The rise and fall of ArsDigita

Diary of a Start-Up

The rise and fall of ArsDigita
and lessons from the world of venture capital

by Eve Andersson, February 2002
Source

This is a story about a company. A company that was profitable from Day 1. A company that built products that were useful to many other companies. A company that had ethics, that treated the breadwinners (programmers) with respect, a company that could afford to help people and give away software and training, while still having enough left over to grow and save a few $million in the bank.

That is, until the venture capitalists arrived on the scene. Lying to customers and employees became commonplace. Greed replaced philanthropy as each of the company's unique programs was dropped. But, this is a company, and the goal is to make money -- any positive impact on the world is secondary, right? The real question is: how much money did they make?

The technical and managerial incompetence of the VCs and those they hired drove the company into the ground. All but 10 of the 240 employees were fired, laid off, or quit. All of the $40+ million in venture capital was squandered. The monthly operating profit turned to loss as more talentless executives were hired who threw out the company's old, useful products and put their blind faith in engineers who spent millions building complicated software that solved no business problems.

This is a story that will teach you something about building a software product, about profitably running a company, and about what can happen if unqualified organizations obtain control.

The Birth of ArsDigita

Since the early days of the Web, I had been building web sites for big-name clients in California with my friend Aurelius Prochazka. Meanwhile, Philip Greenspun and two of his friends, Jin Choi and Tracy Adams, were doing similar work out in Massachusetts. When I moved to Massachusetts in 1998, we joined forces and were quickly able to quickly attract high profile clients, such as Levi Strauss, Environmental Defense, and MIT Press.

Our band of programmers, called ArsDigita ("Digital Arts"), was profitable from the beginning. We had no office, no marketing staff, no letterhead. But we did have a few thousand dollars' worth of computer equipment and our five motivated selves.

The Birth of the ACS

It didn't take long for us to realize that we were solving some of the same problems over and over again for each of our sites. It doesn't matter if a site sells custom-made slacks, lets people share their photography knowledge with other enthusiasts, helps people fight environmental battles against companies polluting their groundwater, facilitates the trading of financial instruments, or helps people find the best combination of red, white, and sparkling wines for their next soirée.

Every site we built needed:

1. to talk to a relational database management system (RDBMS) to facilitate the collection of content from its users
2. a means for registering users and recognizing them upon their return to the site
3. a permission system to enable administration of the entire site, a section of the site, or an individual content item
4. a mechanism for grouping users so that they can be served content or given permissions appropriately

Instead of reproducing this for each client, we wrote a general data model, a few web pages, and some shared procedures, and called this collection of code the ArsDigita Community System (ACS).

We distributed the ACS free and open-source, not merely to be altruistic, but because it made sound business sense. If you are doing professional services, the best way to get your name out, have more clients find you (yes, free marketing), and improve the code base is to share it with the open-source world.

The ACS began as a small core of functionality, but more opportunities for code reuse quickly arose. Many of our clients wanted discussion forums. Quite a few needed ecommerce. Polls and portals and intranets and calendars and address books were repeatedly requested. It would have been foolish to build this functionality one-off for each client. It would have been a wasted opportunity to build this only for ourselves and not let our work reap the benefits of an open-source release. Hence we began the release of ACS modules that could run on top of the core. Dozens of useful modules were created, and that is what led to the adoption of the ACS by programmers and companies world-wide.

Growth

A small group of developers earning lots of money, making clients happy, and developing and releasing a useful software product is wonderful, but ... to make a substantial impact on the world, you gotta grow.

And grow we did. At first it was difficult to hire people because developers don't feel safe working for a company with no office or regularly-scheduled payroll. So, in the fall of 1998, we moved into an office. It didn't cost us much to rent a lovely old house in Harvard Square (complete with showers and kitchen, both which were well-appreciated after nights of obsessive coding). In January, we began a payroll system so that people's paychecks would no longer be tied to when our clients paid the invoices. Even if people were earning a bit less than before the structure was imposed, they were happy with the change. We were still earning a large profit each month. And now it was easy to hire people.

On January 1, 1999, ArsDigita had 5 employees. By January 1, 2000, we were up to 57, almost all of whom were developers. We had very little overhead and profits were rising. Around this time, we started looking in earnest for venture capital in order to accelerate growth and to allow ourselves the luxury of taking developers off of paying client projects so they could work full-time on our core product, the ArsDigita Community System (ACS).

