target market for online retail.
- Fun Seekers (12.1%) tend to be young users (teens and students) who see the Internet as entertainment, and are unlikely to buy online simply because they don't have the means. Consider them to be like the kids who go to the mall to have fun, but not to buy anything. They're a nuisance, not a market.
Customization
A "static" site presents the same experience to all customers.
A "customized" site presents the user with the ability to tell you their interests, then customizes the site accordingly, but generally only for that visit. For example, an online shoe store might ask your size, and show only items for which your size is in stock. The next time you visit, you'll have to re-customize the site.
A "personalized" site collects this information explicitly as well as by observation, and marries its knowledge to a customer profile, which is used each time you visit the site.
The main difference between customization and personalization is a matter of degree: a personalized site knows a lot more about you, and uses that information to a greater extent.
Most significantly, it they differ in the time horizon: a customized site seeks to make your shopping experience easier for a single visit. A personalized site seeks to retain your business and build customer loyalty over a longer period of time.
Retention/Attrition
It costs ____ times as much to acquire a new customer than to get a repeat sale from an existing one. Various sources have said this, and the blank has been filled in with numbers that range from "two" to "twenty" and even "many" - but the problem most Web retailers have is that they do not consider the long-term value of a customer. They want to make an immediate sale, and don't consider the next one.
Most customers buy the same goods multiple times. For example, a person who buys a car today will buy another car in the future. The cycle differs among customers: some people buy one every two years, some every four, some every ten.
And so, your long-term success isn't measured by how many units you sell this month, but how many of those customers will return to you the next time they need to purchase the same item. In general, this is referred to as "customer loyalty", though a number of factors can be measured:
- Recency - When was the last time a person visited your site? When was the last time they purchased? Has it been so long you think you've lost them to another supplier?
- Frequency - How often is a person buying from you? Are their purchases "regular"? Are you their only supplier, or are they also purchasing from others?
- Quantity - Does this customer buy small amounts frequently, or large amounts less frequently?
- Value - How much is this customer "worth" to you, over time?
- Advocacy - How likely is this customer to act in ways that bring other customers to you? Is their opinion influential in third-party forums? Do they "help" other customers in your own forums?
- Handholding - Is this a customer who incurs additional costs to you, from customer-service calls, returns, and other behavior aside of purchasing? Is this costing you more in the long run?
Lifetime Value
Much ado is made in certain circles about the lifetime value of a customer, but it's a bit wonky.
Primarily, it considers that a customer's current purchasing patterns are indicative of a lifelong trend. They won't change their buying habits in unpredictable ways. They won't leave you for another retailer. Not many companies do it well. The author speaks a bit about Harrah's (the casino chain), but can't disclose much in the way of specific detail.
In general, it takes some creative accounting:
- Look at the number of new customers and lost customers over a given period of time, and use that to calculate the average "tenure" of a customer (how many years, quarters, etc. the average person stays with you)
- Look at the amount of revenue generated over the same time, and divide it by the number of customers
- Calculate your operational costs during the same time period (include bad debts and other factors)
From there, it's pretty straightforward: the net present value of the total profit made on a customer over the duration of their tenure yields a dollar amount that is a basic "lifetime value" per customer.
You can also segment your customer based to get a better idea. Certainly, some customer segments generate a higher LTV than others. You can even apply it to the individual level - the only thing you must abstract is the number of years you expect to keep them.
Viral Value
The "viral value" of a customer is the value of business they refer to you.
However, this is harder to calculate, because it is difficult to observe the activities that a person undertakes to refer others to your site. It may be measureable for a campaign (refer a friend), but that's only a small fraction of the impact the person may have on you.
Also, the author fails to consider negative viral value: how much does a person who makes negative remarks end up costing you?
This is all too cloudy to be considered, but it's food for thought.
CRM
The author strays into customer relationship management - but he doesn't do it well. A lot of buzzwords and fluff, but nothing new.
The bottom line is that if you know a customer, you can serve him well, and if you serve him well, he'll come back to you when he needs to buy again, and he'll tell his friends.