Rogue Bloggers and Double Agents

Perhaps you’ve heard of Mini-Microsoft, an anonymous blogger at Microsoft. According to BusinessWeek, he “may be the most notorious blogger on corporate life,” because he frequently attacks Microsoft from within, names people who should be fired, and otherwise airs dirty laundry by the cart-load. Mini claims he’s dishing tough love, telling truths that need to be told before the company can reform itself.

So, at the height of his mainstream-media notoriety, it was interesting to see Mini go ga-ga over Microsoft’s Company Meeting 2005: “I think our customers are going to be delighted silly this coming year! ... Any vestiges of doubt or ennui get blown away once you actually see what we are on the verge of shipping.” (His review of the meeting starts a little ways down on this page, under “Post Company Meeting.”)

I am not doubting Mini’s authenticity, but this turn of events inspires a question: How long will it be until a PR agency orchestrates this type of prodigal-son moment on behalf of one of its clients?

That is, if a company already suffers from bad press and low credibility, who better to be the change agent than a conveniently anonymous rogue blogger? Like Mini, the rogue blogger would have enough inside dirt to make some news, getting the mainstream press to anoint him or her as a tantalizingly credible source. Some of that dirt might even be things the company wants to get out but can’t officially say—for example, that a recent executive “resignation” was really a firing for poor performance (message to shareholders: we know that needed fixing).

Having built credibility with a mix of juicy tidbits and question-authority attitude, the rogue blogger could then, at a critical moment, lock onto company messaging: “I have seen the light!” Or the rogue blogger could subtly change perspective over time, grudgingly giving ground to the relentless progress the company is making.

Of course, this rogue blogger would be a double agent, operating in the darker parts of the ethical gray zone. Most legitimate companies and PR agencies will not go there.

But for the inevitable ones who do go there, will it work? I hope not, but we might never know if it does.

Google Playing Clickstream Catch-Up

Lately, there’s been a lot of informed speculation about Google’s forays into WiFi access points and dark fiber. Last night, Scoble weighed in thusly:

So, why do I fear Google’s wifi? Well, if you own the last few yards in between people and the Internet you can really learn a lot. You can watch everything those people click on, what pages they visit, what browsers they use, how often they turn on Skype, and a lot of other stuff.

Isn’t it the case that Microsoft already has this data on millions of MSN subcribers that use the MSN network to access the Internet? Same for AOL. And depending on its ISP agreement with SBC, perhaps Yahoo too.

The point: When it comes to having access to full user clickstreams (not just those at google.com), Google is actually playing catch-up. So if you fear Google in that regard, you should have plenty of fear to spread around.

Yet elsewhere in the same post, Scoble seems impressed that MSN search “learns from usage patterns.” No indication of whether these are usage patterns just at MSN search or from the larger MSN clickstream pool, but either way it hints at this coin’s flipside: Clickstream data can be helpful not just to companies but to Web users. It can be used in the aggregate for improvements to the audience-wide user experience, or it can be used at the individual level, providing differentiated, personalized experiences.

Today, companies mostly use clickstreams at the aggregate level, because profiling and personalizing via the clickstream requires much more technical effort; it also requires finding an elusive sweet spot where user benefit and trust outweigh privacy risks. But a site that finds that sweet spot, as arguably Amazon.com has done, has enormous advantages.

Now let’s up the ante and talk about the advantages of finding that sweet spot not just at the site level (what you do at, say, Amazon.com) but Web-wide (what you do everywhere): Who will be best at collecting and using people’s full clickstreams in a way that everyone wins enough to participate?

Because of the technical and privacy challenges, the contestants are barely out of the starting blocks, which makes Google’s game of catch-up a potential game of leapfrog. I feel a Google Don’t Be Evil™ opportunity coming on.

Harvesting Power from Human Motion: Small to Large Scale?

You may have read about the backpack or combat boot that generates power from its wearer’s walking. These are small-scale examples of “power harvesting” from human motion. They work by clever use of devices and/or materials that convert motion to power. The power generated is small but potentially enough for a mobile phone or other personal electronics. And if we get a little futuristic, the power could be used for wearable computers or smart clothing.

For me, these technologies invite the question of scaling-up. For example, can a highway overpass be instrumented to harvest the vibrational motion from thousands of heavy, fast-moving cars and trucks? Can the overpass’s surface be adapted to harvest power directly from contact with those cars and trucks’ motion?

