Gillette is a software company.

Hardware companies make money on infrequent higher margin items. Usually they involve productions systems built with large capital investment and are built to run as fast and error free as possible.

The best software companies build for design speed (as opposed to production/operational speed) and identification of errors. These errors or deficencies are then fixed or improved as quickly as possible and released back to the user base. They either run on a continuous subscription based model or charge for major updates/releases.

There is fertile ground to talk more about the distinctions, but most importantly, I want to draw out how the two companies differ regarding innovation (and how that manifests in increased revenue). The basic model of selling hardware — houses, appliances, cars, etc, involves a longer production, buying, and “service” value chain. It’s capital intensive, which induces a sort of efficiency intertia: if it ain’t broke, don’t fixt it.

Selling software — cereal (yes), digital applications, digital cable, etc, may involve intensive capital investment, but selling it means rapid prototyping and constantly introducing new value. Consumers really “consume” software. Its value is in its use rather than its possession. The mantra for selling software is rapid improvement, iteration, and change. The inertia is stuck on “different” — it’s the ball in motion to the hardware’s ball at rest.

The way to apply this distinction is to try to define your company — put it in one of the boxes. Then imagine if you flipped — that as a car manufacturer you became a software company instead of a hardware company. How would you operate differently? What would be different about the organization? How would you make money? Where and how would you invest R&D budgets? Marketing and sales?

Gillette really turned a hardware business model into a software model. They are wildly successful and demand huge premiums on what was considered a commodity — a series of sharpened metal edges mounted in plastic. The key is that they didn’t just build a product to wear out (some would say we need to design our hardware to fail earlier so people will have to repurchase), they actually improved the use through software. They could be better though… they could extend the software metaphor further by allowing backwards compatability to a point that it no longer made sense. All the heads should work on all the bodies unless the connection mechanism changed to improve performance (as far as I can tell, there hasn’t been any innovation on how the razor head attaches to the handle). Imagine them stretching the metaphor even further — what would Gillette as an open source software company look like? How could they design in Web2.0 patterns into the blades? I’m reaching… but you get the point.

There is a company who already thinks like this: Doblin. They have a pdf that outlines a strategy to create business concepts. Even if you are a hardware company (or expand the spectrum to the full array of business models), you can take advantage of software company processes like rapid prototyping. Before you invest to change capital-intensive systems, you can concept artifacts and experieces to evaluate innovation opportunities. You can innovate your business model with very little investment.

The key in initiating something new and different is to find metaphors that link the unknown to the known. Revolutionary items don’t have much of a link to the present — they usually destroy the value of the current system. This is difficult to manage both internal and external to a company (investing, changing, and building in a current organization — and then conveying that to consumers of the product).

All this stems from thinking about the iPhone… Jobs tried to use the metaphor of a do-it-all device. I’m not sure that’s the best metaphor. Consumers have to value the idea of a do-it-all device. Simple is better. I’ll try to come up with better use metaphors for the iPhone.

I’m a little over a hundred pages into Marvin Minsky’s The Emotion Machine and came across this gem:

To understand how our thinking works, we must study each of those “very different things” and then ask what kinds of machinery could accomplish some or all of them. In other words, we must try to design — as opposed to define — machines that can do what humans do.

In my words, he’s taking about Designing to Define. I think that is an apt way to describe an interesting philosophy of doing to learn. In the software community the closest thing is rapid releases, rapid prototyping, etc. But what I’m talking about isn’t designing to improve — it’s designing to learn. That is something I’m not sure that product companies do intentionally. Maybe I’m wrong. But even if they’re not doing it now, they will (and those that are – you’re a step or two ahead).

Marketers, especially those of the direct variety, have always done this to some degree — if they do it right. On a very basic level they understand the values and motives of consumers through emergent design — the audience, the offer, the creative, endlessly reformulated in small increments, a plodding evolutionary approach. What makes that world complex, however, is 1) the spectrum of sophistication (or lack thereof) of other companies’ approaches and 2) the pure volume and breadth of direct marketing communication. Very few marketers actually take into account what else people are getting in their mail as a part of the audience, offer, creative equation. They don’t completely define context.

