The Ultimate CRM Machine

By Tom Steinert-Threlkeld  |  Posted 2001-10-29

Denise Campbell has been hooked on the Internet since 1991. So it was no surprise that she was one of the first Avon ladies to sign up for Avon's "e-representative" program in the summer of 2000. Campbell launched and now operates her own Avon Web site,, checks the status of her orders online and sends her computerized customers e-mail alerts about Anew Retroactive age-reversal cream, Little Black Dress perfume and Skin-So-Soft bug repellent. Campbell even uses the Net to recruit new Avon ladies and to share information with those "downline" representatives reporting to her.

So why was this North Providence, R.I., Avon lady sweltering this summer in a booth on the grounds of the annual South County Hot Air Balloon Festival, rather than sitting in her air-conditioned home and raking in the Web orders?

Because, to Campbell, there's still no better way to sell Courageous Courage lipstick than to let a potential customer see it, handle it and ask face-to-face questions about it. Campbell handed out the latest Avon catalogs to some of her old friends, as well as to prospects. She hawked raffle chances for a basket of Avon goodies. And she shook hands, the old-fashioned way.

View the PDF -- Turn off pop-up blockers!

Campbell, who has sold Avon products for nine years, only gets only about 6% of her orders from her Web site, and another 6% from e-mail, despite her active Web efforts. Her biggest customer still doesn't have a computer, she acknowledges. So, house calls are still the best way to make direct selling work.

Avon's top brass knows this to be true, as well.

"Everything that you'd look at as a textbook version of what CRM is about, the (Avon rep) is doing with her customers," says chief information officer Harriet Edelman. "They know the product lines they like, the birthdays coming up. They do the customization, the personalization, the outbound marketing. It's all very one-to-one."

Indeed, Avon uses 572,000 women like Campbell in the United States—and 3.4 million like her around the world—to produce 98% of its $5.7 billion in annual sales.

But in spite of Avon's rep-centrism, this 115-year-old cosmetics and collectibles company not too long ago was considering whether to abandon its tried-and-true sales channel. At the height of the hype around the Internet and how it was going to revolutionize all of global business, Avon weighed the benefits and costs of the unthinkable: Doing away with the Avon ladies—and just selling to buyers over the Web.

The promises were plentiful. Instant global reach, with thousands of products and prices that could be constantly updated, around the clock. No printing and reprinting of scores of millions of glossy catalogs every two weeks. No handling pages and pages of order forms filled in with No. 2 pencils. Fewer worries about reps defecting or resigning.

In the end, technology for technology's sake wasn't the answer for Avon. The company ended up rejecting a pure clicks alternative in favor of a "representative-centric" Web sales model. And, for now, the company has held off from deploying customer relationship management (CRM) software to codify its customer-support functions. Instead, Avon has relied on the low-tech kitchen kibitzing of its Avon ladies to provide the personalization, customization, customer retention and other deliverables attributed to CRM packages.

: Redesign Processes, Boost Returns">

Goal: Redesign Processes, Boost Returns

Avon's focus on supporting the representatives makes sense now, in a world where old-economy companies are being rewarded for sticking to their knitting.

But it wasn't always so. Back in 1997, Avon embarked on a series of business process redesign (BPR) initiatives that were supposed to increase sales 8-10% a year, and earnings per share by 16-18% a share. Yet, despite efforts to improve forecasting, trim the supply chain and boost electronic commerce, Avon's sales only improved 5% a year. Earnings per share fell below the low-end growth target of 16% a year, until 2000. Cash flow of about $323.9 million was essentially unchanged from 1997 to 2000.

At the time when Avon kicked off Phase One of its BPR program, Wall Street and Avon's board of directors were pushing the company to use the Web to make over Avon's less-than-glamorous reputation. Avon, went the reasoning, could directly reach a far broader, wealthier customer base via the Web.