By the end of March 2000, we had 110 employees (almost double what we had 3 months previously), 7 offices, healthy profits, plenty of cash in the bank, and $20 million in annual revenue, with the revenue figures still on an upward trend.

Company culture

The amazing thing is that greed and ruthlessness were clearly not necessary for us to rake in the cash. ArsDigita managed to make a large profit giving away free software and training. The revenue came from services and support.

During the growth, we never gave up on our fundamental priciples. These principles that we lived by made the company a fun place to work where the opportunities for learning were enormous. It's easy to sleep at night when you know you are doing good things for your clients and for the world. And when the good things you're doing are also contributing the bottom line, directly or indirectly, you know you've got a winning company.

Some of our principles:

* Honesty. ArsDigita had a strict "no lying to customers" rule. This builds stronger client relationships in the long-run.

* Openness. We were completely open on our web site about our strengths, weaknesses, capabilities, our product features. We gave away all our knowledge about building collaborative web sites in Philip & Alex's Guide to Web Publishing. Even the company structure and salary structure were online. We had no desire or need to fool anybody. If companies think you can deliver something that you can't, either you're wasting everyone's time in the sales cycle or, if you get the contract, you're likely to disappoint the customer.

* Respect. Respect can have two meanings. One category of respect is the shallow "don't say anything negative about anyone else's ideas" type, which has been utilized above all else at the post-VC ArsDigita. But there's also a much deeper type of respect, where people's contributions are recognized and rewarded and there are no obstacles to advancement for capable and motivated people, regardless of their background. The latter type of respect was one of ArsDigita's foundations.

* Open Source. We packaged up all of our reusable, non-client-specific software and distributed the source code and documentation free of charge. The business benefits of open source are numerous enough to fill a large essay, but in brief, open source attracts and motivates employees (most good developers get a thrill out of having other developers use their code world-wide), makes sales easier (it's less risky for a potential client to base their business on your software), results in higher-quality code, enables contributions from the community, and establishes your software as a kind of world-wide standard.

* Teaching. ArsDigita employees had the opportunity to teach students at MIT, Caltech, Berkeley, and other universities that adopted the web applications course pioneered at MIT (6.916). We also TA'd boot camps, 1-3 week intensive training courses for developers who wanted to be able to extend our software, and could teach at the pioneering ArsDigita University, a one-year program that taught brilliant non-computer-scientists all the computer science course material in a typical MIT or Stanford computer science undergraduate education. Employees were encouraged to submit articles to the ArsDigita Systems Journal as well. Besides being an educational and satisfying experience for ArsDigita employees, it also benefits the core business by introducing ArsDigita's software to thousands of developers and helping the company identify potential employees.

* Non-profit work. Our programs included the highly-successful ArsDigita University, the annual ArsDigita Prize which gave awards of $1,000 - $10,000 to deserving and talented young people who built useful web sites, and ArfDigita, a site where animal shelters around the country could upload information about their available pets and then users could search for and adopt the perfect pet. Non-profit work is an excellent marketing technique and it can be useful for motivating and training employees.

Venture capital and new management

In late March 2000, ArsDigita received $38 million in financing, primarily from General Atlantic Partners and Greylock, with a bit thrown in by Bain and Trident Capital as well. In early April, Allen Shaheen, recommended by Greylock, took Philip's place as CEO. Philip was quite happy to let a "professional manager" step in and take over some of his day-to-day management duties, leaving him more time to concentrate on company vision, teaching, engineering, and evangelism.

"ArsDigita University was one of the primary reasons I decided to join ArsDigita Corporation."
- Allen Shaheen, early April 2000, less than a year before deciding to shut down the program

Our level of confidence [in ArsDigita] is high. We're pretty fundamental believers that, 10 years from now, a lot of corporations will look at closed software as a way of the past.
- Chip Hazard, Greylock partner and ArsDigita board member, as quoted in digitalMass, October 23, 2000

ArsDigita is clearly positioning itself to generate annual revenues in the hundreds of millions of dollars very quickly.
- Cameron Laird, SW Expert, December 2000

The future looked incredibly promising.

Life post-capital

Revenue continued to rise to about $25 million in 2000. Is this due to the VCs' genius and vision, or merely a result of the momentum built up before their arrival? Was the subsequent downfall, starting in mid-2000, due solely to the effects of the weaker economy in the United States?

