head: Betting on Linux at Andersen Printing

head: As change sweeps an industry - a $1,299 Linux machine saves one company two million a year

DECK: Author Paul Murphy's previous Linuxworld article on making the Unix decision suggested that Unix is usually a smarter business choice than Windows. The current article, sixth in a series, looks at what it takes to implement that knowledge.

The "I" in this scenario is that of a systems consultant brought in as part of a team whose aim is to help a small business owner plot a long term survival strategy for her company.

Ann Andersen Printing, its staff and plans, are fabrications but the situation presented, and the remedies offered, reflect the author's experience with a real-world organization facing a broadly similar problem. Andersen Printing is imaginary, but the conditions, decisions, and outcomes described are broadly based on real events.

2. Background

Ann Andersen began her printing business, in 1982, as a one woman job shop specializing in low volume, high quality color work. In 1984 she got her first monthly magazine contract and recruited two friends to work on production and sales with her. Today, Ann Andersen Printing has 320 employees, a fully modern plant, two sales offices, and expects to close fiscal 2001 with about 61 million in revenues --but the same three people still share executive responsibility.

All three are now over fifty and weighing retirement against the impact selling the company would have on it, the employees, and their own lives. As part of that process they invited two representatives of the staff association to participate in development of their five year succession plan and hired a firm I sometimes work with to facilitate a strategic planning retreat.

According to Julie, general secretary in Tax Planning and the secretary who called to pass on her boss's request that I attend the second day of their meetings, the client was being difficult and refusing to accept the reality that Andersen should be sold to a bigger player. Apparently her boss and a printing industry expert brought in from the firm's San Francisco office for the occasion, had put an M&A [Mergers and Acquisitions] guy on standby to join the meeting and gone in expecting to talk about divestiture strategies only to confront a client executive that wanted to talk about exploring a future in digital media delivery.

In twenty years of early morning meetings I've learned two things:

  1. it is not possible to turn down a hot blueberry muffin; and,
  2. you can tell more about the real positions taken by participants by watching how, when, and with whom, they arrive at these things then from listening to them all day.

The place was empty when Julie and I arrived, but the clients came in, as a tight group of five, right behind us and only Zina (their sales manager) went for coffee before settling at the table. That meant they'd been meeting about us somewhere earlier and underlined Julie's expressed concern about a rift between them and Bill, the tax planning partner facilitating the meeting.

Because I had done work with them before, I got the rest of the story during the pre-meeting chit-chat. Their unhappiness came from the perception that the printing expert was treating them as local yokels and repeatedly explaining the obvious, while they saw Bill as focussed on getting them to let the firm handle the sale of their company.

Bill showed up almost exactly on time, indicating, I thought, a wish to not give offense by being late while avoiding being button-holed. He got started right away too with the re-cap of yesterday's meeting, despite Mark somebody, the printing expert, not having appeared yet.

Ann tried stopping that cold. "Yesterday", she said, "we talked about all the reasons that the company cannot survive. Today we want to talk about making it; not just survival, but growth; doubling our size in the next five years."

But I wasn't with this group yesterday so, knowing who would sign off on my invoice for today, I asked for a brief re-cap on why Andersen could not survive, "just," I said, "to understand what we're up against."

Bill's basic argument is that:

  1. printing exhibits significant economies of scale: the bigger the plant investment, the lower the per page cost;
  2. lower per page costs translate to a lower cost of goods sold per dollar of revenue only if the volume is large enough to reduce the impact of high capital costs;
  3. the cheapest and fastest route to high volume is to buy up smaller printers which, like Andersen, have a few big clients whose work can be transfered to the main production plant; and,
  4. direct competition (buying bigger printers) won't make sense for Andersen both because of the capital cost hurdle and because most of the consolidators don't make money.

This is the same problem that's hurt a lot of airlines. People selling bigger airplanes, or bigger print trains, can make beautiful and convincing PowerPoint presentations because doubling capacity produces a large percentile decrease in variable costs per unit - whether measured in cost per seat mile or cost per page. Throw in a "two thirds utilization" assumption for some pseudo realism and it's not hard to have visions of fat profits dancing in your victim's heads.

Once they own the thing, of course, they find that maintaining two thirds utilization requires them to double their previous volume - something they can only achieve by price cutting. Throw in a competitor or two in the same situation and everybody ends up with a lower contribution margin per unit sold and higher capital and maintenance costs to defray.