One possible way to alleviate the sluggish growth and modest gains in earnings experienced by the company in the late 1990s could have been to get rid of Avon ladies altogether. Selling directly to consumers over the Web seemed like a foolproof path to more sales and profit, almost immediately expanding the potential market for Avon products to everyone in the world with access to a computer.

The newly minted group, led by Avon marketing veteran Len Edwards, decided to hire outside consultants to look at the question of whether to ditch the Avon ladies in the name of progress—potentially, for one thing, saving commissions that total about 40% on each sale. Avon brought in Sapient, a consulting firm founded in 1991 to help businesses develop online and offline strategies.

"There was thinking that the Web provided an opportunity to 'turn the lights on' a massive direct-to-consumer operation almost overnight, without the cost that you would incur for launching a major retail operation with no bricks and mortar," recalls Alex Sion, director of the strategy group for Sapient's Washington, DC, office, and leader on Sapient's Avon engagement.

Sapient's mandate was to help walk Avon through the opportunities and risks for different ways the company might leverage the Web. Sapient, in conjunction with about 20 of Avon's business and technology managers, analyzed how Avon built relationships with customers traditionally; and, how it could change that in the future.

Some of the options Sapient and Avon mulled sound outlandish now, but seemed completely sane during the days of dot-com daffiness. The self-proclaimed "Company For Women" considered transforming itself into a pure content company by acquiring; creating a women-centric Internet Service Provider (ISP) or Application Service Provider (ASP) business; and/or partnering with a content provider like, for which Avon could provide the "commerce back-end," Sapient executives said. Sapient invited a number of Web experts, from companies ranging from Sun Microsystems, to America Online, to Netscape Communications, to act as a sounding board for the team.

"At one point, the whole group was thinking about going direct," says Sion. "They were looking for ways to politely disintermediate the reps."

Eliminating Avon ladies could have been as disastrous a change in a successful formula as Coca Cola's tinkering with its flagship drink and launching the ill-fated New Coke. And messing with Avon's well-established Avon ladies was just one of many strikes against the pure-play Web concept.

: Cut Out Costs, Not Reps">

Goal: Cut Out Costs, Not Reps

The numbers told the tale. Even if the technology to "disintermediate" the reps was available, the data gathered by Sapient showed the costs of going direct would have been prohibitive.

Avon reps have traditionally filled out paper forms that can reach 50 pages to place orders. If reps—or customers—place orders online, the cost falls dramatically. Edwards estimates the cost per online order at 30 cents, versus $1.50 for paper orders. With more than a million orders being handled a week, the difference to the company's annual profits of $478.4 million could be substantial.

But if Avon simply did away with reps, those savings would be outweighed by changes the company would need to make in its manufacturing, logistics, and other processes to support 100% Web-based sales, Sapient found.

Avon manufactures over one billion individual products a year. The company's well-established batch ordering, manufacturing and delivery systems were not set up for the "onesy-twosey" kinds of orders that occasional Web shoppers most likely would generate, Sion explains.

Avon's fulfillment model was based on a basic premise: Avon ships multiple orders for multiple reps in a single package, and reps unpack and customize those orders for each customer before delivering. If Avon had changed its shipment policy, moving to a complete "direct-to-customer" model, it would have had to redesign from scratch its delivery, fulfillment and other back-end processes.

At the same time, Sapient analyzed how much it would cost Avon to acquire direct customers in the way that companies such as had done. One area where Avon would have had to shell out major dollars in order to sell direct would have been in print, broadcast and TV advertising to build brand recognition—an area where the company has invested relatively little money to date. Instead, Avon has largely relied on good old word-of-mouth to spread the word about its products.

While Sion declines to provide the exact data that Sapient provided to Avon, citing client confidentiality as the reason, he did explain the thought process. "We gathered information on how much it would cost Avon to provide customers with the same level of quality as the agents (reps) provided." The bottom line: "Going direct was cost-prohibitive," Sion says.

Even acquiring 10 million customers—about the size of its offline business in Canada—could have cost $350 million, at $35 a customer in acquisition costs. That was considered a cheap rate for acquiring an online customer, back in 1999.