It is the proper atomization or disaggregation into component parts that is the difficult part of Designing to Define. I don’t have to understand all the workings and innards of a complex system, but rather I want to design a system that elicits the outcomes I desire in the appropriate contexts. Therefore:
1. I must be clear about my intended outcome(s)
2. I must understand the relevant components of context
3. I must be able to break the design features into mutually exclusive, collectively exhaustive parts

Doesn’t seem so easy when broken out that way. But the fourth element of DtoD is iteration. The outcomes, context, and design components must all be quantified and measured — then coupled in iterative ways that improve outcomes, are either specialized for specific contexts or rigorous enough for most contexts, and reduce time and cost of design and production.

I know. You’re thinking: this isn’t anything new. This is the scientific method. And I agree. But instead of designing experiments, just design products and services that are emergent and experimental while delivering value. Make the emergent qualities part of the value and assets of the company. You’ll get closer to actual “use contexts”, shorten the time to learn, reduce R&D costs in the long run, and make competively differentiated products and services that people use and will increasingly value.

Imagine the ultimate execution of the long tail in music.

First:
Every track published would be available to purchase via download–at customized levels of compression. Infinite supply.

Second:
You would be able to expose your music collection, playlists, blog entries, etc, as navigable items for purchase — where you take a variable cut of profits. Incentive to sell.

Third:
Discount tools would exist between publishers and sellers that allow for variable incentive arrangements based on conversion. The more demand you create for publishers the more you get paid.

Fourth:
The music is free from DRM and licensed for use in North America, flowing out to all populated countries. Item 2 and 3 help create checks and balances from widespread illegal file sharing.

Fifth:
You could find new music in multiple ways, suited to you. This would be accomplished via collaborative filtering, friend recommendations, related artists and tracks, tagging, folksonomies, genres, moods, etc.

Just rambling these off–but for a purpose. Burn Lounge seems to be at the front of something like this… however, without a change in the user experience, they are doomed to fail.

Read more on navigating to nothing.

(more…)

I’m currently undertaking research to probe demand shaping in the long tail. Interestingly, an HBR article presents another alternative to explain demand (rather than supply) effects of the long tail: network theory.

Duncan Watts and Steve Hasker, in Marketing in an Unpredictable World (you’ll need a subscription), explain why it’s difficult to build formulas to predict hits. Namely, social influence creates very complex interactions that are hard to parse and predict in a reliable way. This is interesting because, as Chris Anderson claims, the “free” digital supply of the long tail removes the barriers that shaped the markets of scarcity of old economics, leaving behind a market of abundance. Perhaps, beyond decreasing friction and cost of supply, new forms of communication technology have also turbocharged mechanisms of social influence.

The article is based on research published in Nature of a study of the roles social influence plays in driving aggregate consumer demand. The findings “suggest that the success of a particular entertainment product cannot be explained by any measure of intrinsic quality or even by ‘appeal’ — the fit between the product’s attributes and consumers’ preferences.” At face value this may be an artifact of the constrained markets of scarcity in that the supply is artifically small and therefore true demand is unable to emerge. While the study design does limit choice sufficiently that this may be the case, there is another paradoxical implication.

Matthew J. Salganik, Peter Sheridan Dodds, and Watts report in the Nature study: “when individual decisions are subject to social influence, markets do not simply aggregate pre-existing individual preferences.” What may be happening is that when social influence is apparent in a system of choice, true individual preference/demand is not revealed. This seems obvious enough — any high schooler can verify the claims of peer influence and social editing. However, this may cast shadows on the democratic nature of the filters, discovery engines, rating systems and aggregation platforms of Web 2.0.