It became increasingly clear to the employees, the customers, and the outside developer community that the VCs and those they had put into management had no idea what they were doing. They discarded the practices that had made ArsDigita a profitable company, destroyed the company culture, and showed their complete technical incompetence by throwing out the ACS, which had been repeately used to solve real business problems, and replacing it with new, partially closed-source software package that was hard to use, had serious performance problems, and met only a small fraction of the business needs that the ACS did.

By the middle of 2000, it was clear that the new management did not understand the mission in front of them. New engineers at our company were routinely complaining about the level of competence and experience we were provided by aD. Communications with the new CEO were unreasonable and frustrating. Contacts at aD went from programmers to project managers to sales VPs. All these people were trying hard, and in some cases doing pretty well - but aD had lost the things that made it unique and successful.
- Josh Stella, former ArsDigita client, April 26, 2001

One of the reasons I joined ArsDigita is its inspiring mission statement, which states that we do not ever lie to customers. The mission statement goes on to say that companies who lie to their customers need to spend time training their employees in the official lies before they can talk to customers, thereby making lying unprofitable.... This week, we were not honest with our customers about the severity of the security holes we discovered in the ACS. We employees even had to sit through a training session so that we would know how to discuss the security holes with our clients. This is a serious contradiction of our mission statement.
- Walt Mankowski, former ArsDigita employee, July 14, 2000

It was sad to see the company slowly being destroyed. It was as if all the life was being sucked out of the company and we had no idea why. Everything that made ArsDigita what it was slowly disappeared until all that was left was just another buzzword-spewing faceless corporation. ACS users moved on, joining in support for projects like OpenACS that still had the spirit of the former ArsDigita... Meanwhile, aD corporate slowly began to shut down programs, including the one that got me to be an ACS user in the first place: ArsDigita Prize. One day a small message appeared that said simply "The prize has been cancelled for 2001."
- Aaron Swartz, past ArsDigita Prize winner, April 24, 2001

It is easy to locate marketing copy on ArsDigita.com. However, it is difficult to get to the information that is actually interesting to anyone building or running an online community, or deciding whether to hire ArsDigita's services.
- anonymous user of arsdigita.com, September 4, 2000

ArsDigita was the company I dreamed of working for when I graduated... In 2000 everything began shifting. ACS 4 seemed an excellent thing, and everybody was excited, but it was never finished and then everything just turned to Java. Nothing against Java, but even I know that you can't sell a product that's not here yet. What happenned to the culture? None of the names and faces we knew posted to the bboards anymore. A few aD faces posted regularly, but mostly we saw a bunch of people only posting questions about the products they should be familiar with, and sporadically. Allan Shaheen only addressed developers twice, in what seemed to be posts typed by his secretary. ArsDigita's website turned to being a brochure more than anything else, which is not bad if you don't forget other things.
- Roberto Mello, computer science student, April 24, 2001

For months now, I've been trying to figure out how aD's management can just ignore the ACS community (and the ACSish market) they way they have. Barely speaking to us; releasing incomplete, unscalable, untested software to us; and finally just abandoning the whole thing midstream. And the only thing I can think of is that they are trying to clear out the community, get rid of the community memory, for when they rollout their closed source, java solution.
- Jerry Asher, developer, April 25, 2001

I heartily agree with you about the unfinished condition of ACS 4.x. We (furfly) have lost at least one prospective client because of it, perhaps more, and the project we're working on now is waaaay behind schedule, much of which is due to our stumbling over ACS 4.x issues. This is not to say that it is unusable, or unfixable, but only that it is unfinished - I don't want to disrespect the hard work that went into it, only the decision to release it before it was ready.
- Janine Sisk, Furfly co-founder, April 25, 2001

ArsDigita, an e-commerce company in the midst of layoffs and a major product overhaul, is bucking the trend of comrades selling open-source software... The change will allow the company to reach profitability by the first quarter of 2002... The company laid off 29 employees in the last week.
- CNET, April 5, 2001.

The founders fight back

Philip Greenspun has never been one to hide his opinions. If he thinks that an idea is stupid, he will bluntly say so. Therefore, Philip had many things to say after April 2000.

Philip's harsh words displeased the other board members (Allen, Chip Hazard from Greylock, Peter Bloom from General Atlantic, and ArsDigita's COO, Ern Blackwelder) so much that they stopped having board meetings after December 2000. Instead, they had "investor meetings" where all the board members except Philip (the chairman) were invited. In March 2001, Peter Bloom sent Philip an email, threatening public humiliation if Philip didn't resign from the board. Philip went to visit his lawyer who reminded him of his rights as majority shareholder.