"The amount of investment with direct brand building that would have had to have been undertaken just didn't make sense for what kind of company Avon was," Sion recalls. "As we moved more and more down the analysis, it became very clear that the rep was Avon's core asset," Sion said. "Avon was a (more than) 100-year-old company with no established relationship with its customers. Their core competency was with their reps."

Instead, the analysis suggested that Avon adopt multiple avenues of selling its cosmetics to customers.

In this hybrid model, Avon's representatives would not only sell to their customers face-to-face, but online as well. At the same time, Avon itself would sell its products online via to customers who didn't like to buy from Avon ladies—customers president Edwards has come to call "channel rejecters."

Avon would also pursue other ways to sell more cosmetics, primarily through retail stores and kiosks that would be located at good old brick-and-mortar malls.

When faced with the numbers, and fortified by the recommendations of Sapient and the outside tech experts, the team threw its support behind the double-barreled approach of selling online, both through Avon's site, and thousands of e-rep sites on a complementary yet competing portal, called The team presented the plan to the Avon board in September 1999, and the board signed off on it. One year later, on Sept. 27, 2000, the revamped business-to-consumer (B2C) and business-to-business (B2B) sites went live, as did the company's online portal for reps, managers and Avon employees.

The distinctions and terminology have proved confusing. An Avon customer can buy Avon products from either or Avon reps manage their sales efforts through, but find news about the company and its rep programs through

And the company still has a long way to go to justify the $60 million or more Avon says it will spend on the Web over the course of three years, in pursuit of a strategy that has—in the words of general manager and president Edwards—produced miniscule revenue and profit results for Avon. Avon is still at least a year away from reaching the $100 million milestone in annual sales over and—combined. And even then, those sales would amount to less than 2% of Avon's business.

The company right now is not even pushing reps to sell products online. Instead, it is looking for the biggest impact to come from getting reps to order products online. If the company can get half its reps to order online, the savings from eliminating the collection of paper order forms across the country could be tens of millions of dollars a year. But, so far, a little more than one year since Avon launched its e-rep program, only 15% of Avon reps order online.

: Replace Paper With Electrons">

Goal: Replace Paper With Electrons

In the end, this is a story of trying to bite off more than one company should chew. From 1997 to 2000, Avon changed CIOs three times, and tried to launch multiple Web initiatives, retail initiatives and redesigns of a good chunk of its business processes, all at the same time. The company nearly had a "corporate nervous breakdown" in the process, current CIO Edelman acknowledges.

Avon traditionally has sold its products door to door through a network of reps, with paper forms keeping track of their progress; one channel, one way of doing business.

But in a matter of a couple of years, Avon tried to transform its paper processes to electronic ones; encourage reps not just to sell offline, but online as well; and move into multiple means of selling its products. Its reps would continue to sell in-person, but also on the Web.

As if all that weren't enough, the company would also start selling in retail stores. By now, Avon was supposed to be selling a new, pricier brand of cosmetics through both J.C. Penney and Sears stores. But Sears pulled out of the cosmetics business earlier this summer, leaving Penney as the only retail partner for the moment. And Avon is left trying to learn how to design, manufacture, distribute and market products in a whole new arena in which it has no experience. Even a seemingly minor change—such as the redesign of Avon cosmetic packaging so that a bottle would sit properly in a retail-store display—has to be thought through for the first time.

Even though Avon had neatly sidestepped the technological dislocation of its century-old representative-centric strategy, while simultaneously moving into the era of electronic selling, Avon's IT staff was still expected to provide tools that would effectively allow the reps to compete against Avon's own online and retail selling efforts.

"It was very stressful. Very, very stressful," says Edwards. "We finally all sat down and said, 'Hey, guys, we're trying to do too much too fast. We need to slow down a little bit.'"

It wasn't just on the Internet side of the house that Avon learned the hard way that it shouldn't tinker simultaneously with too many systems.