How would this work? A core technique of small-scale power harvesting is the use of piezoelectric materials. When these materials bend or stretch, they create power. Instead of having a small wafer of piezoelectric material in a shoe, how about thousands of wafers arrayed throughout an overpass? What if they were connected to the surface so that they bent and stretched with the forces of multi-ton vehicles constantly zipping by?

This isn’t my area of expertise, so I don’t know the answers. But the questions illustrate the more general opportunity of thinking bigger about harvesting (some might say recycling) energy that humans are already expending. From a societal point of view, it would be a welcome area for innovation.

Why You Should Read the Feed

[Update 3/13/2012: This post still exists for historical reasons. Bloglines is gone, and much of what people used RSS for is now being done with Twitter. However, I still find RSS useful. I currently use Google Reader for RSS reading.]

If you have not discovered Web feeds (aka RSS feeds), you should. To illustrate why, I’ll cut straight to an example.

I use a “feed reader” Web site called Bloglines. I have told Bloglines which Web sites I visit regularly. Bloglines now automatically checks them, notifying me when anything new appears.

Why does this matter? I’ve saved time and hassle in two ways: (1) I don’t need to check sites only to find that nothing has changed. (2) When things are new, I can rapidly scroll through all the changes at all the sites, skimming for the stuff I want to read.

And this is not just about blogs. Big publishing sites like the BBC, CNET, CNN, and the New York Times have feeds too. Most sites you like probably have a feed already; those that don’t will have feeds soon.

If this does not seem significant to you, try it. I know very few people who have gone back to the old way.

How You Can Get In on the Action

There are many ways to read feeds. I recommend you start with a Web-based feed reader like Bloglines or (for MyYahoo users) the RSS-reading features in MyYahoo.

A Few Notes to Help You Get Started

Just like a site has a URL for its home page, a site has a different URL for its feed. The difference is, the feed page is designed to be read automatically by systems like Bloglines, not by humans.

When you ask a service like Bloglines to track a feed, you provide the feed’s URL. You are then “subscribed” to that feed. But don’t worry: It’s not like a subscription where the site knows who you are; it’s anonymous. Also, almost all feeds are free, as are services like Bloglines.

When you are at a site, look for icons like this:

  

Although these three images are just examples and thus are not clickable, similar images usually link to feeds. My clickable feed image is on the right side of every page on this site. If you don’t see it, scroll up near the top. The image has a link associated with it. Just copy the link into your feed reader. Or, if you have a feed-discovery feature associated with your browser (here’s an example for Bloglines), it should find my feed when it is applied to any of my blog pages.

Sometimes a page will have multiple feeds in different formats with names like RSS and Atom, perhaps with different version numbers. If you use a major feed-reading service or product, any of them should work.

Companies as Products

With the tech sector heating up again, it’s a good time for venture-funded entrepreneurs to think not only about their companies’ products but also their companies as products: Unlike your actual product, your company-as-product operates in a capital marketplace of investments, IPOs, and acquisitions. Managing this marketplace matters as much as anything else you do, both in normal times and especially when financial markets get exuberant.

A scenario to consider:

Advances in technology have led to a small but fast-growing market with huge potential. Jack’s start-up has the best product and is executing well. Jill’s start-up is a me-too clone of Jack’s, created later and still playing catch-up.

Although her product is less-developed, and her business is well behind Jack’s, Jill is far more aggressive than Jack in the company-as-product marketplace. The IPO market is hot, and she gets there first, selling the concept of her company—which is a clone of the concept for Jack’s company—to the public markets.

As Jack watches from the sidelines, he knows that his business is the one that actually has the substance behind Jill’s story. But when Jill’s company raises $100 million in cash from the IPO, Jack has a problem. Jill now has the resources to shore-up her product and, in the meantime, overwhelm Jack’s marketing and sales efforts with five times the feet on the street. In addition, Jill is bathing in the free publicity of being the poster child of this hot new market.

Now the investment bankers are knocking on Jack’s door, saying he could be like Jill, having a big IPO. But having gotten out first, Jill can use her resources to damage Jack’s IPO. Ouch…she just used some of her IPO cash and inflated stock to buy the one company Jack was actually worried about, a bunch of rocket-science types whose next-generation technology is disturbingly good. Now that company’s technology is in the hands of Jill and her expanding sales and marketing machine. Between this move and her PR from the IPO, she is defining the playing field.

As Jack prepares his IPO, he now is on the defensive for being the me-too player.

Is this fair? If you said no, you missed the point. Jill leapfrogged Jack because she seized an opportunity in the larger system, using the company-as-product market to vault her forward in the actual-product marketplace. It’s fair as long as you realize that the game is played at both levels, especially in times when financial markets are exuberant and thus less discerning than they should be.

