In terms of Internet years, some online grocery-delivery services -- Peapod, Streamline, Webvan, ShopLink -- have been around forever, and some, like HomeGrocer, are just ramping up. But there's no doubt that an industry-wide shakeout has begun in earnest: Cash-strapped Peapod has been struggling against the double whammy of losing its CEO and key investors and a big push by the newer, better-funded Webvan. (International food provider Royal Ahold announced late last week it will buy 51 percent -- a $73 million investment -- of newly issued Peapod stock.)

Even so, Jupiter Communications predicts that online grocery-delivery business could jump from $63 million in 1997 to $3.5 billion in 2002 (even though that's still a bit like a fruit fly buzzing around a $443-billion-a-year industry watermelon). Clearly, despite vastly different business models, there's a lot here that appeals to time-starved consumers.

Because the grocery industry has an inherently narrow margin (an average of 1.2 percent), what will separate the winners from the losers in this online game is likely to be the ability to retain customers by creating Learning Relationships with them, one customer at a time. A recent joint study by Mainspring and Bain & Co. showed that the typical online grocer must retain a customer for 18 months just to break even.

But what's the biggest advantage to these online grocers? We think that, in terms of competitive strategy, it's the ability to observe a customer's behavior patterns, and to use them to build Learning Relationships that get smarter and smarter over time. For example, if a Webvan customer shops with the service long enough to create a "smart" shopping list, that becomes a very big incentive to remain with Webvan, even if a competitor offers a similar service at an introductory price. Home delivery offers a way to learn and remember individual preferences for each customer, to act on those in a meaningful way, to close the gap on missed opportunities through automatic replenishment, and to build bigger and bigger barriers to a customer's abandonment of the relationship.

Despite all the headlines, these firms, without exception, have been losing money. And because they basically offer the same thing -- groceries delivered to your home -- additional, customized services are their primary means of differentiating both the providers' offerings and their respective customers. (For example, I'll never use ShopLink's dry-cleaning service because my local dry cleaner has my business locked up. But a box of Cocoa Puffs is a box of Cocoa Puffs, regardless of who delivers it to my house.)

With all that's been going on lately, we were wondering how well all the players stack up in using the Web to create customer loyalty. So, we polled Peppers & Rogers Group employees across the United States and in the U.K. not only to find out how well they're being treated, but also how well the companies are incorporating our key IDIC strategies (Identify, Differentiate, Interact, Customize). Alas, our informal review shows most can do much better.

Peppers & Rogers Group employees eagerly heeded INSIDE 1to1's call for personal experiences with online grocery-delivery services (Remember, this is a purely anecdotal review!) We asked about several aspects of the services -- registration, ease of ordering over the Web, prices and selection, customer service, delivery, and most important, how the firms were able to identify, differentiate, interact and customize (IDIC).

While most ranked relatively high on customer service, there were some complaints that ranged from annoyance at having to repeat vital information when calling the company a second time and long waits on hold, to 24-hour delays in email responses, or no response at all. These issues tell us right off the bat that a CRM infrastructure hasn't yet been fully established at these companies.

There are a number of different business models used by online grocery-delivery firms, but the standard incorporates building a centralized warehouse to gather and deliver the order via company vans. Only Peapod sends its employees into existing bricks-and-mortar stores to do the shopping, lowering overhead, but shrinking margins. NetGrocer has its packages delivered via FedEx, requiring the greatest time between order and receipt of the groceries, and, as a result, doesn't offer dairy, meat, or fresh produce.

Because online grocery shopping isn't geared towards the bargain hunter, there are no generic lines of products. Respondents said that prepared meals took the biggest chunk out of their wallets, but did the trick in a pinch. Other small problems cropped up (as they do in traditional grocery stores): One respondent using ShopLink, for example, couldn't get a fresh pineapple, and some established name brands weren't available.

In most instances, delivery service was given high scores, but all respondents reported scattered problems -- the most common being missed delivery times and missing items (credits were offered in both situations). In one case, perishables that hadn't been ordered were delivered (the recipient was not charged, and got to keep them). There were few complaints about the quality of meat and produce, and several respondents actually raved about the freshness of the produce, particularly with Peapod and ShopLink. The majority of complaints ranged from "out-of-stock" items and less than seven days left before the expiration date on milk, to slightly melted ice cream (in the fall, not the heat of summer).

Overall, most of our respondents agreed that these services are pretty much what's expected -- worth the price in saved time and convenience. But, for me and most others, these services still don't eliminate a mid-week trip to the grocery store. (My goal is never to have to set foot in a grocery store again, so these services are only a 70-percent solution for me. If they expanded their product offerings or even actively solicited input from their Most Valuable Customers on what products to add, I'd be happier.)

All face similar logistical problems: Does ice cream stay frozen when delivered in August? Will a customer accept Jif if what she really wants is Skippy? Would a customer feel confident that he's getting fresh food if the service allows other customers to return a steak? What if an order left on the back porch has been pilfered -- does the customer take the loss or receive credit? If the reason you have bin-based delivery is that you have no garage or basement, where the heck are you supposed to keep these things so they don't get stolen? How these issues are handled can make or break these online pioneers. But, more important, creating a feeling of trust with the customer from the outset can help dilute negative feelings that result from an occasional flub.

In addition, most firms (such as Streamline and ShopLink) now offer deliveries of videos, dry cleaning, and processed film -- all additional opportunities for interaction and customization. We don't doubt that in the future, we'll be seeing more non-food items being delivered within hours of ordering. (New York Times best sellers, a hairdryer to replace the one you dropped getting ready for work, or a videotape to record that night's "Sopranos" rerun. Firms like Kozmo.com already specialize in more convenience-oriented deliveries in New York, Seattle, and other large cities. Recently, Amazon.com invested $60 million in Kozmo.com, purportedly to be able to deliver goods (CDs, books, and videos) in a one-hour timeframe. (We know of one New York University student who gets her Twizzlers and Diet Coke midnight cravings met within the hour via Kozmo.com.)