While it was undertaking a full-fledged overhaul of its Web-based order-taking processes, Avon also was attempting a wholesale upgrade of its largely mainframe-based order-management, campaign-management, financial reporting, manufacturing and supply-chain-management systems.

In 1993, Avon's corporate IT department, under the reign of then-CIO Ray Perry, instituted a program called Project FIRST (Fully Integrated Representative Sales Toolkit), which was slated to be a complete overhaul of Avon's existing homegrown order-management system. Avon's order management tracks representatives' orders through entry, processing and fulfillment at the company's distribution centers. Avon hired its auditor Coopers & Lybrand—which also was in the business of providing information systems consulting—to assist the company in deploying such a system.

After scrapping FIRST as too unwieldy, Avon toyed with the idea of trying to roll out globally an order-management system developed by its Argentinean subsidiary instead. But, according to former Avon IT executives, that plan also proved too complex, politically and technologically. So, Avon, in conjunction with Coopers & Lybrand, resurrected the FIRST project, this time under former CIO Ron Mastrogiovanni, a one-time Coopers & Lybrand executive.

But FIRST never materialized. The Securities and Exchange Commission currently is investigating the $15 million in costs that Avon wrote off as part of a special charge connected with FIRST. Avon decided in October to write off an additional $25 million it said it spent developing FIRST—which it had previously carried as an asset on its books.

Edelman declines to comment on FIRST, saying the company was not prepared to talk at all about the initiative. PricewaterhouseCoopers, which purchased Coopers & Lybrand, also declined comment, citing the SEC probe.

Says one former Avon IT executive who worked at the company during those years and requested anonymity: "It (Project FIRST) was a fiasco. It cost the company tens of millions of dollars."

: Streamline the Supply Chain">

Goal: Streamline the Supply Chain

Avon had other automation challenges on its hands. Like most major corporations, Avon was dealing with multiple supply chains for different cosmetic, retail and collectibles lines in different countries. On top of that, much of the company's demand forecasting and inventory management was handled manually. In cases where there was some level of automation, custom-coded applications were running on mainframes and were seldom linked to manufacturing systems, order-management systems or other related warehouses of data.

When Avon in 1997 kicked off Phase One of its overarching effort to redesign its business, two of its biggest goals were to upgrade its order-management system and automate its relationships with vendors, called supply-chain management. Avon projected that the first phase of its BPR effort would save the company $400 million over three years, half of which Avon said it would reinvest in building its representative base.

In September of 2001, Avon CEO Andrea Jung publicly stated that Avon did, indeed, reach its $400 million savings target. Avon achieved its BPR Phase One goal of streamlining the number of products and brands it offers its reps worldwide, she said. It made good on its commitments to reduce the number of suppliers it used, automate its call center in Springdale, Ohio, consolidate its returned-goods centers and cut staff where it could. But Jung said the company had not gotten far with its supply-chain initiatives.

"I certainly wouldn't call the first $400 million low-hanging fruit," Jung told analysts at a Prudential Securities conference in Boston on Sept. 5. "But a lot of that was margin expansion and tightening expansion controls. We didn't really get at the true end-to-end processing—with the exception of the beginning of supply chain in the U.S.—and the opportunity to transform the enterprise-wide processes that have been in place for decades and decades."

If several former Avon IT executives are to believed, Jung was being generous. While Avon is claiming that it saved over the past three years a total of $56 million in streamlining its product and material suppliers, and $22 million on improving its transportation management processes, it has done little to actually automate its supply chain, as it promised it would in 1997.

Six years after the company began deploying the first modules of Manugistics Group's demand-planning software in China, forecasting sales remains largely a matter of manual processes and conventional brainwork. Using a combination of automated systems and manual processes, Avon projects how it expects a new product to sell two to three weeks out, based on historical demand and trends. But Avon executives are the first to admit that forecasting still is based on judgment, as much, if not more, than on computerized calculations. And that might not be such a bad thing, since forecasting using computers is still an imperfect science.