The lesson: Although many tech entrepreneurs would like to just build great products, building a great company can also require playing like Jill, or at least actively blocking Jill-like competitors. That requires thinking about companies as products.

Consumer Apps for Real-Time Biofeedback

Look for some interesting companies and products to come from the commodization of technologies related to real-time biofeedback. Let’s illustrate with an example.

Yale Professor Robert Grober has developed a device that lets you hear your golf swing. Ingredients: an instrumented golf club, with sensors that wirelessly transmit data to a receiver, which in turn converts the golf club’s telemetry data into an audio soundscape. Different swing parameters contribute to the soundscape, allowing the golfer to “visualize”—through sound—his/her swing as it happens.

The “as it happens” part is critical, because the core skill in golf is developing muscle memory to swing and putt correctly, consistently. Getting the feedback of what you’re doing while you’re doing it is much better than, say, reviewing a video of your swing.

Grober’s company, Sonic Golf, has a Web site, which explains why this product is becoming real now:

Sonic Golf products are enabled by the convergence of four technologies, each one of which is driven by an existing industrial base: 1) silicon based sensors, including MEMs accelerometers (automotive industry) and gyroscopes; 2) low power hand held electronics, including micro-power microprocessors and micro-power A/D converters (PDA and cell phone industry); 3) digital music synthesis (sound cards, computer multimedia/gaming applications); and 4) the IEEE 802.15.4 communications protocol (Zigbee) for extremely low power (mW), relatively low bandwidth (100s kB/sec) wireless communications.

In other words, the key enabling technologies are available off-the-shelf, at reasonable cost. Besides the obvious variations on Sonic Golf for other sports, how else might these technologies be applied?

Let’s start with a system for improving posture. If you are unaware of this market, a Froogle search for “posture” yields 19,000 listings for product prices at retailers. As with a golf swing, learning to sit and stand correctly is well-suited to biofeedback using real-time sound. However, the market is no longer just golfers; it’s everybody. (Cue ominous voiceover: “Bad posture not only affects how others perceive you. It hurts your health.”)

What about a smart chef’s knife, which uses sound to tune the user’s skills at slicing, dicing, chopping, mincing, and such? How about a bicycle that teaches kids to ride it?

If we assume that component costs drop so low that almost anything can be “sonified,” what happens? Smart chopsticks, anyone?

High-Definition Lettuce

In terms of lettuce, I grew up a member of the Iceberg Generation. Our lettuce was a greener shade of pale, but what it lacked in taste it made-up in crispness. On sandwiches, that crispness broke the mouthfeel monotony of Oscar Meyer mystery meat on Wonder bread. But if one were to experience iceberg lettuce as the featured attraction, such as in a “salad,” it was just roughage.

In the past decade, the U.S. population has begun an ascent up the lettuce hierarchy of needs. Where once our ancestors foraged only for iceberg at the local supermarket, now they return with a plenitude of choices: hail-caesar romaine, post-Popeye spinach, the weedeater’s frisee, and other varieties with code-names like “arugula” and “butterhead.”

Living in Northern California, epicenter of lettuce actualization, I bring a report from the future. One of the best restaurant dishes I ever had was a recent salad. It was a small head of organic butter lettuce and a simple mustard-vinaigrette dressing. That’s it. No croutons, no crumbled feta, no nothing.

If iceberg lettuce was like black and white television, and typical Northern California organic greens are like color TV, this was high-definition lettuce. Beyond that phrase, I won’t try to relate the experience. My only point is to say that the trend toward better lettuce continues, not just in variety but in taste. So if you see a suspiciously spartan lettuce dish at a high-quality restaurant, try it.

[For the record, my butter-lettuce epiphany dish was at a San Francisco restaurant called La Suite, corner of Embarcadero and Brannan, now defunct.]

It’s Like Golf But with a Shotgun

I’d like to thank Eli Marcus, who in a recent conversation enlightened me to the existence of “sporting clays.” When I heard, “It’s like golf but with a shotgun,” I had to know more.

Turns out his description was not an embellishment. Wikipedia’s Sporting Clays article tell us: “Sporting Clays is a clay pigeon shooting sport. Often described as golf with a shotgun, the sport differs from skeet and trap shooting in that it involves shooting clays at various locations which are launched at different velocities and angles.”