There are two well proven solutions:

  • go through bankruptcy to ensure that the successor company gets the plant at five cents on the dollar; or,
  • establish a local price-point monopoly through vertical integration and/or by buying up the competition.

Refusing to sell out, Bill says, is not an option. Although buying Andersen is the cheapest and fastest way for the consolidators aiming at the local market to get Andersen's biggest clients, it is not the only way. Epicentre Graphics will, he says, go after those accounts by trying to buy the parent companies or by directly selling into those accounts. Eventually Andersen's higher per page cost will count against it and the business will go to Epicentre with, or without Andersen's co- operation. That being the case, selling out now is, he says, the only way to preserve business value for Andersen's owners.

The Andersen people know this. Their two highest volume contracts print regional editions of nationally distributed specialty magazines on locked in contracts that run to June, 2003 and January, 2004 - but their cost for paper is higher than Epicentre's variable cost per finished page and they know perfectly well that the numbers are running against them.

The counter, one of the association representatives says, is to move to an all digital print system and specialize in fast turn- around on high value added jobs with smaller volume.

This idea sounds pretty good - but it doesn't look like a long term winner to me. When Bill asks what I think; what I think is mainly that I now know why he wanted me here. If the inverse relationship between page costs and printer sizes is crushing Andersen from the top, improvements in PC software and low volume printer technology are squeezing out the company from below.

This unhappy state of affairs has come about now rather than 15 years ago because of recent changes in the adoption of Postscript and basic font display technologies in the PC world. When Postscript first came out in the early eighties, it offered significant advantages over previous technologies and was almost instantly adopted throughout the printing industry. The PC people, however, rejected PostScript, possibly because of its close association with the Macintosh, to ghetoize themselves behind the 80 character PC-DOS screen and the canonization of embedded printer control languages like HP's PCL.

The result gave the print industry a tremendous competitive advantage over in-house production for the design and proofing stages of the print workflow. Since printers compete on printing cost but make money on design and fulfillment, companies like Andersen could leverage this advantage to grow quickly and profitably.

Today, however, newer Windows products are starting to offer improved display functionality, Mac-like font management, and access to Postscript device drivers for use with cheap printers. As a result the "PC tariff" behind which the printing industry has been protected from the changes affecting other manufacturing businesses is now breaking down and taking Andersen's competitive advantage over in-house production with it.

You can see this happening anywhere you look at the print business. Even two years ago, for example, Andersen did lots of high margin work on things like designing company newsletters - but that business is drying up, the customers are simply doing the design work themselves and printing it internally or at digital print shops like Kinkos.

Last year Andersen did some personalization work - the kind of thing where the client wants to print custom ads on the monthly statements they send their customers- but there are fewer of those jobs this year, and by next year most customers will be doing that in-house too and for the same reason: the combination of PC software and PC printing is getting better and more usable all the time.

It's the same story with newer product offerings like media asset management (tracking and indexing workflow elements like text or imagery) as a component of a digital print workflow. Some of Andersen's clients have tens of thousands of stored images and whole pieces of pre-pub workflow - much of which they can't find when they want it. Offering these guys an asset management service sounds like great value - they could store stuff on Andersen's servers, work with low resolution imagery in-house, and print from high resolution stuff without paying for the enormous bandwidth required to transmit those images back and forth.

But, a simple package like Cumulus is 99 bucks for the PC - and will handle several thousand pieces with no problems. By next year a lot of these clients will be doing this kind of thing for themselves - and then printing on low volume gear in-house.

The result, as shown here, is that Andersen is caught between two irresistible forces:

  1. attempts by big companies to capitalize on returns to scale are making big jobs unprofitable; and,
  2. improvements in PC software and low end printing technologies are reducing demand for Andersen's services on smaller, more profitable, jobs.

It gets worse, too. There's a third force at work: internet distribution both de-centralizes in-house printing and cuts into Andersen's remaining market by reducing its scale.