It was time to regain control of our mis-managed company before it became nothing but a shell. On April 5, Philip Greenspun and Jin Choi, together holding a substantial majority of ArsDigita's shares, had a shareholder vote that gave the founders majority control of the board. They demoted Allen from President and CEO to only President, elected Philip CEO (one spot on the board is reserved for the CEO), promoted Tracy Adams and me (both already Vice Presidents) to Executive Vice President and placed us on the board as well, removing both Allen Shaheen and Ern Blackwelder. The two venture capitalists retained their seats on the board. But we had control.

Yes, it had required drastic measures to ensure that our ideas would be heard, but we were willing to now work hand-in-hand with Allen and the VCs to return ArsDigita to a state of health. But we didn't get the chance. Six days later, on April 11, 2001 Allen Shaheen, Ern Blackwelder, General Atlantic, and Greylock filed a lawsuit against Philip, Tracy, and me:

ALLEN SHAHEEN, ERNEST )
BLACKWELDER, GENERAL ATLANTIC )
PARTNERS 64, L.P., a Delaware )
limited partnership, GREYLOCK )
X LIMITED PARTNERSHIP, a )
Delaware limited partnership )
and ARSDIGITA CORPORATION, )
a Delaware corporation, )
)
Plaintiffs, )
)
v. ) Civil Action No. 18821
)
PHILIP GREENSPUN, EVE A. )
ANDERSSON and TRACY E. ADAMS )
)
Defendants. )

Now, I'm not a lawyer, but I never understood how it could possibly be legal for ArsDigita Corporation to be listed as a plaintiff, since that would have (obviously) been against the wishes of ArsDigita's majority shareholders. And how could it be legal for General Atlantic Partners, Greylock, Allen Shaheen, and Ern Blackwelder take hundreds of thousands of dollars out of ArsDigita's bank account and use it to pay their lawyers to sue ArsDigita's majority shareholders?

The case came very close to going to court but, at the last possible moment, the plaintiffs came to a settlement agreement with Philip. Perhaps the plaintiffs were afraid they would lose the case because the judge had looked at their "evidence" and said that he was uninterested in approximately 90% of it. The plaintiffs' lawyers had wasted their time and, more importantly, ArsDigita's money, preparing a case about Philip Greenspun's personality instead of thinking about whether someone's personality gives one a right to try to illegally seize control of a company.

In the settlement, Philip received $7.6 million. In return, he gave up more than half of his stock, resigned from the board, promised not to attempt to execute any control over the company, and signed a non-disclosure agreement. Before the settlement, Philip owned more than half of ArsDigita's stock. Now that the VCs owned more than half the stock, they had no need to settle with any of the other defendants.

The VCs poured a few more $million into ArsDigita so that it could take a few more gasps of air before collapsing.

VCs push out the founders one by one

The marginalization of Philip Greenspun started taking place quickly as Allen Shaheen discovered that it was difficult to work with someone who wouldn't let him get away with incompetence and dishonesty. Over the course of 2000, more and more responsibility was taken away from Philip and given to so-called professional managers who didn't understand the Web or software development. Instead of firing Philip outright, he was banned from decision-making meetings and was put in charge of the less profitable parts of the company.

In March, 2001, Jin Choi, the company's 2nd-largest shareholder was fired. Jin didn't seem to mind so much because he has never had any tolerance for anyone he deems stupid.

Around the same time, the VCs pushed out Aurelius Prochazka. Aure had built up most of ArsDigita's west coast operations. Most of the clients were there because of Aure. He had trained most of the developers. He had personally built some of the most important modules of the ACS. He could construct an enterprise-quality site himself that would normally take a team of 4 or 5 developers to create.

Aure's crime was that he cared about the company. When he saw ArsDigita executives making poor decisions (or no decisions at all), he wouldn't just sit there and let it happen. He tried to explain to them why other paths may be better to take, and when they ignored him, he wrote to the VCs. Apparently, the VCs didn't care that Aure generated millions of dollars for the company; they were more concerned about not being questioned. They told him that there was "no place for him at ArsDigita" and fired him.