Over to the Sporting Clays Magazine FAQ for a little extra color:

With variations in trap position, trap speed, shooting position, and flight paths of different types of clay pigeons, targets can come through the trees, from under your feet, straight down, over your head, quartering, going away, left to right, right to left, and in any path a real bird might choose. The key words are unpredictable, variable, and sometimes bordering on impossible.

If bass fishing can be a televised sport, and my cable system has 500+ channels, I’m wondering what kind of market failure is responsible for the lack of sporting clays coverage on TV.

SRI Media Futures Program

SRI International is one of the world’s largest independent research organizations. When I was there in the early to mid-1990s, you could walk across the campus and pass groups working on artificial intelligence, economic development programs for post-Soviet states, improvements to the public-education system, an easy-clean oven surface, military communications networks, and cancer drugs. If you’re into inventions and innovations, definitely check out the SRI timeline.

At SRI, if you could find government or corporate entities to fund your research, you could largely do whatever you wanted. In 1991, I and two colleagues, Ed Christie and Paul Di Senso, decided that the world needed a research program about the future of digital media. This was before DVDs, DirecTV, digital cable, Tivo, and the rise of the commercial Internet. The outlines of these technologies were becoming visible on the horizon, so we bet that many companies would want to know what was going to be real when.

We somehow wangled internal seed money to get the program started and thus was born the Media Futures Program. It was a “multiclient” research program—that is, many client companies would each kick-in a yearly fee and then receive back the sum-total of research. In essence, it was a market-research business, albeit with an SRI twist. We combined quantitative survey research about consumer demand, an engineering view of technology feasibility and costs, and business analysis of strategic, competitive, and regulatory factors. Put another way, we did original research on what people wanted, what technology could deliver at what cost, and which companies were likely winners.

I was one of the two research leaders. That meant that I managed half the projects, presented results to clients, and participated in much of the sales effort. I also had the fun of getting hands-on with a lot of different research tasks: evaluating the early HDTV prototypes, creating a Monte Carlo simulation of factors involving video-on-demand uptake, and authoring a lot of documents about audio and video compression (technologies that would enable MP3 players, digital-cable boxes, and many other devices).

[Now that Google has made the Usenet archives searchable back to the 1980s, the technically inclined can see what we were talking about when today’s common digital-video technologies were being hashed out in 1992-1994.]

As the idea of “digital convergence” heated up—thank you, Wired magazine, for casting cable and telco CEOs as rock stars—our little research program got big. Apple, AT&T, Disney, Microsoft, Philips, Sony, and most of the “Baby Bells” were clients, as were dozens of others U.S., European, and Asian players.

Ironically, we got most of our notoriety from making negative assessments. In 1992-1993, interactive TV and video-on-demand were the big ideas, which everyone was pushing hard. Our research indicated that the proponents were living in a fantasy world in terms of their ability to deliver such services at anywhere near economical costs.

We were not trying to be contrarians. In fact, we said that interactive television and video-on-demand would make sense someday, but a set of intermediate products and services would emerge before. The most important of those were DVDs and broadcast digital video, which we called correctly as the big winners of the late 1990s. But because we questioned the near-term commercial viability of interactive television and video-on-demand, that was “news.”

In market research, being known as negative is not usually a ticket to success, but we called things as we saw them. And for those willing to read past the headlines, we highlighted plenty of opportunities in what were then seen as less-sexy areas. Ultimately, however, the final verdict came when the clients, whose visionary CEOs we were undercutting, kept renewing their participation in the program.

We also did consulting projects for specific companies. My favorite was a job we did for Technicolor’s CEO and executive team. We helped them decide to get into the manufacturing and distribution of prerecorded CDs and DVDs. This was well before the first DVD was released, yet they went with our call that DVD was going to be big. It was not a slam dunk given the failure of two previous formats, analog videodisc and VideoCD. However, Technicolor went on to become a dominant player in DVD manufacturing, producing more than a billion DVDs per year.

It was a great time and a great team: the three founders plus honorary founder Michael Gold (abducted from SRI’s engineering labs) and exceptional colleagues Adam Gross, Dave Rader, and Joyce Thom.

In 1994, I and two different SRI researchers had started another program, iVALs, which focused entirely on the Internet as a medium for measuring and understanding what consumers wanted and, ultimately, using that knowledge to personalize media back to them. That meant I did progressively less Media Futures until I and the iVALs colleagues left SRI at the end of 1995 to form Personify.

But I’m pleased to report Media Futures kept going strong, continuing to scan the horizon of new digital-media markets as they emerged. Its current incarnation is Digital Futures, part of the SRI spin-off SRI Consulting Business Intelligence.