Avon discovered just how imperfect demand-forecasting software is, as it attempted to roll out more of Manugistics' demand forecasting, transportation and fulfillment modules worldwide, according to former company technology executives. Two years ago at Christmas, Avon's supply chain problems were illustrated rather poignantly by an incident that remains fresh in the minds of a couple of those execs.

Manugistics' Networks Demand software was showing that Avon reps wouldn't sell many Pokémon watches—at most, a couple of hundred thousand worldwide. But 1998 to 1999 was when the Nintendo video-game characters were the hottest things since Barbie dolls. Executives attributed the resulting shortfall to a problem with the feeder data upon which the Manugistics' software was reliant in order to generate forecasts. This feeder data, stats on fast-moving products, likely sales, and other related information, were collected by Avon's various legacy systems and should have provided the kind of trend data that would allow Manugistics to predict future sales patterns.

"Manugistics should not have become the de facto standard at Avon," said one of the former Avon IT officials. "They didn't know where they'd get the information needed to feed into Manugistics. Basically, the feeder data was broken. The data was there, but it wasn't integrated into Manugistics and wasn't feeding into it in a timely manner."

Edelman, who was heading up the company's supply chain efforts at that time, was not amused, former company executives said. Ed Woroniecki, the Avon executive who recommended Manugistics, abruptly left the company. Some former Avon technology staffers say that Avon decided against renewing his contract. Woroniecki did not return requests for comment.

: Cut Through the Web Chaff">

Goal: Cut Through the Web Chaff

Starting this year, Avon is going to take another pass at tuning up its supply chain. Supply-chain renovation is one of the four main goals of Avon's Phase Two BPR initiatives, which the company outlined in May of this year. As part of Phase Two, Avon has committed to cutting eight to ten days a year from the total amount of inventory it keeps on hand. (In May, Avon said it had 119 days of inventory on hand; in September, that rose to 122.) In order to achieve these goals, Avon is retrofitting its demand-forecasting processes and systems, starting with a pilot "total demand visibility" system it is launching this fall in Mexico . Over the next three to five years, the company plans to roll out similar demand-forecasting systems in 10 markets, including the U.S.

The company recently created a dedicated demand-planning organization that will be completely separate from marketing. The organization's mission: Improve forecasting accuracy by 30%. At that point, Avon would be able to forecast sales to within 10 percent of what they would be.

Avon still is relying on Manugistics' Networks Demand system for general forecasting, but in January of 2001 added Churchill Software's Promotional Demand Forecasting package to the mix. A custom-developed marketing system feeds in data on known factors that cause sales to bump up or down, the effects of promotional campaigns, and other variables into Churchill's software.

The Internet also is big on Avon's agenda for the second phase of its business redesign. Avon is trying to make it easier for its reps to order online; find more incentives for Avon ladies to become e-reps; and, get more customers to buy online.

That Avon reps should be heading to the Web shouldn't be a new phenomenon. actually has been in operation for six years.

Cybergrrl/Webgrrl founder Aliza Sherman built a small Web site in 1995 for Avon to enable the company to promote Avon's breast-cancer crusade work. In 1996, Avon hired On-Ramp, a consultancy that is now part of AnswerThink, to build and host a full-featured site for the company. A revamped site, which was mainly an abridged, online version of Avon's paper catalog, went live in fall 1996.

Avon was not fully satisfied with On-Ramp's work, Edwards says. The company was not up to the task of hosting the site, he claims. But On-Ramp, according to one former AnswerThink consultant who requested anonymity, had warned Avon that it planned to exit the hosting business, since it was outside On-Ramp's core competency area. Their contention: Avon execs had failed to take heed.

In the end, Avon decided to take back control of the site and to host the site itself. AnswerThink officials declined comment on the company's work with Avon.