Around this time, I was a member of the product team, led by Richard Buck, an Oracle refugee, recruited by the VCs. Richard Buck's arrival gave everyone great hope (could he revive the software so poorly mismanaged by Michael Yoon that it was way behind schedule and didn't meet any of our clients' needs?). Unfortunately, Richard Buck had a detrimental effect on ArsDigita by:

1. letting all technical decisions be made by a group of programmers that had never managed a client project
2. failing to put his foot down and reject bad architectures that had been selected merely because the technologies were considered fashionable
3. not realizing that, without modules, ACS is a useless product
4. not being able to motivate the product team to work more than 40 hours/week
5. allowing the product team to release untested code that racked up many megabytes of memory leaks after only a few hundred page loads
6. encouraging gossip and rancor within the company (for example, when ArsDigita's marketing department foolishly decided to discontinue one of ArsDigita's free services without notice, Richard sent out an email to the product team speculating about whether Philip was at fault [Philip had nothing to do with it])

Naturally, the VCs and Allen adored Richard Buck.

In my typical, non-political way, I tried to make the best of the situation and worked hard to improve the software and the processes. Richard Buck was very impressed with my work and told multiple people that I could do as much work in one day as most people do in two weeks. Given what I've written about Richard Buck so far, one might be disinclined to lend much credence to his opinion of my work. But my performance reviews have concurred: dedicated, brilliant, super-efficient, "capable of doing anything she sets her mind to," extremely high standards, a "tough but fair" manager. So I was certainly no slacker by anyone's standards.

So why did ArsDigita decide to fire me? Could it have been my candid response during my deposition when I was asked what I thought of Allen Shaheen? (I had replied that the economic downturn couldn't possibly be responsible for all of ArsDigita's ills.) Or might it have been because I was dating Philip Greenspun?

On July 2, 2001, about a week and a half after the settlement was finalized, Allen and I had a little talk about my future at ArsDigita. He "suggested" that I take a leave of absence, after which he -- and he said this with the utmost sincerity -- would try his best to find a good place for me at ArsDigita. I said that was fine; I would spend three months traveling. Allen and I agreed to talk again on October 9 and explore my options together.

Did Allen truly intend to "find a place for me at ArsDigita"? There is evidence to the contrary. On July 16, two weeks later, a terminated employee was FedEx'd an "exit package" by the Human Resources department, as is the standard practice when people in remote offices are terminated. Along with the exit package was a list of Employees Selected For The Termination Program. One may assume that HR did not mean to include that list; somehow they had accidentally put it in the FedEx envelope with the other materials. I do not know the purpose of the list. All I know is that I'm right there on the list ("VP, Operations 27"), which means I had been selected for some termination program on or before July 16. And thus, Allen had no intention of brainstorming with me on October 9. Another case of ArsDigita lying to its employees.

On October 9, Allen was "unable" to meet with me. Same for October 10 and 11. On October 12, I met with Allen in his office. We had a pleasant chat about my vacation, and I saw that he had a copy of my recently-published book Early Adopter VoiceXML (Wrox Press, August 2001) on his shelf.

Allen then handed me a packet of papers, the first of which was a letter dated October 11, 2001:

As you are aware, ArsDigita has decided to realign the organization to better achieve our revised corporate strategy. As a result of this decision, a number of positions will be eliminated. The purpose of this letter is to notify you that your position is among those that will be eliminated and that your employment with ArsDigita will be terminated on October 15, 2001...

Given that I was qualified to do more than half of the jobs at ArsDigita, I know that I did not lose my job because of a re-org. The true reasons are obvious.

Now, What?

What has ArsDigita been doing over the past year?

Lying to the employees and to the press:

Tech startups aren't exactly on the front burner nowadays in the market for initial public offerings, but ArsDigita is steaming ahead anyway with plans to go public next year... The four-year-old Cambridge, Mass. Web software company is hoping to kick off its initial public offering some time in 2002... If it keeps growing and building its customer base, and the IPO market continues its gradual comeback, ArsDigita will be able to take itself public, Shaheen figures.
- Steve Gelsi, CBS.MarketWatch.com, May 24, 2001, a few weeks before Allen Shaheen admitted during his deposition that he didn't think ArsDigita would ever be able to go public

Firing people, spending money:

