3
COLUMN THE PRACTITIONER’S BOTTOM LINE DavidShepard f This issue of the Journal of Interactive Marketing is about predicting, tracking, changing, and understanding customer behavior, which should mean that there’s something for everyone. Let us know what you think. FROM SIMPLE DECILES TO DIFFERENTIAL EQUATIONS Rick Courtheoux’s Analysis of List Segmentation Efficacy and Cam- paign Optimization proves the point that on any given Sunday the best practitioner can make issues just as complicated as the best academic . . . not saying that that’s necessarily bad. Rick takes the relatively simple decile analysis so popular among direct market- ers and, with some fancy mathematics and simple financial anal- ysis, demonstrates that two of the most common problems asso- ciated with decile analyses (higher than would be expected response rates in the bottom decile and splitting marginal de- ciles) can be overcome by “smoothing” the decile curve (using a fairly sophisticated smoothing method—not your father’s simple exponential) and applying some basic economics to the smoothed results. Rick’s equations also suggest ways to evaluate alternative contact and pricing strategies, but as always, and as Rick points out, the critical issue is not the spread of the pre- dicted response rate but the accuracy of the prediction itself, © 2004 Wiley Periodicals, Inc. and Direct Marketing Educational Foundation, Inc. f JOURNAL OF INTERACTIVE MARKETING VOLUME 18 / NUMBER 1 / WINTER 2004 Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/dir.10075 2 DAVID SHEPARD is president of David Shepard Associates, Inc., a direct marketing and database consulting firm in Dix Hills, NY; e-mail: [email protected]

The practitioner's bottom line

Embed Size (px)

Citation preview

C O L U M NTHE PRACTITIONER’S BOTTOM LINE

D a v i d S h e p a r d

f

This issue of the Journal of Interactive Marketing is about predicting,tracking, changing, and understanding customer behavior, whichshould mean that there’s something for everyone. Let us knowwhat you think.

FROM SIMPLE DECILES TO DIFFERENTIAL EQUATIONSRick Courtheoux’s Analysis of List Segmentation Efficacy and Cam-paign Optimization proves the point that on any given Sunday thebest practitioner can make issues just as complicated as the bestacademic . . . not saying that that’s necessarily bad. Rick takes therelatively simple decile analysis so popular among direct market-ers and, with some fancy mathematics and simple financial anal-ysis, demonstrates that two of the most common problems asso-ciated with decile analyses (higher than would be expectedresponse rates in the bottom decile and splitting marginal de-ciles) can be overcome by “smoothing” the decile curve (using afairly sophisticated smoothing method—not your father’s simpleexponential) and applying some basic economics to thesmoothed results. Rick’s equations also suggest ways to evaluatealternative contact and pricing strategies, but as always, and asRick points out, the critical issue is not the spread of the pre-dicted response rate but the accuracy of the prediction itself,

© 2004 Wiley Periodicals, Inc. and

Direct Marketing Educational Foundation, Inc.

f

JOURNAL OF INTERACTIVE MARKETING

VOLUME 18 / NUMBER 1 / WINTER 2004

Published online in Wiley InterScience (www.interscience.wiley.com).

DOI: 10.1002/dir.10075

2

DAVID SHEPARD is president ofDavid Shepard Associates, Inc., adirect marketing and databaseconsulting firm in Dix Hills, NY;e-mail:[email protected]

which involves such decisive and often uncon-trollable factors as the economy, seasonality,competition, and internally generated fatigue.

SO, HOW ARE THINGS GOING?Suppose you knew, for each of your online cus-tomers whether their behavior pattern waschanging, and by behavior pattern I mean thelength of time between visits to your Web site.Obviously, the choices are: The length of timebetween visits is (a) getting shorter, (b) gettinglonger, or (c) staying about the same. If youknew this with some degree of certainty, youcould probably think of a number of marketinghypotheses to test.

Well, apparently this is possible, so say WendyW. Moe and Peter S. Fader in their article Cap-turing Evolving Visit Behavior in Clickstream Data(Actually, not being anything like an expert inprobability distributions, I’m assuming theirconclusion is that this can be done; otherwise,why would they have written the article?) So, ifyou think that this information could help youmanage your site better (everyone’s speedingup, better get up to speed) . . . or foresee acoming crisis (everyone’s slowing down), or ifyou’re just looking for help in better targetingyour online budget, give this article a shot. Butbeware: It contains more symbols than The DaVinci Code.