AnswerThink isn't the only vendor that Avon ended up dumping. Avon hired FutureBrand HyperMedia, a Toronto-based branding and retail consultancy, to help move sales rep catalog copy onto the Web site. According to Edwards, FutureBrand was unable to deliver on this project, with the company claiming that the content in Avon's rep catalogs could not be captured and reconstituted easily on the Avon Web sites. Avon switched to P2i Group, a Bethlehem, Pa.-based company focused on converting print catalog and other content into Web content, to publish to the Web site its many pages of catalog material. FutureBrand executives did not respond to several calls requesting comment as to why their contract with Avon ended.

There were other false starts for Avon on the Web side of the house. The company briefly established its unit as a separate, standalone subsidiary. But Avon quickly saw the potential channel and management conflicts such an organization could spawn and rapidly reabsorbed the group

Avon's seemingly innocuous move to include the address on one of the spring 1999 catalogs in the hopes of driving more Web business was ill-received by the rep community. Some reps thought their in-person selling efforts were being used by the company to take away sales from their customers. After all, this was before the launch of, where reps were able to set up their own Web selling sites. The company ran

"The decision to put the URL on the catalog caused quite a firestorm," recalled one of Avon's integration partners, who spoke on the condition of anonymity. The reps thought "Avon was going around them and that caused some backlash."

To its credit, Avon did listen. Avon hasn't made any additional attempts to feature the address on the catalogs, the prime selling tool of reps.

Not only did the company listen, it also learned. When it originally designed an automated order-entry system for e-reps, Avon discovered that the most streamlined user interface for rep ordering wasn't the most intuitive one for the e-reps. Avon had to backtrack and reintroduce a simpler method of ordering online that it first introduced in 1997 as Now, Avon reps ordering products at can enter all details of orders they place by using the numeric keypad on their computers, much like using an adding machine or calculator. They don't have to use the alphabet. If they want, they can see the names of the products they've ordered—or just the stock-keeping numbers, if they are confident of what they're doing.

Avon's Web strategy has been a work in progress. So, too, has the infrastructure powering the, and sites.

In order to drop AnswerThink as the primary developer and hoster of, Avon needed to build a capability to host a transactional e-commerce site. Such a site would need to be integrated with Avon's existing order-management, product management and pricing systems.

To build such a system, the Avon CIO in 1999, Sateesh Lele, hired nearly a dozen consultants and software vendors. Lele, a highflier who had served as CIO for General Motors Europe, selected IBM Global Services as the lead integrator on Avon's Web sites. Other players Lele brought in included Modem Media, which did design and branding consulting; Verilytics (which acquired iBelong), which designed the portal; and Kinzan, which developed the B2B portal.

"Sateesh championed the whole idea of bringing in outside experts to get things done," recalled a former Avon IT executive, who requested anonymity.

But Lele lasted in the CIO job less than a year. Integrators who worked with the company said he never really had corporate support. One former Avon IT executive said Sele was "less fluid" than other Avon executives. Lele declined several requests for comment on his tenure as CIO.

Lele was out, but his legacy remained. Managing the consultants turned into a job of its own, and one that lead integrator IBM Global, in conjunction with Avon's own IT staff, was charged with accomplishing.

IBM Global suggested that Avon adopt IBM's WebSphere Net.Commerce middleware (now known as IBM WebSphere Commerce Suite) as the crux of its e-commerce system. At the heart of Websphere Commerce Suite is IBM's application server, which provides the business logic for a variety of e-commerce-related applications, ranging from Web presentation to transaction processing. The platform allows for the development, deployment and management of custom-developed and/or third-party-developed Web applications.

IBM Global executives said the company offered Avon other possible choices for its middleware platform, but declined to be specific, citing confidentiality restrictions.

In late 1999, when Avon was deciding on which commerce platform to build upon, there were only a handful of robust middleware options to provide glue between the company's databases and Web applications. Finding good middleware was especially key since Lele and his team had decided that they were going to rely on Sun Solaris servers to power Avon's Web sites.