Personalisation goes one-on-one with reality
By David Walker (Google profile)
The technology called "personalisation" had its genesis in a 1993 book by marketing experts Don Peppers and Martha Rogers called The One to One Future, a tome about the possibility of selling in a different way to each different customer. Around 1995, technology-oriented marketing types put Peppers and Rogers together with the Internet in one of those mad-scientist lab-explosion moments. When the smoke cleared, they had decided that online commerce would be one-to-one marketing's killer app.
The revelation went like this. I like hearing "hello David" and being shown pants in my favorite style when when I walk into a store, and direct marketers have for many years observed that most people named David responded better to letters addressed to "Dear David". So I'd like that on a Web page, wouldn't I - along with lots of other stuff based on the same principle, enough to build a "close relationship". And after customers customers entered these close relationships, they would face a higher cost in leaving to establish a new relationship at some other site.
In short, the Internet strategists concluded, online stores needed unique, one-to-one and automated relationships with all their customers.
[A note on terminology: this article adopts the loosely-accepted distinction between personalisation, where technology automatically analyses your preferences and adapts to them, and customisation, where you simply state what you want - which stock prices, which sports scores and so on.]
Business managers fell easily for the Web personalisation story. After all, it suggested that Internet retailing might lock in consumers and churn out margins as fat as bricks-and-mortar retailing. Web personalisation systems were delivered into a receptive market.
The strategists had to report problems right from the start, but they assumed new technology would fix everything Some Time Real Soon. Typical was May 1997's dramatically-titled Forrester Research report, Personalize or Perish?. The report admitted breezily that "products that promise personalized Web experiences at the click of a button currently fall short of the mark". That would soon change, it suggested.
The change never came. Two years later, in a July 1999 report titled Smart Personalization, Forrester concluded that "typical Web personalization efforts fail to produce results that match market expectations". An August 1999 report by Forrester rival Jupiter Communications struck a similar tone, reporting that personalisation was "evolving into a core element of site infrastructure" but also describing business managers as "overwhelmed by the sheer quantity of data that a typical Web site generates". Added Jupiter: "Sites are struggling to understand their users, isolate the characteristics of valuable customers, and gain a full appreciation of the returns ...". In other words, businesses have bet on the personalisation hype and bought the technology, but can't make good use of the data it produces.
One of the deepest analyses of personalisation's potential came in 1999 from Mainspring, a strategy firm. Mainspring went out and talked to nine financial services firms trying to implement personalisation.
Mainspring's chosen firms all hoped personalisation would bring users back to their site over and over, keep them there longer and - especialy - encourage them to transact more. Like most firms running substantial Web sites, they worried that they had created expensive new Web sites and Web processes which weren't attracting enough real customers. They were seeking new ways to turn their visitors into buyers.
But Mainspring found all these firms struggling with personalisation's planning, staffing and technology issues, and unable to show that their effort would ever pay off. None of the firms had created measurable goals for their personalisation project. None could provide hard data showing that personalisation had boosted their profits or even their sales. They'd spent sums running over $1 million, but they didn't know whether it had done any damn good.
Organisations often struggle to measure the results of Web site intiatives, just because the typical site environment changes so fast. Big, stage-by-stage intiatives like personalisation present a tougher measurement challenge still. But at the very least, Mainspring's study suggests personalisation won't turn a so-so site into a success. Rather, personalisation may give sites an incremental boost. That's a rather different picture from the one painted by its most enthusiastic backers.
"Financial institutions will have to weigh the expected benefits of personalization against the cost of redesigning their site to successfully integrate personalization in order to make a sound investment decision ... There is little doubt that the Internet of the future will be a more personal place, but firms need to decide where and why to use personalization before deciding how to do it."
More recently, Internet research firm Jupiter Communications has dug down further into personalisation's problems. Jupiter has generally treated personalisation as an exciting new development, but its July 2000 report sounds all sorts of warning bells. It notes many firms are having trouble integrating personalisation with other site features and analysis tools, and that many personalisation systems scale poorly as sites grow. The report describes existing personalisation technology as "immature" and predicts that personalisation software tools will "creep forward" rather than taking off with a rush.
"Many Web ventures that Jupiter contacted are unhappy with second or third attempts at creating a personalization delivery chain and are actively pursuing a third or fourth attempt at enormous costs ... This rejection rate will remain high until tool vendors can deliver on promises and the landscape settles 18 to 24 months into the future. ... So tough is the personalisation challenge that Jupiter expects that in time, many firms will outsource the task to data application service providers."
Typical of the failures was Levi Strauss's Stylefinder feature. It had personalisation experts pointing excitedly to it - right up to the moment when Levi's dumped it from their site.
Even the keenest of personalisation enthusiasts now expect managers to start treating the technique more sceptically. Business Week magazine has run several gushing pieces about personalisation's potential. But in an August 2000 article it quoted McKinsey & Co's Marc Singer (author of the Internet commerce-boosting volume Net Worth) warning that personalisation's dream run is about to end. "There'll be some form of backlash in the next six months," said Singer, "because so much money is being poured down the tubes".
And so it went. By 2001, StreetAdvisor.com founder Kevin Prigel could tell eCompany Now magazine that personalisation had been his worst technology mistake. "We built a personalization-engine service, and it turns out people didn't have much interest in personalizing content. They are happy going on and reading what's on our front page and navigating their way through it. About 1 percent of our users used it. We spent $200,000 and ultimately canceled it."
Why the difficulties turning theory into practice?
Because the Big Concept of Internet personalisation is currently flawed on four levels:
- Customers don't want relationships with corporations. As I smile at the sound of the store attendant's greeting, I'm actually responding to a person. I'm certainly not responding to a corporate message, and I certainly don't want warm personal "relationships" with the 100 retail businesses I use most. The image of the exploitative corporation is embedded in modern popular culture, underpinning Hollywood movies all the way from "Erin Brockovich" to the "Alien" series. A colleague of mine at the Internet loan-finding service eChoice, a former senior ANZ executive, is struck by his children's lack of attachment to financial institutions. "They don't have any loyalty," he remarks. He isn't condemning his kids; he's simply noting a change in consumer behviour over the past two decades.
- Personalisation requires data. Amazon.com, the poster child of personalisation, makes the technique successful because book purchases send it a constant stream of finely-detailed data about individual customers. Most transactions, though, don't provide such rich preference data.
- Personalisation technology mostly understands consumers poorly. Even Amazon.com will start recommending needlepoint books to you as soon as you order that ideal gift for your great aunt. At this point, even the thickest consumer is apt to realise he and his online bookstore aren't that close. Technological progress will shrink this problem, but not eliminate it.
- Personalisation costs. US group Mainspring estimated in a June 1999 report that "any significant personalization initiative will cost hundreds of thousands of dollars" - a conclusion with which every major player in the field agrees. Implementing serious personalisation systems will burn not just cash but the time of valuable staff. As Jared Spool of User Interface Engineering noted in a recent newsletter: "You're more likely to get a good return on your efforts - and encourage repeat visits - by fixing other problems, such as difficulty in locating content". System costs will fall with time, though.
Personalisation does work effectively. It simply doesn't work everywhere. In fact, right now it appears to add business value only to a small and select group of sites - books, CDs and auction sites. Outside these categories, you run a far higher risk of doing your dough. The oft-cited Amazon personalisation success is not a harbinger of Internet commerce's future, but an atypical example, to borrow a notion from Jakob Nielsen.
Internet personalisation is not a Big Idea but a Small Idea, a Special-Circumstances Idea, a Use-With-Care Idea. It is an idea that most Web sites should, for now, dismiss.