Beginning Thursday, ArsDigita Corporation has cut a confirmed 24 (from 165) employees and announced plans to close the Berkeley, CA, office. Rumor from several connected inside sources has the final tally climbing to 67 (or higher) over the next few weeks. Late last week the CEO was demoted by the VCs and replaced with Dan Keshian (http://www.greylock.com/team/DanKeshian.asp), one of Greylock's partners, while they wasted another $10 million (reportedly) on the company, presumably to help it stumble along to a fire sale. Those not yet gone are those in engineering, upper management, or are working on billable projects. Those on billable projects will be terminated upon project completion. Standard severance is reportedly 1 month salary plus unused vacation.
- FuckedCompany, October 11, 2001

Building products that nobody wants to use:

When the VCs and Allen Shaheen seized control in April 2000 the company had revenue, profits, a customer list of AOL, HP, MIT, Oracle, Siemens, etc., a $160+ million valuation, $41 million in cash. In about 1.5 years the same company had a product that nobody wanted to use (though it was based on more fashionable tools), no profits, no cash, and few customers. Despite the infusion of more cash and a new CEO, a Greylock partner, in the fall of 2001, the company continued to spiral downward. Now they are dead and RedHat is buying some bones.
- FuckedCompany, February 7, 2002

And trying to make some money for themselves out of the scraps:

ArsDigita, a privately held software company in Cambridge, was shut down on Tuesday. At least some of the company's assets, primarily from its professional services division, will be sold to Red Hat Inc., an open-source Linux software company in Durham, NC, according to sources close to the company... Dan Keshian, a venture partner at Greylock, became ArsDigita's CEO in the second half of last year to lead the company to profitability. Keshian was not present when the shutdown and sale were announced to employees.
- Jeff Miller, Mass High Tech, February 7, 2002

Lessons learned

Over the past 1.5 years, the VCs and their management team have taken a profitable, healthy, interesting company and:

* spent the profits that ArsDigita had saved
* spent all the capital raised
* destroyed an excellent software product
* released a horrible product a year behind schedule
* hired a slew of incompetent managers
* fired the people who made ArsDigita profitable
* repeatedly lied to customers
* repeatedly lied to employees
* repeatedly lied to the press
* repeatedly lied to the outside developer community
* and given themselves big bonuses as a reward

Greylock and General Atlantic Partners have mis-managed ArsDigita into the ground.

What can we learn from this? Be clear about control. Don't assume that people with MBAs know a thing about business, let alone technology. Don't throw out your prime source of revenue before another one is in place. Fashionable programming languages don't equal useful software. Don't lie. And steer clear of General Atlantic Partners and Greylock.
Postscript: Relections, a Year and a Half Later

Written July 2003

Michael Yoon has written a thoughtful essay about ArsDigita. He writes well and makes many good points. I don't fully agree with every one of his conclusions, but this isn't surprising given that -- to state the obvious -- he and I are different people who had different roles and perspectives within the company.

There's no question that Michael was a very important contributor to ArsDigita. Yes, there were things I was frustrated by (the spareness of ACS 4, causing my London team to have to work so many extra hours to recreate the missing functionality; the lateness of ACS Java, making it impossible to demonstrate our product to potential clients). But there are many things I admire about Michael Yoon, things that benefitted ArsDigita greatly: his development of policies and procedures; his careful thinking through of issues during team leader meetings; his writing (his Job Description for Programmers is far more interesting than the average job description); his good client relationships; his incredible diplomacy (it's hard to imagine anyone else calling me a [sacred] cow without giving offense!). I remember a meeting, probably the same one Michael mentions, circa 1999, where Michael did question the wisdom of an IPO, where he questioned what ArsDigita would be like once it was beholden to masters outside the company. He was the only one either brave or smart enough to put forth that question.

I'd like to respond to a few things that Michael wrote and to some of the comments that people posted on my site in the year-and-a-half that have passed since I wrote my ArsDigita story.

Mistakes We Made

As many people have commented, my ArsDigita History is not complete. Not even close. My goal when I wrote the story was to teach a few lessons we had learned the hard way -- the most important being that a company should be very careful about giving control to the wrong people.

The "complete" ArsDigita story is far from black and white (it's hard to come up with things in the real world that are black and white). We (pre-VC ArsDigitans) made many mistakes. As Michael says, we took the VC money and gave up control (though some of that control was abused by the VCs and Allen when they started excluding the chairman from decisions). And yes, there's no doubt that we opened too many offices (with enough tropical fish to feed a small village).

We hired people too quickly. The bootcamps gave useful training to our software developers, but this training was primarily technical. Our new employees would have benefitted from more one-on-one time to teach them our principles, infuse them with company culture, instruct them on our coding standards, show them how we interact with clients.