Personify Retrospective

Personify started before the dot-com boom and outlived most of those claimed by the bust. Yet its six-year run included an intense dosage of the good, bad, and ugly of that era.

I led the original founders in January 1996, and I stayed for the whole thing, through August 2002. My official title was CTO, but at one time or another I also sold our products, served on the board, deployed software in the field, was interim CEO, hacked out random fixes for bad customer data, cut partner deals, and did whatever else needed doing. Together, these activities were the best input possible for my most frequent role: something like a VP Products, working with a team of exceptional people in defining, designing, and delivering products that were among the most advanced of their type.

The Technology

Personify was one of the early players in enterprise Web analytics. For the sake of a definition, “Web analytics” means software that helps measure and analyze a Web site for the purpose of making it more effective. “Enterprise” means the software does enough—and costs enough—that buying it is a corporate decision, not something a Webmaster expenses on a credit card. At this level, our initial competitors were start-up companies Accrue, Andromedia, and NetGenesis.

The name Personify aptly summarized the vision: (1) enable a Web business to analyze and personify its audience, to understand what people wanted not as a whole but as segments and even individuals; (2) then use the learnings to make better business decisions and, ultimately, to personalize what people see.

Today’s analogy is when Amazon.com learns about you and then adapts your experience accordingly. For the late 1990s, it was a big vision to fulfill as a software provider. Your solution needed to work across a wide variety of sites, which didn’t have Amazon.com-like resources.

Although the competition could spin similar visions, their technical foundations were Web-site reporting tools. These tools were engineered for large-scale reporting on site traffic (page counts, session durations, bytes downloaded, and such) but ill-suited for what came to be known as CRM (customer relationship management), where the center of the data universe is customers (or, in this case, Web-site users). We were the only ones taking a CRM approach to Web data. We had privacy-protected user profiles, data-mined segmentations, and on-the-fly analysis of behavioral, purchasing, and demographic attributes. For 1998 and many years after, it was a unique feature set.

The Value

With Personify, a marketer could discover (via data mining) a set of behavioral segments, evaluate each segment’s relative value to the business, and then explore what made each segment click: Which outbound advertising campaigns are delivering my “Core Buyers” segment? Are my on-site content investments paying-off for the “Researchers” segment? Is my latest wave of promotion just drawing the low-value “Price Predators” segment?

From day one, Personify customers were able to explore and answer these questions instantly. Later, we added a module that integrated with email and Web-site-serving systems, so Personify data could drive personalization—for example, “Target this special email only to ‘Core Buyers’ who have not visited in three months.”

In sum, we were in the business of enabling “difference marketing,” where a business adapts itself to the key differences among customers, rather than making every decision based on Joe or Jane Average. In traditional direct marketing—the businesses behind the catalogs and credit-card solicitations you get in the mail—difference marketing had already proven the superior approach. But because difference marketing was fueled by data, and Internet data was like a new kind of jet fuel, there was an opportunity to do even better.

The Ascent

We were seed-funded in spring 1996 by U.S. Venture Partners. We spent the rest of the year as a garage shop called Affinicast, proving the basic opportunity and technology. Having passed that test, we spent 1997 building the product and the core of a company to market, sell, and support that product. We also changed the name to Personify.

In February 1998, we launched the company and previewed the product at Internet Showcase, where we won a “Best in Showcase” award.

The first official release was in June 1998, by which time we already had paying, referenceable customers and a fan club of analysts and industry types.

1999 was pure growth, as we scrambled to hire sales and support personnel to chase the demand. To reinforce the core, we acquired a 30-person professional-services company, Anubis, which specialized in data-warehousing technology. It was a great injection of talented, dedicated people.

The Peak

By early 2000, Personify was widely regarded as a hot commodity. We had just raised private money at a $500 million valuation. We had turned away multiple acquisition approaches and were preparing an IPO filing. Having watched companies in our competitive set go public to billion-dollar market caps, our investors were hungry for a home run, as were the employees, all of whom already knew somebody who got dot-com rich.

By the peculiar financial standards of the day, we looked good. We had already doubled our 1999 revenue in the first quarter of 2000 alone. For that quarter, we reported $1.25 million in revenue, using a conservative accounting practice that only recognized the value of a contract over time. When you combined the fast growth with the conservative accounting’s trailing indicator, it was obvious that we were heading for something like $10 million in new business that year.