One of my mainframe clients owns two Xerox Docutech printers that produced about 40 million pages last year - mostly for reports that were then distributed by courier and intra-office mail. He's got a few months to go on his printer lease but we shut down variable maintenance and nearly all of his materials and distribution costs by redirecting the mainframe print output stream to a $1,200 Linux box. There a PERL script calls txt2pdf ($99 from sanface.com), converts each report to an indexed PDF, and then loads the results into ZOPE for delivery via the company intranet. What's left of the print job is handled on a much smaller IBM printer and is mainly head office stuff that doesn't require courier delivery.

As a company they're getting three clerical level FTEs in Systems and over two million a year in cash savings out this - even though the per page costs on the IBM are nearly 40% more than for the Xerox machines.

One of the side effects is that the managers who get these reports; people who previously pulled out the pages that interested them and threw the rest away, are now printing a few pages locally and throwing the PDF away. At 17 cents a page for printing on an inkjet that nets out to a cheaper way to print the report because so many fewer pages are printed and distribution costs are gone.

For the next step he'll be changing the process. Instead of generating the PDFs and pushing them out via the web, we'll be generating PDF pages on the fly as people ask for them. The reason for the change from push to pull is that he'll be also changing his approach to low end printing. All those inkjets and personal lasers he now has to support are going - they'll be replaced with 16 and 24 page per minute departmental lasers. Since those would tempt people to go back to printing whole reports, we'll use "pull" to only give them the pages they actually want.

This is a win-win for everybody in the company: he'll look like a hero because users will get better print performance, full color, and less hassle; departmental managers will get budgetary savings; and senior management can start a drive to cut paper use - something that will eventually produce savings and efficiencies dwarfing what he got by de-commissioning his Docutechs.

What works for big companies like his, will work for a lot of Andersen's printing clients too. Joe's Tools, or whatever, will be putting in a website, and letting his customers download a catalogue PDF - one he can change overnight if he wants too and for which he doesn't pay Andersen anything.

3. Solutions

After a break, we focus on trying to develop some viable options. Mark deGroot, the production manager at Andersen, suggests using something like an Indigo [a combination laser/offset press with full digital input and control] to provide backend services for companies which do all of the workflow except printing and delivery. Wouldn't that combine economies of scale with customer expectations about rolling their own?

It would, those things get close to paper costs for full color printing, run around 240K, and can turn out 8-12 thousand pages per hour. They're too rich for most of Andersen's customers to own in-house, but might give Andersen a price point that makes sense for many. It's an intriguing concept; except that it cannot, by itself, support a 320 employee company.

The core of Mark's idea is to provide a completion service that lets the customer pick whatever elements of the workflow the customer wants to do, then completes the rest to get the customer's information product out there. In most cases, therefore, Andersen would provide some media asset management services and archiving, expert advice, and output services - consisting of printing, cutting, binding, personalization, and delivery or fulfillment co-ordination.

The big problem with this is that the volume times the margin doesn't produce enough revenue to run the business -even if selling costs are effectively zero. $61 million in annual revenues means that you bill someone a quarter of a million bucks every working day. If you made 3 cents on every page, you'd have to print almost half a billion pages a year - roughly 832 a minute, 24 x 7 - just to meet payroll for 320 people.

Mark's basic idea is a good one -for a smaller company- the volume needed to continue the existing company is just not going to be there but, create a diversified portfolio of Andersen companies that do different things, and one of them could do this very successfully.

So I suggest that Andersen be split into two groups: Andersen Printing would hold the two big contracts and the people and facilities that support them. This company would be structured for an easy sale to Epicentre or somebody that wants to come in here and compete with them. It will actually be worth more as a stand-alone than in combination with the rest of the business because the buyer won't be facing shutdown costs on the 230 or so people, and 2500 small customers, he doesn't want.

Think of the other group as Ann Andersen Venture Capital; a company with initial investments in the other major divisions, but restructured as a bunch of competing entities. There would be a warehousing and fulfillment company, a digital art company, a traditional small job printer, a digital print company, a bindery, and, of course, an integrated IT services company.

There's a big advantage to this plan: the owners get to sell the business and keep it too. A sale doesn't happen instantly; there's time to to move people around and brief customers before the sale, and time afterwards to re-tool the spin-offs and start them on the path to growth before the competition can react.