WHAT WERE THEY THINKING ABOUT?What motivates consumers to articulate themselves onthe Internet? What a great title for a journalarticle: Think of all the spin-offs, actually it’s thesubtitle of the Electronic Word-of-Mouth via Con-sumer-Opinion Platform, an article written by ateam of academics whose names will undoubt-edly appear in the table of contents, so I won’trepeat them here. Basically, what they did isthey searched the literature and came up with11 possible reasons why someone would be-come involved in eWOM (electronic word-of-mouth) communications. Then they conductedonline research among 2,063 voluntary re-sponders, all of whom had participated ineWOM. Twenty-seven questions, one PCA, eight

factors, four clusters, and one discriminantanalysis later, they came up with the answersthey were looking for.

To my great surprise, the motivations I ex-pected to see come to the top of the list—venting negative feelings and extraversion/pos-itive self-enhancement—were at the bottom ofthe motivation list (I should mention that theresearch was done in Germany. Bet the resultswould be different if done in New York.) Fi-nally, the authors make some easy-to-follow rec-ommendations as to how the results of theirresearch can be put to use. So, if your chatrooms get a lot of attention, I recommendspending some time with this article.

GETTING YOUR FAIR SHARE?Do you think it’s possible to measure your site’svisibility (defined as the extent to which a usermay come across a reference to a company’sWeb site in his/her online or offline environ-ment) without extensive and expensive surveyresearch? The authors of The Measurement ofOnline Visibility and its Impact on Internet Traffic,Xavier Dreze and Fred Zufryden, think the an-swer is yes.

Here’s what they did: First, they studied 100companies in 10 categories and found 53,017links, on average 530 links per company. Butnot all links are of equal value. Intuitively, onewould think, and prior research showed, thatlinks higher up on a page are more valuablethan links lower down on the page, and if youbelieved that visitors followed a common paththrough your site, that links on the pages onecomes to first would be more valuable than thelinks that followed. (The former argumentturned out to be true; the later hypothesis wasnot supported.)

The other two “online” factors which logicallyimpact visibility are listings in online directoriesand the results of online searches. Finally, theauthors added some measures of offline andonline advertising expenditures and quantifiedall five factors. Next, they interviewed 5,000 on-line users to determine if their perceptions/recollections/impressions could be predictedby the variables described earlier. The short

T H E P R A C T I T I O N E R ’ S B O T T O M L I N E

JOURNAL OF INTERACTIVE MARKETING ● VOLUME 18 / NUMBER 1 / WINTER 2004

3

answer is yes. The managerial implications arethat by following the methodology outlined inthe article, companies can create their owntracking system, measuring/indexing their per-formance against their competition.

TRUST ME . . . THERE MAY BESOMETHING HERE FOR YOUGetting the first sale is always the hardest . . .especially for small online retailers without anational reputation. So what is one to do? Theauthors of Signaling the Trustworthiness of SmallOnline Retailers have some suggestions for thingsto try.

Their basic hypothesis, as I read it, is thatonline retailers can emphasize at least fivefeatures that should intuitively affect trust.

They call these features “clues of interest” andinclude (a) seals of approval, (b) return pol-icy, (c) awards from neutral sources, (d) se-curity disclosures, and (e) privacy disclosures.In the absence of a “real” retail environment,they built their own site and asked 400 or sostudents to take part in the study. Each stu-dent was exposed to one of 16 versions of thesite, each version differing in the emphasisgiven to each of the five variables outlinedearlier. The bottom line . . . depending onwhat you emphasize, you can influence behav-ior measured in terms of “willingness to pro-vide personal information” and “bookmark-ing” a proxy for the intent to revisit the site.But, not all factors are of equal importance. . . as I encourage you to see.

J O U R N A L O F I N T E R A C T I V E M A R K E T I N G

JOURNAL OF INTERACTIVE MARKETING ● VOLUME 18 / NUMBER 1 / WINTER 2004

4