The other standout for Personify was the customer list: Our IPO filing listed more than 50 customers, many of whom were household names. Given the media’s interests at the time, we probably got more buzz out of serving Drugstore.com, eToys, Petopia, and DrKoop.com than we did Barnes & Noble, Volvo, J. Crew, L.L. Bean, or REI. A lot of these companies were dabblers in what we enabled—sophisticated clickstream data mining, analytics, and targeting—but they paid real money nonetheless.

We filed to go public in May 2000, two months after the NASDAQ peaked. At the time, no one knew it had peaked. When we filed, the NASDAQ had bounced back near its position at the start of 2000. But the markets were jittery, and the Internet IPO window was in the process of closing—for years, it turned out.

The Fall

Despite the non-IPO letdown, we ended up exceeding $11 million in new business for 2000 anyway, acquiring customers such as American Century, Bose, Continental Airlines, the New York Times, Nieman Marcus, and Volkswagen. These names reflected the fact that in early 2000 we stopped pursuing the dot-coms, raised our prices, and aimed squarely at corporate America. As a result, the average deal size climbed above $250,000. But despite doing bigger deals, with bigger customers, and capturing far more revenue, the Personify of early 2001 was no longer worth $500 million, or anything close to it.

This irony was not unique to Personify. As the Internet boom hit its limit, the super-inflated valuations of Internet companies gave way. To put it in airline-safety speak, a loss of cabin pressure occurred, but the oxygen masks didn’t drop. Valuations had gotten so disconnected from business fundamentals that when the fall came, the decompression was massive, overwhelming whatever other factors might normally contribute to a business’s value. For a lot of Internet companies, the plane outright crashed; for those like Personify, it leveled-off just above the treetops.

In 2001 and 2002, our (former) dot-com customers were not the only ones suffering. More important for us, big corporations were hurting. They were putting enterprise software purchases on indefinite hold, whether from Internet upstarts like us or from the likes of Oracle and Siebel. Everybody’s numbers were way down. The real question was who had the cash to survive until things got better, whenever that might be.

It didn’t get better soon. Although we had enough substance to raise some additional private money in the early part of the bust era, it was a far cry from an IPO’s proceeds. So we did as virtually all others and managed the business’s costs against the falling revenues. That meant multiple waves of layoffs in 2001 and 2002.

By 2002 Personify reached the point where our products would win deals and then the company’s financial condition would un-win them. A prospect would want to buy but be concerned about whether we’d be around next year. Assuming that other prospects would have the same concern, each would walk away, putting us on the wrong side of a self-fulfilling prophecy: No one buys because they think no one else will buy.

We had not helped our cause when, in 2001, we absorbed a start-up that wasn’t viable alone but had some cash. It also had some debt. The deal gave us the appearance of being a consolidator in a consolidating market, but it was too little, too late. The cash was nice, but the debt was a time bomb, with a fuse set for mid-2002.

The Final Cut

By mid-2002, we were running out of time and alternatives. In an effort to circle the wagons, we agreed to be acquired by the only other then-independent company from our original competitive set, Accrue. As of late July 2002, we had a signed term sheet, the final documentation was prepared, and all Personify employees and customers were going to be carried forward.

Then a bad thing happened, symptomatic of a time when bad things happened a lot: A creditor we had inherited from the 2001 cash/debt deal was now in bankruptcy. Too many of his boom-financing deals had gone bad. So, despite having signed-off on the the Accrue deal’s term sheet, the creditor scuttled the deal on the day it was to be finalized. Instead of signing, he made a play for our remaining cash by effectively killing the company.

This event took down Personify, which paid-out what was left to creditors and closed in August 2002. Accrue later ran out of gas as well. After a bankruptcy filing, its assets were resurrected as a different company.

Meanwhile, the next generation of enterprise Web analytics players had emerged in the late 1990s. They were either new companies (founded a few years after the first generation) or smaller, lower-end providers that were graduating to serve the enterprise. The new generation’s offerings had less functionality but were packaged as easy-to-buy application services. For the bust era, cheap and easy was the right value proposition. Although corporations could not make six-figure, enterprise-software purchases, they could find a few thousand dollars a month out of a marketing budget. And because they could pay monthly, the “going concern” issue didn’t matter as much.

The best of that new generation (companies like Omniture, Coremetrics, and WebSideStory) would later fill-out their features and, with the economic recovery, raise prices to resume where the first generation left off.

Clever versus Stupid

“It’s such a fine line between clever and stupid,” observed David St. Hubbins in the movie This Is Spinal Tap. Let’s explore that line.

In retrospect, Personify should have been acquired. We declined multiple opportunities. But at the time, an acquisition was not so obvious. Here is why.