Social networking, social ranking and rating, and other collaborative forms of content generation and communication may be governers to long tail and niche growth. If I make it easier for social influence to take effect — the most diggs on digg.com, the most listens on last.fm, or the most popular books on Amazon, all made more viral through email, IM, blogs, social networks, message boards, etc, peer influence may take on more powerful and demand-stifling forms.

Therefore, some of the same technology that is helping to create the long tail may be making it HARDER for individual’s to 1) understand what they desire and 2) act on that desire — due to hypercharged social influence mechanisms — those same systems and filters that drive demand into the long tail.

Business 2.0 has an interesting feature this month: Road Warrior’s Guide to Travel (at the time of publishing, the site didn’t have the April issue content up). The best of the Road Warrior (?) tips involve air travel. Check out Flight Stats.

The best way to see the value immediately is through the Flight Report. I can imagine scheduling all my business travel through an engine like this. And imagine if you actually paid for performance? I’ll use LGA to ORD as the test flight. It would work something like this:

First, if you delve into on-time performance metrics, you’ll see that on-time is relative. I can travel from New York to Chicago and have a scheduled flight time of 2 hours and 26 minutes if I leave at 6 am and 2 hours and 35 minutes if I leave at 7 am. The later flight has 9 extra minutes built in. Why should I pay the same for a flight that is scheduled for 6% longer for me, all things equal.

The key there is “all things equal.” If price adequately reflects demand (which we shouldn’t assume) we have to understand the components of demand. Comfort, speed, convenience, meets schedule needs, etc…

So let’s go back to the LGA to ORD example. There are five flights that leave on a weekday between 6 and 7 am. Say I need to get to a 10 am meeting in the loop. My primary concern is making the meeting on time–but I also don’t want to get up any earlier than I need to. These five flights are all in play.

So I’ll start with the two 7 am flights. How much time will I have from the time I land until the meeting? I want to give myself 60 minutes from the arrival. That means I have to arrive by 9:00. Looks like I won’t have a problem with either flight. But what are the chances of landing on time? The United flight (669) is on time (within 15 minutes of sked arrival) 78% of the time. The American (305) does a little better at 81%.

The next check is, if it is late, how late will it be? United averages 14 minutes delay with a standard deviation around 11. American comes in with an average of 15 minutes and a high standard deviation of around 31 minutes. So while they have a good chance of being on time, the American flight has much more variance in arrival time–if it is late it will have a much greater chance of being significantly late.

So based on performance, how much more or less will I pay for the American flight over the United flight?
Rather than using Flight Stats on-time percentages, I’m using their average and standard deviation on delays to come up with the 90% Minutes metric: I have a 90% chance of being no more than x minutes late.
For American flight 305, my 90%m is 55 minutes — 90% chance of arrive by 9:29 am.
United flight 669 90%m is 29 minutes — 90% chance of arriving by 9:03 am.

or

I have an 85% chance of arriving by my target time on United 669–subtract a 6% chance of the flight being cancelled = 79% chance of success.
I have a 62% chance of arriving by my target time on American 305–subtract a 6% chance of the flight being cancelled = 56%

This is just the start. If the meeting start time is very important–I’m depending on other people also arriving at certain times, 79% might not even be good enough. This also is using an average that hasn’t accounted for seasonality or additional weather concerns, that I can tell. Without taking this all the way through, so far I decided I have a 24% probability advantage on United. I’ll work some more on this in a later post.

Is Godin’s latest thing really a new perspective on marketing or a not-so-veiled advertising ploy for Squidoo? I’ll let you be the judge.

If you’ve read it–another question: does the megaphone metaphor work? If the funnel works, as Godin even agrees, by prospects entering at the wide side and customers exiting at the small end–how can the idea of flipping that same funnel, and rearranging which end = in and which out, make any sense? Wouldn’t the obvious, and more accurate, way to illustrate his point is a funnel like the figure on the right?