Mark somebody, the guy from San Francisco, doesn't like this idea. "These companies will be priced out of the market," he says; "the job printer will be stuck with a couple of old Heidelbergs, five times Epicentre's cost per page because of all the hand work." Right, but it also has skilled craftsmen, a loyal client base, and the ability to print on substrates, and in sizes, that are out of scope for most in-house printers. The jobs it does won't attract competition from Epicenter; they're too small. Even if Epicentre had the gear and people for this, their big company overheads would kill them.

"Why", Zina asks, "have an IT services company? that's not an expertise we have."

I have a short answer for this: "companies which don't control their IT investment, don't control their businesses" but a more careful answer is that there are profit opportunities that go well beyond saving money on meeting in-house needs.

Andersen's past earnings have come mainly as returns to capital - because the owners made investments in plant and equipment their customers needed to use; and as returns to knowledge - because Andersen has expertise the customers lack. That candle is being burnt at both ends as its markets erode: companies like Epicentre have more capital and so gobble up the big stuff, while the combination of better PC software, the web, and lower cost departmental printers devalues Andersen's expertise to reduce margins and opportunities at the low end.

For there to be a future beyond selling out, Andersen needs to get earnings from something else - that's going to be returns to organization - the ability to change business direction and processes on a dime, ahead of the customer's own recognition of the need for those changes. Returns to organization come from combining people and information; that means having the IT house in order.

In the past Andersen's approach to systems has consisted mainly of benign neglect. That's worked because the business pressure to change just hasn't been there - most orders traditionally come by fax. But, that's changing very quickly as Andersen competes for smaller, more sophisticated, business. Andersen needs better functionality and greater reliability. Right now Andersen's sales people rely on a few big contracts to keep the bucks rolling in; reduce the average contract size and the sales people will be under new pressure to perform - which means they won't be willing to wait five minutes for their email to come up - or be forgiving when some of it magically disappears.

The question is therefore one of how to do that - how do you make systems work better? transform that systems cost into a business advantage? what's the infrastructure for the IT company going to look like?

Everyone here knows my answer: Unix with smart displays and a few Macs or PCs where there's an advantage to be had from using them. That's a general answer which may, or may not, apply to you. We don't know what printing requirements will look like next year, never mind being able to predict all the way to 2005. That uncertainty exists for any architecture, so the question is where to place your bets.

In this context the most important thing about Unix is stability. You get your staff on smart displays and that part of the equation doesn't change for the next five years. The software will change, you may change the servers, but the basic access skills and desktop devices don't change. That makes it easier to adapt to the things that do change, and avoids not just hardware cost -functionally trivial anyway- but retraining and user stress. That's big stuff, get the hassles out of the way and use the freed up mental bandwidth to go after bigger stuff -returns to organization; the combination of people with information- just being better at the business than the other guy.

There are some problems and limitations. Right now some of the software Andersen needs isn't available for Unix; but its not all available for the Mac or Windows either so you'll have a mixed environment for some time. The growth right now is in software for Linux, and that doesn't look likely to change. More importantly, you can help it along. Every time you spend a nickel on software, demand a Linux solution - even if you have to settle for something else upfront, your dollars will help bring better products forward.

Unix will help the Andersen Ventures companies succeed, but the advantage game can be played in the negative too. Windows is a very big thing with lots of people who don't understand technology and that can be used to Andersen's advantage: if the enemy has costs you don't, you win. Windows is expensive in ways that go well beyond checks written. There are security issues, staffing issues, organization issues, productivity issues - all of which cost money and tie up management, so encouraging your competition to use Windows will help you beat them.

Someone suggested earlier that Andersen Printing would be worth more to Epicentre as a stripped down spinoff than as an integrated company. You can do even better in terms of raising your selling value, remove the question mark about your accounting system by having someone from here [the accounting services firm I was working with] migrate that business to Great Plains on Win2K. That'll help make the sale, and you can upgrade the new companies from the SCO/RealWorld you use to Samco Realworld on Linux and give yourself a big competitive advantage.

I'm working on a tirade here, but I'll say it one more time and stop: when the other guy uses Windows, you'll always get a competitive advantage out of Unix with smart displays. To succeed in the coming years the new Andersen companies will have to combine small scale and wide scope, tight management and entrepreneurial free rein. Those aren't contractions, they're components of a strategy built around getting returns to organization instead of capital or specialized skills. I can't tell you how this is going to play out, but it's time to place your bets on the information system to support it - and my recommendation is that the place to double down is Linux.