When we did our private financing at a $500 million valuation, Personify was substantially undervalued against companies with similar financials that had already gone public. Yes, a huge portion of everybody’s valuation was hot air, but those who went public transformed their hot air into piles of cash, maintaining independence in the process. It was an appealing path. Looking back, it was arguably the only path for an Internet enterprise-software company like Personify to have a shot at remaining independent through the bust, taking the boom’s loose money and squirreling it away for the long winter.

And it wasn’t just about the money. Personify had taken the big-vision approach in a time of great change. It was a rare chance to build a company of consequence. A quick way to kill that chance would have been to pull the ripcord on an acquisition, thereby becoming a cog in someone else’s machine.

Of course, the IPO path was only open for a limited time. We narrowly ended up on the wrong side of the line, that fine line. And thus history tells us that we should have sold out to an acquirer at the top, notching a $500 million win for the employees and investors. But this lesson amounts to little more than “buy low, sell high.” (Thanks for the tip.)

Endings as Beginnings

The good news is that the dozens of people who comprised the Personify core have long since gone on to great things. (Because Personify had so many heroes, I’ve avoided naming-dropping a few at the expense of others. They are all deserving. I’ve also omitted discussion of a few villains because life is too short.)

Of the first 80 employees, 15 have since been founders of companies, often together in pairs or threes. For my part, I went on to start another company, which was acquired by CNET in late 2004. Although it was a different kind of venture, it was informed by many Personify lessons.

As for the technology, last I heard it was still running at several sites. That included one doing the whole vision—analytics integrated with personalization—on a terabyte-scale Personify database. My heart warms at the thought.

And on that note, I say this: If you do a start-up, you’d better enjoy what you work on and who you work with, because a big payoff may come or not—and statistically speaking, you’ll usually be looking at not. Even on Personify’s worst days, there was always the quality of the core people—of whom there were many, in all parts of the company—who thrived on doing what had not been done before and doing it well. If instead we were slogging something just to make a buck, the game for most of us would have been lost before the outcome. I’m glad I was a part of it.

SRI iVALS Program

In 1994 and 1995, I led a research program at SRI International called iVALS (Internet Values and Lifestyles). It was the dawn of the commercial Internet, and we were among the first to measure, analyze, and segment the Internet audience.

I had been at SRI since 1990, mostly in another research program, the Media Futures Program, which was about the commercialization of digital-media technologies. A regular theme of the Media Futures Program was that when a networked, interactive form of digital media arrived, the game would change. Not only could the consumer actively control what he or she saw, but an electronic newspaper or catalog could adapt itself to the consumer’s interests by learning from the consumer’s behavior. Compared to broadcast media or print, it would be more like a two-way conversation.

In the early 1990s, interactive television (ITV) was supposedly going to be the medium that made all this real. We studied ITV in enough depth to realize that the technologies of the day were too costly to use at large scale, no matter what the visionary CEOs proclaimed.

But something else was happening that most people didn’t notice, whereas we had a front-row seat: the dawn of the commercial Internet. SRI had been on the receiving end of the first Internet packets in 1969 and since then had a deep Internet culture. Thus, we were among the first to see and use new Internet technologies of the early 1990s, such as Gopher, WAIS, and this thing called the World Wide Web.

When the Web appeared, we realized that it could economically scale to reach millions of people in the near term. People already had computers, and it required far less network bandwidth or server technology than ITV. The Web could be the networked, interactive medium to change the game.

At the time, the only other contenders were commercial online services like AOL, CompuServe, and Prodigy. However, they were more about email and message boards than enabling anything like Web sites. In addition, they were closed, fee-based networks that were not compatible with each other. In contrast, the Web was free and open, the two key ingredients for its hypergrowth.

I and two colleagues at SRI, Adam Gross and Bruce MacEvoy, bet that the Web would matter. We created iVALS to explore what it meant to measure and analyze the most data-rich medium ever. I led the program but special credit goes to Adam for being the earliest evangelist of the Internet’s importance.

By a coincidence of history, SRI was the birthplace of VALS (Values and Lifestyles), the best-known system for analyzing the U.S. population psychographically—that is, by people’s beliefs and values. Bruce was a consumer psychologist and statistician who worked on VALS2, a revamped system designed to analyze and predict consumer behavior. His consumer-research angle plus Adam’s and my digital-media angle equaled iVALS.

Our first project, in 1994, was the creation of a Web site where people could take a short questionnaire and receive their VALS2 type, including a dynamically generated description. It was one of the earliest database-driven Web applications with adaptive content. More than a decade later, a descendent of the site still exists, although I see they’ve changed the names of some of the types and removed the dynamically generated descriptions.