We also took on new projects too quickly, overworking our developers. Sometimes useful code developed for client projects never made it back into the ACS. Or if the code was put into the ACS, the developers might not have had time to adequately generalize it. Once we had finished a project (or transferred it into "maintenance" mode), there usually wasn't time to do a post-mortem before jumping into the next project.

Management (e.g., me) was inadequate. We were inexperienced and stretched thin. We had a philosophy that a manager should be able to do the job of every person on his or her team. It's a nice idea, but it made it hard to bring in outside people to do the job. It was also hard to move people who were doing very important work on client projects into management. The result: a few people, people who may have been good at many things but who had no formal management training, were in charge of up to 30 direct reports -- in addition to having quite a few other responsibilities.

If we had had a larger, more experienced management team, and if we had spent more time inculturating our new hires, we would have had a stronger foundation, one that would have been difficult for the VCs to shake. [In retrospect, this would have been a very good lesson to have included in my original story.]

Bitter?

My experience at ArsDigita was incredible. I was lucky to work with so many brilliant and fun people -- not only programmers, but people from a variety of disciplines: client services, marketing, sales, HR, project management. I was lucky to be able to help build a product that was used by people around the world. I was lucky to be able to travel, to spend a few months in the London office (what a great team!), to learn from clients who were experts in their fields. Despite the challenges, I was even lucky that we were stretched thin because how else do you get the chance to do management and development and sales and recruiting and marketing and writing and teaching and learning continuously?

Sure, I'm disappointed that a company that had so much going for it was reduced to rubble. I'm disappointed by the egregious actions of the VCs -- and by my own part in relinquishing power to them. But it was a learning experience and, in the whole, it was wonderful.

VCs are Not Necessarily Evil

ArsDigita's venture capitalists in particular showed a profound lack of judgement and morals. Sadly, a large number of people have sent me their own, similar stories about how venture capital was detrimental to their businesses.

But I've also known venture capitalists who do not fit that mold whatsoever.

My advice is this: just be careful. The balance of power between a company and its investors is delicate and the goals of the two parties are not always aligned.

My Team

Different people respond well to different management styles. Yes, some people did leave my team. Others joined it. In the team-leader days, my team was consistently the biggest group in ArsDigita, which isn't necessarily a good thing (see "Management (e.g., me) was inadequate," above).

Some people weren't afraid of giving me feedback, bluntly and to my face (thanks, Caroline!). I made many mistakes. Sometimes I didn't delegate enough and sometimes I delegated too much without proper follow-up. I had unreasonable expectations. I spent too much time with project leaders and not enough time with developers. I didn't give enough feedback. I fought too many fires and didn't do enough fire prevention. But I did learn and improve over time. Michael is probably right that I would have learned more if not for being a Sacred Cow.

I'm proud of my team. They did beautiful work and I'm still using some of their software!

Being the Girlfriend of the Chairman

This was advantageous in one respect: he and I had the same goals. We could work as much as we wanted without annoying the other person. On the other hand, it was an uphill battle to dispel people's assumptions and to earn their trust.

The Ecommerce Module

It's hard to muster the energy to defend something that, by today's standards (and even by the standards of a few years ago), sucks. It was a very early module; when it was written, it had about as many database tables as the rest of the ACS, combined. At that early date, we had few coding conventions and even fewer "support modules."

My implementation of user sessions may have not have been graceful (no one liked usca_p=t), but ACS user sessions didn't exist yet. Henry hadn't yet created his cool spam module, so I had to do my own bulk email system. The file storage module was still just an idea brewing in the back of Aure's mind, so product photos were stored in the file system instead. We had no personalization package (don't know if that package was ever finished) so I made up my own way to personalize product recommendations and do collaborative filtering. I have no excuse for categories/subcategories/subsubcategories.

Given the time and budget constraints (two person-months to implement the whole thing) and the inflexibility of the release date (four client projects waiting on it), it served its purpose. But I won't hold up the ecommerce module as an example of beautiful, scalable, general, modular code.

Conclusions

Interspersed with the more personal and not-generally-applicable things I've written in this postscript are two lessons that may be worth remembering:

* Make sure that your company has a strong and stable foundation with a unified set of values.
* Be careful when taking venture capital. Understand the tradeoffs and protect your company's ability to control decisions whenever possible.


Eve Andersson (eve@eveandersson.com)