One final note before we get to the IDIC discussion: One of our U.K. staffers told us about a neat deal with Tesco, one of that country's largest grocery chains. Tesco is online, but it also supplies a CD to customers so their orders can be compiled offline -- saving phone time, which is essential in the U.K. The CD connects to Tesco via the Internet; the database of items and prices are updated, as well as any of the browser pages on the CD. The CD logs on to Tesco, you provide your credit-card details (no one-click order here!) and book your delivery slot. It's possible to save lists and reuse them. But a major downside, according to this respondent, is a lack of information or photos of the items on the site. Nor is there a recipe feature or interactive offers or suggestions. One note: The user has stopped the service because the most convenient delivery dates are booked two weeks in advance. Tesco has announced it's investing to fix this, but it will take some time. (The Sunday Times of London recently reported that Tesco has allegedly decided the way to further its Internet operations is to spin it off as a stand-alone operating unit, a la Walmart.com, focussed on the development of the Internet channel.)

Here's how the companies ranked against our IDIC model:

IDENTIFY: All respondents said they were "remembered" once they signed up for service -- obviously a requirement of all online grocery services. But on at least one occasion, we found that a customer's data wasn't readily available across all departments. One respondent who placed a phone call to ShopLink was asked to repeat all of the information she had already supplied to the Web site -- a clear sign that ShopLink wasn't integrating its offline and online data very well. But, on a positive note, ShopLink was the only company to offer two months' free service as a way to entice users to register.

DIFFERENTIATE: How well do these services differentiate between Most Valuable Customers -- say, a family of five with a weekly bill of $250 -- and the single professional who buys only a few frozen dinners and a six-pack of beer each week? The only evidence of differentiation we heard about from respondents were offers of different levels of service (depending on how much you're willing to pay) by ShopLink and Streamline. Both provide the option of supplying the customer with a refrigerator in the garage or basement for deliveries when no one's home, and a remote-code keypad for entry into a secured garage. No one reported receiving personalized purchasing recommendations, meaning that perhaps none of the sites are using matching engine technologies.

There's a real incentive for these sites to do this: Because you're not actually going into the store and seeing new items, you can't make an impulse buy. Therefore, it's in the best interest of the company to use matching-engine technology for product recommendations. For example, if a store knows you buy a lot of barbecue-type foods, it could recommend what others with similar buying habits have purchased. It could even partner with some brands to promote a new product by offering it free and asking for feedback. Recipes featuring the new product could be emailed to customers. (Unfortunately, even the free samples from ShopLink I received had nothing to do with my past buying behavior. I got coffee samples and I live in a caffeine-free household!)

There are still some bugs to be worked out -- for example ShopLink's site didn't recognize "string beans" (we found them under "green beans" instead). We also wondered how regional dialects would be handled -- suppose a user, originally from the Midwest, was looking for Pepsi and typed in "pop"-- he would see Pop Tarts and Orville Redenbacher's popcorn among other items, but no Pepsi. (Of course, typing in "soda" would bring up Arm & Hammer baking soda, along with Coke, Pepsi and 7-Up.)

INTERACT: In most cases, interaction with customers started immediately after registration. But, for my purposes, this was clearly for ShopLink's convenience and not mine because there was nothing the information taker told me that I wouldn't have been able to figure out on my own. In addition, the prime time for home visits was booked for six weeks out. So, they lost $200 per week in revenue from me.

After creating an account (either online or by telephone), in most cases an agent came to respondents' homes with a scanner to create a personalized list by recording what's already in their cupboards and refrigerator. No respondents mentioned being annoyed at the amount of questions they needed to fill out during registration. But because the customer logs on at least once a week to order, we feel there's an opportunity to use "drip-irrigation" questioning -- a little at a time, rather than all at once. After all, allowing the customer to shop and buy groceries hassle free, is the idea here.

Responses about the actual ordering process were a mixed bag. Respondents said the front ends of all the sites tended to be a bit clunky. On the bright side, ShopLink's drivers are CRM professionals -- they make notes on any evidence of a change in the lifestyle of a customer -- a new baby, a new dog, etc. This information helps ShopLink make an up-to-date profile. Other firms have established similar customer-service practices -- for example, Webvan does pre-and post-delivery email contact, as well as some targeted direct mail. (Webvan appears to be the most technologically advanced final-mile provider of the online grocery stores, and the most proactive in terms of CRM.) For example, of all the services we examined, Webvan is the only one to offer a live-chat support option, in addition to email and a toll-free number. Also, Webvan users have the ability to "jointly collaborate" on lists from different computers -- for example, a couple could each add to their joint list from work during the day, before it's submitted after work. (NetGrocer should do better: One respondent reported that it took seven calls to straighten out one order's worth of mistakes.)

CUSTOMIZE: In all cases, shopping lists can be saved for future reference, a service we associate with the most advanced Web sites. Other than that, we heard of no real standouts in this area from respondents. No one remarked about customizable Web sites, (because none of the services employ the technology) or multiple billing options, with the exception of Webvan. All services required an online credit or debit card, or authorized withdrawal from a checking account to open an account, but Webvan gives users the choice to pay by check or money order, but only for attended deliveries. With the exception of NetGrocer's "recurring orders," which allows the customer to receive a predefined order weekly, the saved shopping-list function is the closest these services came to offering automatic replenishment (except for Streamline's "Don't Run Out") -- a big disappointment. If online drug stores and even underwear sites can provide automatic replenishment, why not grocery stores? (I order two gallons of milk and two dozen eggs every week without fail!)