With thousands of people taking the questionnaire and getting typed, we had one of the first images of the early Web audience. 50% of the respondents fell into a segment of highly-educated professionals, which comprises only 10% of the U.S. population. It underlined the point that the 1994-1995 Web audience still largely reflected the Internet’s academic and corporate-research heritage.

The specialized nature of that era’s Internet audience called for a specialized psychographic system, which led in 1995 to the iVALS typology, an Internet-specific set of segments. The purpose was to capture factors inherent to the new, technical nature of the medium. For example, to understand online behavior, it didn’t matter if a person was highly sociable if that person didn’t know how to use a chat application (this was before the days of chat being built into Web pages or user-friendly instant-messenger apps).

The point of the exercise was to model why people did, or did not, do things online, using a combination of psychographic and technical-capability factors. Companies were just starting to ask how the new media was different from the old, and we were among the first with answers. Being at the edge of that frontier was a great experience.

As we did our work, we would run into people from the relatively few Web companies of the day, like Yahoo, Excite, and HotWired. Everybody loved what we were doing, but they all said the same thing: “You need to start a company.”

We did, and that company became Personify.

ExactChoice: The Two-Minute History

ExactChoice was a company I co-founded with Howard Burrows, a friend and colleague from the early Personify days. At ExactChoice, we developed Web-based applications that did analytics and mining of complex product data. The company started in November 2002 and was acquired by CNET in December 2004.

Our best-known app was the ExactChoice Recommender. Using various forms of personalization, it helped people quickly identify, learn about, and buy the best computer for their needs. First among its technical innovations was that it recommended not just products but also product configurations.

For perspective, a typical recommender represented a Dell computer as just that: a single computer, usually the default configuration. ExactChoice represented the same computer as 5,000 different possible configurations, each with its own price and features. If you could configure it at the Dell site, we could represent it. Not surprisingly, when consumers told the system what they wanted in a computer, the results were better when considering 5,000 times the possibilities. (This was not Dell-specific. We covered all the major computer manufacturers, each of which offered both packaged and customizable models.)

ExactChoice built other apps on the same technical foundation. We had one that let call-center reps instantly identify their best answer to a competitive challenge, complete with hints for how to counter-pitch. We also had an app for executives and marketers, who could track the relative feature/price-competitiveness of their offerings versus those by competitors; we even threw-in a “what if” feature for running scenarios with different components and pricing.

From ExactChoice’s inception to acquisition, we did it all with two people, in 25 months, with a total capital outlay of less than $15,000. By design, ExactChoice was owned and operated by the founders, based in a virtual office. (I once had to fill-in a form that required the square footage of our office, so I provided the dimensions of our P.O. box slot.)

In many respects, it was an experiment to see what we’d do with total control and accountability, not to mention the lack of a safety net. If it failed, it was our money at stake—not just the dollars-in but also the bigger cost of not taking a normal salary. When the term “burn rate” gets personal, it’s a focusing thing.

At the end of the day, ExactChoice succeeded. We had enormous freedom to do the right thing, designing and implementing not just the product but the business with continuous user feedback. When you are constantly testing and evolving toward what people will use or buy, creating value is inevitable. That said, the process is neither easy nor predictable, which is why it is rarely done end-to-end.

Another key factor was a good call we made in 2002: that online shopping was going to rebound and be increasingly ready for a new wave of innovation. Never underestimate good timing.

Because ExactChoice had a happy ending, there’s a tendency to forget the unglamorous parts along the way: trudging over to the P.O. box to deal with the latest small-business tax forms, cold-calling prospects, and other stuff that magically happens elsewhere in bigger companies. But everything has trade-offs, and in this case, they were well worth it.

[Update, 12/10/2007: After running the original ExactChoice Recommender for two years, CNET in 2007 created a set of new, ExactChoice-inspired recommenders in several categories. Details here.]

Catching Up

I just wanted to apologize for my lack of postings over the past 15,000 years—no good reason, just the usual excuses: work, plague, Dark Ages, etc.

For the sake of continuity, I’ve included an archaeological photo of my last post, “Untitled” (from 12,789 BC), as seen in the recent Discovery Channel documentary, Instinct for Expression: Lost Blogs of Prehistory.

Although I miss the spit and grit of the cave-as-canvas, I appreciate productivity-enhancing technologies like the Internet and written language.

So going forward, I expect a big jump in my number of postings per geologic epoch, starting now.